Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Rural earthen roads impact assessment in Phewa watershed, Western region, Nepal

Rural earthen roads impact assessment in Phewa watershed, Western region, Nepal Background: This work describes current research being conducted in the Phewa watershed, near Pokhara in Nepal’s Middle hills, a moist sub-tropical zone with the highest amount of annual rainfall in Nepal (4,500–5,000 mm). The main purpose of this study is to focus on the increase and impact of rural earthen road construction in the Phewa watershed as part of land use changes over 30 years in one of Nepal’s most touristic regions. Research methods were interdisciplinary and based on a combination of remote sensing, field observations and discussions with community members. The study compared 30 year-old aerial photos with current high resolution satellite images to detect changes in the watershed road network. Secondly, 42 % of the watershed’s roads (138 km) were surveyed in order to inventory and quantify shallow landslide events. Using an erosion-characteristics grid, their main features were measured (location, size, type and dimensions of damaged areas, etc.) and a Geographic Information System data base was created. We then estimated economic impacts of these events in terms of direct agriculture lands losses and road maintenance. Results: Results of the remote sensing analysis demonstrate that the road network increase is following an exponential trend, which enables us to predict future watershed road network scenarios. Field work investigations have demonstrated that unplanned road excavations are producing mainly embankment shallow planar soil slides and/or gullying which primarily affect roads themselves, wiping them out and blocking vehicle circulation, and secondly, destroying or burying agriculture lands. Along the 138 km surveyed, we calculated an amount of soil material released of about 70,000 cubic meters, which amounted to 99 cubic meters per kilometer annually. Of 179 cases of roadside erosion processes sampled, about 85 % directly impact roads or agricultural lands. Conclusion: The current mode of road construction which is currently occurring in Nepal is largely related with erosion and shallow landslide processes. Considering the exponential growth of rural earthen road networks, we would expect an increase of sediments released by roads and serious consideration must be taken if roads continue to be made without more careful methods. Through simple technologies using low cost and local resources along the lines of ‘green road’ or ‘eco-safe road’ approaches, it may be possible to reduce the impacts of rural road construction. Keywords: Rural earthen roads, Erosion, Sedimentation, Shallow landslides, Nepal * Correspondence: geoffroy.leibundgut@gmail.com Institute of Earth Sciences, University of Lausanne, CH-1015 Lausanne, Switzerland Full list of author information is available at the end of the article © 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 2 of 21 Background rural road construction has become a priority for the main Nepal has the greatest altitude variation on earth which village and district authorities, in other words the Village ranges from 60 to 8,848 m above sea level. The high Development Committees (VDCs) and the District Devel- mountains are the result of the collision between the opment Committees (DDCs) (World Bank, 2013). Many Indian tectonic plate and the Eurasian tectonic plate communities are collecting their own funds to rent bull- (Avouac, 2015). The uplift is currently active and makes dozers and build rural roads without proper technical and the country prone to earthquakes. The 147,000 square geological expertise (Das Mulmi, 2009). The road con- kilometers of this landlocked mountainous country are struction boom and lack of enforcement of government divided into five physiographic units from north to regulations has led to a low quality Nepal rural network, south: the high Himalaya, high mountains, the middle with roads which often do not provide year round access hills region, the Churia or Siwalik range and the Terai re- (World Bank, 2013). Slope cut during rural earthen gion in the plains (Fig. 1). About 80 % of Nepal is moun- roads construction certainly affect the frequency of tainous or hilly and 20 % are plains area of Terai, which shallow and in some cases larger landslides and local are located at the northern Ganga Basin (Agrawala et al., soil erosion/deposition processes, although the exact per- 2003). With about 200 km separating the plains area of centage of road induced shallow landslides is unknown Terai (south) from the high Himalayas (north), the coun- (Fort et al., 2010; Furniss et al., 1991). This human- try presents great topographic diversity. The country induced phenomenon directly impacts communities and therefore has high climate and vegetation diversity which infrastructure (Sudmeier-Rieux, 2011). varies from tropical in the Terai to High-mountains cli- In 1979, Laban demonstrated that 5 % of Nepal’s land- mate (snow and ice) in the north (Agrawala et al., 2003). slides (above 50 m ) were induced by roads or trails and Annual rainfall varies from 1,000 to 5,000 mm. As the formulated a serious warning about this high value since middle mountain range (around 2000 m.a.s.l) are the first the percentage of the land covered by roads network mountain barriers for the monsoon clouds moving north was at the time very small. The road and transportation from the Bay of Bengal, around 80 % of this precipitation network has clearly increased dramatically over the past occurs during this season, which lasts from June to three decades and we should expect a drastic increase of September (MoHA, 2009). Poor rock mass conditions road-induced landslides in Nepal since 1979 (Laban, makes Nepal’s slopes prone to fluvial and landslide 1979; Petley et al., 2007; Fort et al., 2010). Green roads, erosion. Moreover, geological context of the Middle or the use of simple engineering structures for drainage hills region is dominated by metasedimentary rock and slope stabilization, have been promoted in Nepal for with metamorphosed granites in upper part and, car- several decades (Das Mulmi, 2009). They are character- bonate and clastic sedimentary rocks, which are weak ized by the use of locally available deep-rooted grass spe- (Hashimoto et al. 1973). cies based on a participatory approach from planning to According to the World Bank (2013), the road net- implementation, and environment consideration. work has tripled in Nepal in the past decade. In 2013, The Phewa watershed is actually an area that has expe- the Strategic road network (SRN), which is managed at rienced an exponential increase of rural roads network the central level, was approximately about 11,000 km (as it will be developed farther in this paper) partly due to while the Local Road Network (LRN), which is managed the presence of the adjacent town of Pokhara and the at local level, was about 60,000 km (World Bank, 2013). presence of the Phewa Tal, one of the most prominent lake Roughly half of the SRN and more than 95 % of the in Nepal. Both have contributed to attract people for agri- LRN is unpaved, giving a paved network of 8,000 km culture and tourism. Therefore, we selected this watershed only or 11 % of the total road network (World Bank, as the study area to quantify roadside erosion events. 2013). The SRN road density increased from 3.22 km of The objective of this paper is to share observations on roads per 100 km in 1998 to 8.49 km of roads per the current state of rural earthen roads and their im- 100 km (DoR, 2015). According to the World Bank pacts in terms of an acceleration of erosion rates and re- (2013), the Nepal road density is actually high as com- lated costs for road maintenance and agriculture land pared to other mountainous countries and largely due to losses in the Phewa watershed. Sedimentation rate and the trend in the past decade of opening up new roads in costs results could provide a basis for further studies on Nepal. The government has placed a great emphasis on conventional versus green road construction. developing the roads and the transportation infrastruc- ture as a real means of development for the rural popu- Description of the study area lation (World Bank, 2013). Phewa watershed is located in the western part of the Moreover, according to Sudmeier-Rieux (2011), the re- Pokhara valley of Kaksi District in the Western Develop- sult of the 2008 Decentralization Act produced authority ment Region of Nepal. The watershed lies within latitude and budgets transfer to local governments. Consequently, of 28° 11’ 41.7” to 28° 17’ 26.0” north and longitude of Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 3 of 21 Fig. 1 Nepal’s physiographical regions, and Phewa watershed location and boundaries as considered in the study. Note on the bottom map the erosion areas measured along the surveyed roads. The surveyed roads network makes part of the total roads network (Source for the top map: Forest Resource Assessment Nepal, 2015; Source for the bottom map: Pleiades Satellite Imagery, 2013) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 4 of 21 83° 47’ 53.2” to 83° 58’ 04.3”east. It covers an area of lithology is mostly gritty phyllite (46 % of the total area) about 111 square kilometers (Fig. 1). and fine grain massive quartzite (25 % of the total area), The outlet of the Phewa Lake is situated at 784 m in the south-western part of the watershed (Harpan sub- above sea level (Sthapit and Baila, 1998) where the high- watershed). The watershed is also comprised of colluvial est point is marked by the Panchase ridge summit at in homogeneous deposits (silt, sand, gravel and boul- 2517 m.a.s.l. The confluence of two main tributaries ders) and debris flow deposits. The flood plain is mainly Sidhane and Adheri, form the Phewa river drains into gravel, sand, silt and clay deposits. the Phewa lake. These two sub-watersheds cover 24.5 The watershed climate is based on annual monsoon and 28.4 % respectively of the total watershed area events, which bring more than two thirds of the annual (Sthapit and Baila, 1998). According to our data, in rainfall between June to September. This season is char- 2013, Phewa lake covered 3.3 % of the total watershed acterized by intense rainfall; events of 300 mm in 24 h area. This figure is similar to recently published data, are not uncommon in this area (MoHA, 2009). Between which determined the lake’s surface to be 3.96 % in 2013 1982 and 2012 the mean annual rainfall at Lumle me- (Rimal et al., 2015). The capacity of the lake was estimated teorological station (located at North West of the water- as 42.18 million m in 1998, and the annual average sedi- shed, 1740 m.a.s.l.) was about 5,506 mm. At the Pokhara mentation rate in the lake was about 180,000 m (Sthapit Airport meteorological station (827 m.a.s.l.), it is about and Baila, 1998). If a constant rate is considered, 80 % of 3,875 mm. These extreme natural events contribute of the lake’s storage will be silted up in about 190 years ac- the natural degradation of steep terrain including the cording to the same authors. Moreover, Phewa Lake is triggering of landslides or flash flooding. also experiencing accelerated eutrophication, land en- croachment, and massive invasion of water hyacinth and Methods exotic carp fish species (FEED, 2014; JICA, 2002). Data available and field investigation According to the 2012 land use classification based on The present survey was conducted through data available a 2012 RapidEye satellite image (5 m resolution) under- and data field collected during November 2014 campaign. taken by UNEP (Sharma et al., 2013), land use in Phewa Available data: watershed was comprised of 41 % productive (agricul- ture/grassland terrain), 49 % forest (trees and bushes), ▪ Digitalized topographical map of 1996 from the and 5 % water bodies (lake, river and swamp area), 3 % Government of Nepal, NGIIP, Survey Department, built up area, 1 % sand area (near rivers and lake). The Ministry of Land Reform and Management. area was part of a watershed land use management pro- This digitalized topographical data set is directly usable gram, which started in the 1970s (Fleming and Puleston on a Geographic Information System (GIS) program. Fleming, 2009). As far as the difficulties to reduce ero- The data are compounded by Shapefiles of sion are concerned, the program focused on the conver- transportation and hydrographic networks, sion of “critical landscapes” (Fleming and Puleston administrative and built zone boundaries, 20 m Fleming, 2009: 38), such as degraded shrubs, grazing topographical contours lines and land cover areas. land and unmanaged forests, to managed community ▪ 1979 aerial photography and 2013 satellite images. forests or managed pasture. According to the authors, The 2013 images come from Pleiades Satellite Imagery watershed forest land increased from 28 % to 36 % be- and are of 2 m resolution for the 4 bands (multispectral) tween 1978 and 2006 while the terraced arable land and of 0.5 m resolution for the panchromatic. The 1979 remained constant. Forests managed by Community aerial photos come from the Department of survey of the Forest User Groups exceeded 60 % of total forests in the Government of Nepal. Both fully cover the watershed. watershed in 2006. Forest (and bush) cover steeper ▪ 1998 Geological map of Pokhara valley published by areas, possibly due to improved community forest devel- Department of Mines and Geology in cooperation with opment in the watershed, and could play a role in the Bundasanstalt für Geowissenschaften und Rohstoffe - protection of soil from mass movement and failures Federal Institute for Geosciences and Natural (Papathome Koehle and Glade 2013). Resources, Hannover, Germany (Koirala et al., 1998). The watershed lithology is compounded by intensively ▪ 2012 Land use base map prepared by UNEP for the weathered rocks and weak soils, highly prone to erosion Ecosystem-based Adaptation in Mountain Ecosystem and shallow landslides (Agrawala et al., 2003). According project in Nepal. The classification was undertaken on to the geological map made in 1998 by the Department a 2012 RapidEye satellite image (5 m resolution) using of Mines and Geology of Nepal in cooperation with the segmentation and object classification method on the Bundesanstalt für Geowissenschaften und Rohstoffe - eCognition software tool (Sharma et al., 2013). Forest Federal Institute for Geosciences and Natural Resources, areas, agriculture areas, and water body and sand areas Hannover, Germany (Sikrikar et al., 1998), the watershed were classified. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 5 of 21 During field work in November 2014, about 42 % of Nevertheless, not all the events directly affected the the total watershed roads network (138 km) was sur- surrounding environment, infrastructure or persons. veyed to detect and inventory erosion processes directly induced by road construction. We created an erosion- We then created an Excel sheet summarizing the field characteristics grid to characterize and classify the obser- work measurements. One important step of the inven- vations and measurements undertaken during field tory work was to measure relevant dimensions of each work. All these data were gathered in a Geographic failure event with a laser distance meter to propose volu- Information System (GIS) project using ArcMap 10.2 metric estimation using basic 3D geometric shapes (see (ArcGis program) in the projected geographic coordin- further for the volume estimation methodology). For ate system WGS 1984 UTM, Zone 44 North. each damaged area, the road deposit volume and the agriculture land lost surface were also characterized. Road network detection and comparison for the period Moreover, by representing event GPS location points 1979–2013 (as Shapefiles) in the GIS project, it was possible to rec- Three sources of data enabled to detect and compare ord some relevant information that could be directly in- changes over the period 1979–2013 in the Phewa water- tegrated in the database. For example, we noted the shed road network. The 1979 aerial photographs were 2012 land use and slope angle for each landslide event digitalized and geo-referenced by our care using the pro- point location. gram ENVI 5.0. Roads were manually detected respect- The main idea of creating an erosion-characteristic ively on the 1979 aerial photos and on the 2013 satellite database paired with a GIS project of the Phewa water- images. We define roads as the transportation ways for shed was to be able to generate statistics about simple vehicles in general (black top and earthen roads) but volume estimations of the road side events. earthen roads are clearly the major type of transporta- tion way in the watershed (both in 1979 and 2013). Sha- Classification details pefiles from the 1996 digitalized topographical map Fluvial and hill slope erosion processes were inventoried, provided a detailed overview of the entire road and path and differentiated considering their main conditioning network, which was re-classified to differentiate the two. factor (natural or human induced). The landforms iden- The main goal of this remote sensing work is to com- tified differ in type, shape, road location and potential pare the length of the road network over the three past damages involved. decades in order to characterize road network trends of The following list/sub-sections details each relevant the watershed. parameter of the erosion-characteristics grid: Road induced erosion and damaged area events 1. Road-induced erosion events inventory Event type: The mass balanced method of excavation About 138 km of the watershed’s roads were surveyed (‘cut-and-fill’) is commonly used for road construction over 3 weeks of field work to measure main features of on hill slope (Keller and Sherar, 2003;Cornforth, the erosion processes (shallow landslides, gullies) in- 2005). The ‘cut-and-fill’ design (Fig. 3) could be at duced by road construction (see Fig. 1). origin of landslides and road embankment failures: “an We directly separated this inventory work into three approximately equal cut-and-fill cross-section can (i) different general classes (Fig. 2): undermine the upper slope, causing it to fail, (ii) overload the downhill slope, causing it to fail, or (iii) 1) Road-induced erosion events: we measured their cause the entire slope to become an active landslide.” position along the road and elevation using a (Cornforth, 2005: 12). Following the definition of Geographic Positioning System (GPS), their Hungr et al. (2013), roadside erosion events observed dimensions using a laser distance meter; we also in Phewa watershed are commonly shallow and planar noted failure event activity/age, the material soil slides. Upper and lower roadside embankments involved, etc. can be affected. Moreover, gullies induced by runoff, 2) Classification of the road through observations and especially during monsoon, on the ‘cut-and-fill’ design discussions with community members. road embankments are also an erosion phenomenon 3) Potential damaged areas due to road-induced erosion observed in the watershed. Erosion events were events. Two damaged areas were considered: the classified as follows (Fig. 3): road itself that could be wiped out or blocked, and agriculture land that could be buried or destroyed. Each one represents an economic impact and, in some a. Embankment shallow planar soil slide or shallow cases, could injure persons living in the watershed. soil slip: Shallow soil volume sliding along an Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 6 of 21 Fig. 2 General architecture of the erosion-characteristics grid summarizing field’s measurements. The three main classes - erosion events, surveyed roads and damaged area - are inter-related and are characterized for each point measured inclined slip and planar surface. They are embankments and form gullies. Moreover, this characterized by small size and thickness, with phenomenon could be at origin of channelization in volumes up to few cubic meters (Crosta et al., 2003). the lower slopes; it is thus considered to be gully They are occurring directly on the upper or lower erosion induced by road construction. The road embankment. consequences are deposits produced by bedload b. Extended shallow planar soil slide: Corresponds to a and/or small debris-flows. larger shallow planar slide and doesn’t occur only on the roadside embankment but also on the Activity surrounding hillsides. They are characterized by This parameter is important and is mainly signaled by length up to 20 m and width up to 10 m, and the age of deposits. Numerous signs indicate activity, involve larger quantity of material. As far as is such as the rock deposit patina and weathering, if the difficult to be sure to consider road as the driver of deposit is re-vegetated, etc. this larger event, the few cases measured here were clearly triggered by the roads. ▪ Recent: less than 2 years; c. Gully: In some cases, road construction affects water ▪ Medium: between 2 and 5 years; drainage that could cut upper or lower road ▪ Old: more than 5 years. Fig. 3 Cut-and-fill design of road construction and related slides and failures events potentially occurring in Phewa watershed (modified from Keller & Sherar, 2003) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 7 of 21 1.5 Event volumes V = 0.05 × S (with S the landslide surface and V the The volume of material failures was estimated as the vol- landslide volume). ume of sediment removed assuming interpreted ‘pre ero- Slide surface could be easily approximated as an sional’ land surface and simple geometric 3D erosion ellipse depending on length and width measured shapes (Cossart & Fort, 2008; Campbell & Church, 2003). during field work. The erosion event type is, defined as a horizontal shape in two dimensions of the erosion area used as base Slope to calculate the volume. The formulae used to estimate Contour lines of the 1996 digitalized topographical map erosion volume are based on the basic dimensions mea- were used to create a raster DEM 30 m resolution sured in the field with a laser distance meter (length, through interpolation tools on ArcGis 10.2 (based on width and thickness). For example, shallow soil slide discretized thin plate spline techniques). The DEM has events are largely defined by a parallelepiped volume of enabled then to compute slope angle values. Due to the material of a given width (W), length and thickness slid- small DEM resolution, the slope layer represents only ing along a plane (L); in this case the 2D base shape is a the angle value of the regional slope of the watershed. parallelogram. Figure 4 illustrates the flowchart detailing We considered 4 slope classes: [0°; 13°], [13°; 27°], [27°; the volume calculation process. 39°] and [39°; 90°]. The list below details the volumes estimation method according to each event type. Basic dimensions and 2D Lithology base shapes are illustrated by Fig. 5. This parameter was processed by joining the 1998 Geo- logical Map, digitalized and vectorized to be available on ▪ Embankment shallow planar soil slide or shallow soil a GIS program, with event points shapefiles. slip: The 2D base shape could be of 2 types: semi-ellipse or parallelogram. Volume is calculated as the product Land use type of the base shape area (depending on length and width This parameter was processed by joining the 2012 Land parameters) by the thickness of the sediment volume Use layer with event points shapefiles. removed. ▪ Gully: The 2D base shape section is defined here as a Surveyed roads semi-ellipse (similar to the U-shape section of a gully) A large part of the watershed road was traced using a and volume estimation is the product of the surface Global Positioning System (GPS) during the field work and length of the gully. and then classified according to the observations and ▪ Extended shallow planar soil slide: When it was discussion with the community members. Each event difficult to measure the slide thickness on the field, we also contained road information according to its location used an empirical relation (Hovius et al., 1997): (as it is classified). VOLUME ESTIMATION PROCESS Erosion event FOR EROSION EVENTS Embankment shallow planar soil Extended shallow planar Gully Event type slide or shallow soil slip soil slide Ellipse Semi- Parallelogram U-shape 2D base shape ellipse Parallelogram surface × Empirical relation 1/2 ellipse surface × Semi-cylinder with Event volume estimation thickness thickness ellipse base section (Hovius et al., 1997) Equation Fig. 4 Volume estimation process for erosion events. L, W and T correspond respectively to length, width and thickness of the event and are illustrated in the Fig. 5 Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 8 of 21 Fig. 5 Basic dimensions as measured on field used to determine: (a) embankment shallow planar soil slide event volumes according to the two types of base shapes, (b) gully erosion event volumes Surface type: Volumes to remove or repair: The volumes that are removed or used to repair are estimated as ▪ Paved road: Black top surfacing. parallelepipeds of given length, width and thickness. ▪ Earthen road: Earth surfacing. They were measured during field work. In some cases, the full volume of the involved material is Road access: considered to be removed; this depends largely on the observations undertaken on field (Fig. 6). ▪ Drivable: By a 4 wheel drive car. This parameter is directly used to estimate the ▪ Not-drivable: Because of a bad road surfacing or direct cost of maintenance. deposit/failure cutting the road. B. Affected agriculture land Type of damage: Road maintenance: a. Failure: Failure of terrace areas destroying crop area due to a collapse/slip of the upper road ▪ Maintained: Shows numerous signs of maintenance embankment. (protection and stabilization infrastructure, drainage b. Deposit: Burying of crop areas by soil material system, etc.). However, this does not necessarily reflect coming from upper-sides. how well the road is maintained. Crop type: Type of crop cultivated before the ▪ Unmaintained: The road is left in its actual state. damages occurred. Dimension of agriculture surface lost: The productive Damaged areas crop surface lost is considered as parallelogram of a given Characterizing the damaged area was undertaken mainly length and width measured during field work (Fig. 6). to estimate direct costs of the road maintenance and the This parameter is directly used to estimate the agriculture land losses according to several scenarios. direct cost of agriculture lands losses. A. Affected roads The flowchart (Fig. 6) illustrates a simple way to esti- Type of damage: mate roadside volume to remove or repair and agriculture a. Deposit: Burying of road surface by up-slope soil surfaces directly affected by road induced erosion events. material. Maintenance will consist of removing deposits from the road. Results b. Cut in road shoulder: Road surface failure due to Slope angle distribution of the watershed a collapse/slip of the lower road embankment. We have divided the Phewa lake in four classes accord- Maintenance will consist of filling in the eroded ing the slope; each class has a different land use (Fig. 7): part with soil deposits. The first class [0–13 °] corresponds to the flood plain, Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 9 of 21 VOLUME AND SURFACE Damaged ESTIMATION PROCESS FOR area DAMAGED AREAS Agriculture Damaged area type Road land By material By material By cut in road By failure of deposition on deposition on Type of damage shoulder terrace road field Mainly Others event Mainly gully or Mainly shallow deposition Shallow soil slip Type of event types Shallow soil slip soil slip downslope of road Volume of the Surface of fallen Surface of deposit Volume of the Volume or surface Total volume of affecting the eroded cut terrace buried field the event road segment calculated as a calculated as a calculated as a estimation calculated as a parrallelogram parallelogram parallelepiped parallelepiped Fig. 6 Volume and surface estimation process for damaged area. Note that simple shapes, such as parallelepiped and parallelogram, are used to estimate volumes and surfaces fluvial terraces, distal parts of alluvial fans and the lake. than 39°; this area corresponds to rock slopes with low Most of the settlements, schools, and agriculture and vegetation and steep terrain covered by forest. grazing lands, occupy this area, accounting for 25 % of It is important to note that the 5 m resolution of the the total watershed and part of Pokhara city, which lies 2012 RapidEye satellite image allowed a rough classifica- in the watershed. The main reason for the land use of tion of the land use type of the watershed. Boundaries this sector is the soil conditions, better ability for terra- between two classes are not well defined. Also, an iso- cing, water drainage, and access to roads. The second lated element of a land use type could be not detected. class [13°–27°], also corresponds to low relief area. It is However, in this case, the 30 m DEM that has allowed primarily covered by agriculture terrains and settlements computing only the angle value of the regional slope of whereas the third class [27°; 39°] is mostly covered by the watershed, it is relevant to assess the slope angle dis- forest. Finally, 3 % of the total watershed area is steeper tribution in relation to this rough land use classification. Fig. 7 Slope classes distribution (see below for calculation) according to the 2012 land use type. It has been standardized to take into account the same proportion of surface covered by each slope class area in relation to the total watershed area; in that way, land use type area for a given slope classes based on the same surface calculation and so, we represent here the weight of the different land use types for each slope class in the same proportion (Source: 2012 Land use base map prepared by UNEP for the Ecosystem Based Adaptation in Mountain Ecosystem project in Nepal. The classification was undertaken on a 2012 RapidEye satellite image (5 m resolution) using segmentation and object classification method on the eCognition software tool (Sharma et al., 2013)) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 10 of 21 Road network increase and trends is physically limited and the road network length will In 1979, there were no black top roads in Phewa water- therefore tend to a maximum value. The model would be shed whereas in 1996 about 5 km of the road network therefore characterized by an exponential trend reaching was paved. A classification between earthen roads and an asymptote corresponding to the maximum value. black top roads for the year 2013 is based on the road Aderamo (2013) has also characterized growth of cities survey conducted during field work, where most of road networks (Ilorin, Nigeria) using aerial photos and im- black top roads of the Phewa watershed (about 10 km) ages and he showed that it conforms to a logistic curve. were surveyed. In addition, we included a section of the However, to make earthen roads length predictions for the Pokhara road network belonging to the Phewa water- next decade, we assume here that in the short run, the shed to reach a total black top road network length of growth trend will not have reached its maximum value. about 18 km for the year 2013. The total length of the Regarding the total road network growth according to 2013 earthen roads in the watershed was calculated as slope angle classes (Fig. 10), the curve fitting of the road the difference between the total number of transporta- construction indicates the largest growth coefficient tion ways detected on the 2013 satellite image and the (superscript of the exponential relation) for the range black top roads length calculated as explained previ- [27°; 39°], which correspond to medium steep areas. ously. Following Fig. 8, the watershed black top roads Then, follow the ranges [13°; 27°] and [0°; 13°], which length increase is more defined by a 2nd order poly- correspond respectively to the low steep areas and the nomial relation whereas the earthen roads length flood plain. Both show a quite similar road construction grew exponentially. We then focused only on the increase coefficient. Moreover, the largest increase quan- earthen roads network, as we will demonstrate further titatively occurs for the range [13°; 27°]. This means that earthen roads are the main trigger of erosion events the major road network length is located in low steep in the watershed. areas and in the flood plain, with respectively 54 and The watershed earthen roads network length has 36 % of the total watershed road network. This observa- considerably increased over the past three decades, tion is clearly related to the land use type in these slopes from about 23 km in 1979 to 310 km in 2013, due to classes; agriculture activity and settlements location have the new earthen roads boom in Nepal (Fig. 9). The played an important role in developing transportation increasing trend is clearly defined by an exponential infrastructure in that areas. If the road network growth curve and we can consider the trend line (in Fig. 8, of these three slope classes follows an exponential trend, plot (B)) as representative of the earthen roads net- we can note that results for the range [13°; 27°] are not work length vs year, as the correlation coefficient is really relevant as the correlation coefficient is low. Fi- close to 0.95. nally, road length growth in the range [39°; 90°] is better Nevertheless, the transportation network growth model defined by a linear relation and is little represented in would be better defined by a logistic curve as the territory the total watershed road network. ab Black top roads network increase in Earthen roads network increase in Phewa Phewa watershed over the past 3 decades watershed over the past 3 decades 20 350 y = 0,0135x + 0,0632x + 3E-14 0,0764x y = 19,399e R² = 0,949 0 0 010 20 30 40 010 20 30 40 Year (from 1979) Year (from 1979) Fig. 8 Increase of (a) black top roads network length and, (b) earthen roads network length in Phewa watershed over the past 3 decades. The horizontal scale starts in the year 1979 (0 corresponds to the year 1979). Note that for the black top roads increase case (plot (a)), the growth is defined by a 2nd order polynomial trend. Only 17 km of black top roads built more than 30 years illustrates well the few considerations in developing a paved road network in Nepal. For the earthen roads increase case (plot (b)), the growth is well defined by an exponential trend and clearly illustrates the earthen roads construction “boom” occurring in Nepal over the past decade Length (km) Length (km) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 11 of 21 Fig. 9 1979, 1996 and 2013 roads networks in Phewa watershed. Note that the watershed roads networks length has increased significantly over the period of 1979–2013 Erosion events and damaged areas: from inventory to partially or totally and in 21 cases, agriculture lands were statistics affected. Most of the shallow landslides identified locate Along the 129 km of earthen roads surveyed, 179 ero- in the northern slope of the watershed (Fig. 11), where sive landforms induced by road construction were iden- the road is cutting perpendicularly a steeper part of the tified. In 155 cases, the road network was washed out watershed. In the south, fewer roads were built up, but Road network length increase according slope degree classes [0° ; 13°] [13° ; 27°] [27° ; 39°] 0,0764x y = 10,334e 140 [39° ; 90°] R² = 0,908 0,0743x y = 8,8072e R² = 0,9947 0,1149x y = 0,6108e R² = 0,9933 y = 0,0377x R² = 0,7529 0 5 10 15 20 25 30 35 40 Year (from 1979) Fig. 10 Total roads length increase trends according to slope classes. The horizontal scale starts in the year 1979 (0 corresponds to the year 1979) Length (km) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 12 of 21 Fig. 11 General view of Phewa watershed mapped with erosion type (with illustrations) measured on roads surveyed during 2014 field work. Extended and embankment shallow planar soil slides and gully are the three type of road-induced erosion events observed in the watershed Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 13 of 21 also fewer surveys were carried out. The contrasting to failure processes but also due to land down-slope buried landslide occurrence between north and south reflects by deposits. Extended shallow planar soil slides are also the link between road construction and shallow landslide playing an important role in burying agriculture lands. occurrence. More erosion events were detected in the medium range Results focus on counting the number of landforms in activity than in the recent one, as 3 years are included in addition to estimating the volume relation between ero- medium range while 2 years are included in the recent sion events, damaged areas and surveyed roads, first for range. Nevertheless, total volume released for the medium the total area of the watershed and, secondly according and the recent ranges activity, respectively led back to 1 to zonal parameters: slope angle classes, lithology and year, give an amount of about 12,500 m /year for each land use type. one; this result traduces that the event activity parameter is relevant for the recent and medium ranges. However, Main statistical results due to high land use changes and quick vegetation growth, Most of the roads on the Phewa lake are earthen roads old range activity total volume is not relevant due to the (Fig. 12 and Table 1). Along them, landslides and gullies difficulty in detecting older events. Otherwise, the dam- developed in relationship with the cut of slopes during aged areas cases were mostly induced by recent events. road construction and when by diverting run off. More- Amounts have been estimated for the total watershed over, in the same way, the majority of cases measured roads network: along 310 km of earthen roads, 166,982 m occur on the unmaintained road segment, wiping it of material is released, 46,835 m need to be maintained them out and leaving the road undrivable in more than and 89,445 m of agriculture lands is lost. These values let two third of cases. Nevertheless, about the half of the assume than road-induced erosional conditions are well network surveyed clearly showed signs of maintenance. homogenous along watershed roads. As mentioned above, Given that the watershed research primarily targeted ac- the remaining watershed roads which were not been sur- cessible roads by jeep, the remaining watershed road veyed during the field work are in worse state than the sur- network would obviously being worse state and certainly veyed one: this observation could significantly increase unmaintained, that means that the roads surveyed dur- erosion events volume and damaged areas cases. ing field work may be not representative of the total Assuming homogenous erosional conditions along watershed roads network. earthen roads in the watershed, we can estimate kilomet- Table 2 and Fig. 13 summarize total results of erosion ric indexes for erosion events volume, roadside mainten- events volume, roadside maintenance volume and agri- ance volume and agriculture land lost surface. Therefore, culture land lost surface amounts. It also classifies these this human-induced phenomenon is responsible for re- values according to type and activity of erosion events. leasing 544 m of soil per kilometer of earthen road. To Along the 129 km of earthen road surveyed, an esti- take into account the time dimension, we can use totals mated 70,133 m were directly released along roadsides. estimated according to the event activity parameter: about 3 3 Of this total amount, 28 % or 19,671 m of material de- 198 m per kilometer of earthen road is released over posited need to be maintained and 37,567 m of agricul- 2years (total ‘recent’ events volume per 129 km), which ture land were buried or destroyed. gives an annual volumetric erosion rate per kilometer of 3 −1 −1 Extended shallow planar soil slide is the mechanism about 99 m .km .year .The ‘recent’ value was used to releasing the most of material volume even if the num- calculate the rate because it is considered as the most ex- ber of cases measured is lower in relation to embank- haustive events volume measurements in the watershed. ment shallow planar soil slide or shallow soil slip (13 The same assumptions was used to determine annual cases detected for extended shallow planar soil slide roadside maintenance volume per kilometer and annual against 150 cases for embankment shallow planar soil agriculture land lost surface per kilometer which amounts 3 −1 −1 2 −1 −1 slide or shallow soil slip). This observation could be ex- respectively to 39 m .km .year and 86 m .km .year . plained by the fact that extended shallow planar soil This Figure is comparable to the Figure given by Validya 3 −1 −1 slides are clearly a much larger phenomenon than the (1985, 1987) of 55 m .km .year of debris produced by fewer embankment slips defined as embankment shallow rural earthen roads (cited by Sharma and Maskay, 2009). planar soil slide or shallow soil slip and involve greater quantities of soil material. Moreover, 16 cases of signifi- Influence of various parameters on roadside erosion and cant gully erosion were detected and involved lower damaged areas quantities of soil material. Nevertheless, regarding the Additional analyses were undertaken on key watershed pa- damaged areas, embankment shallow planar soil slide or rameters related to erosion events and damaged areas shallow soil slip are clearly responsible for most of the (measured along 42 % of the watershed roads network). road maintenance needed and also represent a large part The study therefore focused on the lithology, on the 2012 of agriculture land losses, especially in terraces subject land use and on slope angle classes with the goal to Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 14 of 21 Fig. 12 Kilometric distribution of erosion events and damaged area cases along the surveyed roads. Note that to avoid the length factor, the number of cases is standardized in the distance unit, e. g. by kilometer of road. Results are clearly linked with the state of roads; erosion events (as defined in the methodology section) and induced damage areas (e.g. the road itself and the agriculture lands) cases occur most of time along unmaintained earthen roads calculate the total erosion events volume, the total roadside a. Influence of lithology maintenance volume and the total agriculture land lost sur- face per classes of slope, per classes of land use and per Erosion events occur in colluvial soil and in the fine classes of lithology. However, it is likely that greater vol- grained and massive quartzite formations (which are umes or surface value are to be found on a larger area of located in the in Harpan sub-watershed - south-west detection. Similarly, we standardized the erosion events and of Phewa watershed), constituting 40 % of the total maintenance volumes, the surface areas lost and the road volume released and 60 % of material volume affecting length by the surface of class of detection. Finally we plot- road network (Fig. 14, plot (a)). According to the ted the ratio volume/km or surface/km for each class in second revision of the geological and soil cover map order to represent only the influence of the parameter, not (Sikrikar et al., 1998), colluvial soils are indeed taking into account the surface value of detection. considered the most erosion prone of the geologic Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 15 of 21 Table 1 Erosion events and damaged area cases number measured along 42 % of the watershed roads network Surveyed road Erosion events Damaged areas Road Agriculture land Classification Length (km) Ratio Cases number Cases number Cases number Total road 138.2 100 % 179 155 21 Earthen road 128.9 93 % 179 155 21 Black top road 9.3 7 % 0 0 0 Drivable 124 90 % 144 122 12 Undrivable 14.2 10 % 38 33 9 Maintained 72.1 52 % 39 29 2 Unmaintained 66.1 48 % 140 126 19 Results are given also according to surveyed roads classification. Note that the majority of the watershed road network surveyed is earthen surfaced and drivable formations of the watershed. According to Sikrikar b. Influence of slope angle et al. (1998) and as observed in the south-west of the Phewa watershed, the quartzite formation is prone The erosion events detected and road networks to deep gully formation, rock fall, rock slides and surveyed up to a slope angle of 39° clearly make the wedge failures involving rather rough material (rocky higher slope class the most prone to erosion events deposits, as gravels and boulders). Some deep and and volumes of soil released and thus induced road large gullies, due to road induced modification of damaged cases (Fig. 14, plot (b)). The volume water drainage at the origin of channelization in the released per kilometer is clearly increasing with the steep lower-slope, were also observed in this area. slope angle, which is realistic considering that slope Moreover, Gritty phyllite formations, covering roughly angle is a main triggering mechanism for mass half of the total watershed area, is also susceptible to movement and slope failure. Nevertheless, seeing mass movement and slope failure and represents that the road network increased coefficient in the about 20 % of the total material volume released and range [39°; 90°] is very low (Fig. 10), we cannot 25 % of material volume affecting road network. expect a real significant increase of the material Agriculture land lost surface are mostly (almost 70 %) released in slope area up to 39° for the next decades. occurring in Gritty phyllite because most agriculture The real road impact will be more significant in the areas are located in this formation (south aspect of slope ranges [13°; 27°] and [0°; 13°]. Regarding the watershed). agriculture land lost surface, it appears theoretically Table 2 Total volume released on roads surveyed, total volume that need maintenance and total agriculture land lost surface amounts along 42 % and 100 % of the watershed roads network Inventory of soil volume losses Erosion events Damaged areas on roads surveyed Road Agriculture land Erosion events Volume Volume to remove Surface buried or 3 3 2 number release (m ) or repair (m ) destroyed (m ) TOTAL Along 42 % of the 179 70133 19671 37567 roads network Along 100 % of the 426 166982 46835 89445 roads network Event type (42 % of Shallow soil slip 150 22469 18576 20457 the roads network) Gully 16 8817 356 0 Extended shallow 13 38847 740 17110 planar soil slide Event activity (42 % of Recent (<2 years) 72 25580 9998 22,202 the roads network) Medium 78 37178 5767 15,320 (>2 to < 5 years) Old (>5 years) 29 7375 3907 45 Results are also classified according to event type and event activity (for measurements on 42 % of the watershed roads network) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 16 of 21 Fig. 13 Distribution of erosion events volume released, roadside maintenance volumes and agriculture land lost surface according to event parameters along 42 % of the watershed’s road network that it occurs on range [13°; 27°], area mostly Road length, roadside erosion volume and direct costs covered by agriculture land. forecasts In addition to the volume of erosion created by road con- c. Influence of the 2012 land use struction, this study also calculated direct economic losses Erosion events and roadside slope failures are mainly for rural earthen roads due to roadside maintenance and occurring in forest lands (60 %) and agriculture direct agriculture land lost surface. This analysis was lands (40 %) because these two land use type are undertaken considering community-based road mainten- covering the main part of low to highly steep areas ance scenarios using three different types of maintenance: (Fig. 14, plot (c)). A map has been prepared to put labor based, equipment based and mixed technology into evidence this observation: most of the roads (combination between labor-based and equipment-based network where erosion events are occurring is maintenance). A more detailed description of the method- located in these two main land use classes (Fig. 15). ology used is published in Additional file 1. Even if forests can be considered as protection against Table 3 summarizes data for a medium excavator erosion and landsliding, in this case, slope steepness is (Daewoo Solar 130LC – V) and incorporates the cost of the main factor explaining why more failures are land losses for the three scenarios. Results are given also occurring in the forest area than in agriculture area. according the event age (recent, medium and old). Con- Slope angle appears to be the most relevant parameter sidering the use of ‘mixed technology’ for road mainten- for explaining the erosion phenomenon induced by ance, economic impacts due to rural earthen road earthen roads in the watershed. Note finally that the construction in Phewa watershed amount to 49,260 USD agriculture surface lost found in the forest land use for 42 % of the network. Assuming homogenous ero- type is not relevant and could be translated by a lack sional condition along roadside in the watershed, the of precision of the land use classification undertaken total cost for 100 % of earthen roads will rise to 117,287 on a 5 m resolution satellite image in relation to the USD and the annual earthen roads economic impact per −1 −1 field work measurements. Earthen roads are of about kilometer would be 116 USD.km .year . 3 m maximum of wideness and measured GPS points Furthermore, road length, erosion and cost values could easily be taken into account into the bad land were forecasted in Table 4 for the period 2014–2030. It use class due to the rough resolution of the provides forecasts assuming constant rural roads growth classification. and homogenous erosional conditions in the watershed. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 17 of 21 Fig. 14 Kilometric distribution of the erosion events volume released, roadside maintenance volume and agriculture land lost surface according to: (a) lithology, (b) slope angle classes and, (c) the 2012 land use type. Note that all erosion and failure events occurred in the three main types of lithology of high susceptibility for landslides. Volume released and maintenance volume clearly growth with the slope angle, whereas land lost surface keep on the range [13°; 27°] as it mainly covered by agriculture land For instance, in 2030, 955 km of rural earthen road in impacts. Results demonstrate that road-induced erosion the watershed will produce a total erosion volume of in the watershed is mostly occurring on roadside em- about 231,000 m , which represent an economic impact bankments, which produce small and shallow soil fail- of about 192,000 USD. Even if erosion and cost indexes ures which often make the road drivable for no more were characterized for 42 % of the watershed road net- than one year until the next monsoon season, which was work, they are applied for the total road network. Note the main trigger of all failures observed. that the surveyed roads network may be not representa- We note that physical conditions present in the water- tive of the total network and the remaining roads may shed naturally favor the occurrence of erosion: geology be in worse state: this observation could significantly in- with low soil cohesion, high rates of weathering, steep crease erosion and cost indexes and, therefore the fore- slopes and one of the highest rates of rainfall in Nepal. casts for the future. These conditions are amplified through human activities in the watershed, often attributed to deforestation, agri- Discussion culture or construction. However we note that the forest This study is the result of field work conducted in 2014 cover has actually increased over the past two decades, which resulted in an exhaustive inventory of the road agriculture is in decline, whereas the rural earthen road network of Phewa Watershed (42 % of the total water- network has expanded exponentially. Unplanned road shed road network) and road-related landslides. It is construction without proper drainage tends to accumu- based on observations and measurements, which were at late water and channel it in ways that lead to gullying, times roughly estimated (e.g. such as the age of an ero- greater release of soil volumes, roadside failures and sion event), yet supported by a statistical analysis. The shallow landslides, especially in areas with such intense study evaluated the direct and forecasted impacts of rainfall events, such as Phewa watershed. This situation road construction as measured by volumes of soil re- leads to immediate problems: a high number of road leased by road construction and related direct economic failures, higher road maintenance costs and road cuts Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 18 of 21 Fig. 15 Surveyed roads network in relation to the 2012 land use classification in Phewa watershed. Note that the road-induced erosion areas have been placed on the map and events occur mostly in the forest and agriculture land use types leading to reduced mobility and access to employment, scale of Phewa watershed during field work in 2014 with health care, education, etc. an update of the situation after the extreme rainfall This field work was conducted before the 2015 Gorkha event of July 29, 2015. As this study demonstrated, the earthquake and 2015 monsoon season, with one follow- current mode of road construction which is currently up field visit in September 2015. Although little was ob- occurring in Nepal is largely related with erosion and served due to the 2015 earthquake, the 2015 monsoon shallow landslide processes. Considering the exponential produced exceptionally high amounts of rainfall, includ- growth of rural earthen road networks, we would expect ing one extreme rainfall event on July 29, 2015 an increase of sediments released by roads, These road- (150.4 mm, in 24 h recorded at Panchase metrological induced sediments contribute to the Phewa watershed station and 135.2 mm at Gharelu the weather station sediment budget, however identifying other erosion established by UNIL in 2014). The event created five sources (erosion from terraces, other natural events, debris flows in the Harpan sub-watershed (Simpani vil- etc.) was beyond the scope of this study. Thus, we can- lage in Bhadaure VDC) which caused 9 casualties, the not estimate which percentage of sedimentation is due destruction of at least 10 houses and numerous fields. A to roads as compared to other sources of erosion. As a first inventory reveals at least 50 roadside failures, land- real ‘boom’ of rural road construction has occurred in slides and debris flows with a large number of roads Nepal during the past 15 years, erosion and shallow completely or partially destroyed. A more detailed inven- landslides have obviously increased around the country tory of events is being established. and, as Laban (1979) already warned 30 years ago, ser- ious consideration must be taken if the road network Conclusions continues to grow without more careful construction This study characterized the effect of road construction, methods (Fort et al., 2010; Petley et al., 2007). It would a human-induced environmental phenomenon, at the be clearly more than the 5 % of all landslides due to Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 19 of 21 Table 3 Synthetic results for the cost estimation of road maintenance and agriculture land losses due to road induced erosion phenomenon considering a medium capacity machine, Daewoo Solar 130LC – V Scenarios Part of the watershed roads Event activity network considered Recent (<2 years) Medium (>2 to < 5 years) Old (>5 years) Total (USD) Road maintenance costs 42 % 37186 21577 14520 73283 (USD) LB technology 100 % 88538 51375 34571 174484 Road maintenance costs 42 % 29391 11506 7370 48267 (USD) Mixed technology 100 % 69978 27395 17548 114921 Road maintenance costs 42 % 12615 7407 4948 24970 (USD) EB technology 100 % 30035 17635 11782 59452 Agriculture land 42 % 587 405 1 994 losses costs (USD) 100 % 1398 965 3 2366 TOTAL Scenario LB (USD) 42 % 37773 21983 14521 74277 100 % 89936 52340 34574 176850 Total Scenario Mixed (USD) 42 % 29978 11911 7371 49260 100 % 71376 28360 17551 117287 TOTAL Scenario EB (USD) 42 % 13202 7812 4950 25963 100 % 31433 18599 11785 61818 Results are based on the measurements undertaken along 42 % of the watershed roads network. They are also estimated for 100 % of the network assuming homogenous roadside erosional conditions Table 4 Forecasts of erosion volumes, roadside maintenance volumes and damaged agricultural surfaces, and direct costs for the total road network of the watershed assuming constant annual indexes and exponential roads increase trend Year Total watershed Cumulative value for Damaged areas earthen road network erosion event volume Cumulative value for Cumulative value for Cumulative value for cost length (km) release (m ) volume to remove agriculture land lost of damaged areas with 3 2 or repair (m ) surface (m ) Mixed maintenance (USD) 2014 310 166982 46835 89445 117287 2015 318 167732 47130 90096 118165 2016 328 168732 47524 90965 119337 2017 354 171307 48539 93202 122355 2018 382 174087 49634 95617 125612 2019 412 177088 50816 98224 129128 2020 445 180327 52092 101038 132924 2021 480 183823 53469 104075 137020 2022 518 187597 54956 107353 141442 2023 559 191670 56561 110891 146215 2024 604 196067 58293 114711 151366 2025 652 200813 60162 118833 156927 2026 703 205935 62180 123283 162929 2027 759 211465 64358 128086 169408 2028 820 217433 66710 133271 176401 2029 885 223875 69247 138867 183949 2030 955 230829 71987 144908 192097 The 2014 values were calculated for the total watershed area using the amounts characterized on 42 % of roads network considering homogenous erosional conditions in the watershed. Following year values are cumulative Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 20 of 21 roads as observed by Laban in 1979 for all of Nepal, technical help and general support in this research by helping in the interpretation of results. All authors read and approved the final manuscript. given that the total volume of sediments released by roads today is exponentially higher as compared to the Funding watershed area. This research was conducted as part of the Ecosystems Protecting To minimize the amount of damage caused by rural Infrastructure and Communities project through the International Union for Conservation of Nature, made possible through funding from the road construction, existing road construction policies International Climate Initiative, supported by the German Federal Ministry for should be enforced by providing better technical support the Environment, Nature Conservation, Building and Nuclear Safety (BMUB). to communities, which could take into account environ- Author details ment and intrinsic parameters of the area for construc- Institute of Earth Sciences, University of Lausanne, CH-1015 Lausanne, tion, such as slope angle and geological setting. 2 Switzerland. Institute of Engineering, Department of Civil Engineering, Moreover, simple technologies using low cost and local Tribhuvan University, Patandhoka Road, 44700 Patan, Lalitpur, Nepal. Norwegian Geological Survey, Postal Box 6315Sluppen, NO-7491 Trondheim, resources along the lines of ‘green road’ or ‘eco-safe Norway. road’ approaches to reduce the impacts of rural road construction. Such techniques include roadside drainage Received: 23 November 2015 Accepted: 26 May 2016 to control run-off to avoid high erosion occurring at road interfaces (especially gullies formation) due to lack References of proper water drainage. Furthermore, bio-engineering Aderamo, A.J. 2013. Monitoring of road network growth in developing countries: techniques for slope and soil protection (e.g., planting a case of Ilorin, Nigeria. European International Journal of Science and Technology 2(7): 98–105. local deep-rooted species on bare roadside embank- Agrawala, S., V. Raksakulthai, M. Van Aalst, P. Larsen, J. Smith, and J. Reynolds. ments to reduce soil erosion and stabilize slopes) 2003. Development and Climate Change in Nepal: Focus on Water Resources (Howell, 1999) are well-known by government officials and Hydropower. Paris: OECD. Avouac, J.P. 2015. Mountain Building: From Earthquakes to Geologic but not systematically implemented for new roads. Fi- Deformation. In Treatise on Geophysics, 2nd ed, ed. G. Schubert, 381–432. nally, as current road construction methods require a Oxford: Oxford UNiv. Press. large amount of maintenance, any improved construc- Campbell, D., Church, M. 2003. Reconnaissance sediment budgets for the Lynn Valley, British Columbia: Holocene and contemporary time scales. Canadian tion methods including bio-engineered roads with still Journal of Earth Sciences 40: 701–713. require maintenance yet to a lesser extent (Howell, Cornforth, D.H. 2005. Landslides in practice: Investigation, Analysis, and Remedial/ 1999). Bio-engineering constructions and roadside drain- Preventative options in soils (1st edition). Hoboken, New Jersey: Wiley. Cossart, E., and M. Fort. 2008. Consequences of landslide dams on alpine river ages will obviously be more efficient if routine and pre- valleys: Examples and typology from French Southern Alps. Norsk Geografisk ventive maintenance are planned with the community Tidsskrift-Norwegian Journal of Geography 62: 75–88. (e.g., cleaning drains, cutting and caring for plants, etc.). Crosta, G.B., P. Dal Negro, and P. Frattini. 2003. Soil slips and debris flows on terraced slopes. Natural Hazards and Earth System Sciences 3: 31–42. This type of routine work can be directly carried out by DoR 2015. Road Network Data. Government of Nepal, Department of Roads. hand and will clearly be less costly than the current large http://www.dor.gov.np/road_statistic_2008/Report%20Pages/tables/1.pdf. maintenance requirements, environmental damages and Accessed 20 June 2015. Das Mulmi, A. 2009. Green Road Approach in Rural Road Construction for the losses caused by road inaccessibility. The nuisance and Sustainable Development of Nepal. Journal of Sustainable Development 2(3): damages caused by road failures in Phewa watershed are 149–165. a clear hinder to population mobility and development FEED. 2014. Development of Ecosystem based Sediment Control Technique & Design of Siltation Dam to Protect Phewa Lake: Herpan Khola Watershed Kaski. which are the result of the lack of proper initial road de- Kathmandu: FEED (P) Ltd. sign and maintenance programs. Fleming, B., and J. Puleston Fleming. 2009. A watershed conservation success story in Nepal: Land use changes over 30 years. Waterlines 28(1): 29–46. doi: 10.3362/1756-3488.2009.004. Endnotes Fort, M., E. Cossart, and G. Arnaud-Fassetta. 2010. Hillslope-channel coupling in In total 138 kilometers were surveyed, of which 9 km the Nepal Himalayas and threat to man-made structures: The middle Kali Gandaki valley. Geomorphology 124(3–4): 178–199. were paved roads. Furniss, M.J., Roelofs, T.D., Yee, C.S. 1991. Road construction and Maintenance. Influence of Forest and Rangeland Management on Salmonid Fishes and Additional file Their Habitats. American Fisheries Society, Special Publication 19: 297–323 Hashimoto, S., Y. Ohta, and C. Akiba. 1973. Geology of Nepal Himalayas. Sappdra: Himalayan Committee of Hokkaido University. Additional file 1: Economic estimation of the impact of rural earthen Hovius, N., Stark, C. P., Allen, P. A. 1997. Sediment flux from a mountain belt road construction in Phewa watershed. (DOCX 20 kb) derived by landslide mapping, Geology, 25: 231–234. Howell, J. 1999. Roadside Bio-engineering. Kathmandu: HMG Nepal, Department of Competing interests Roads, Babar Mahal. The authors declare that they have no competing interests. Hungr, O., S. Leroueil, and L. Picarelli. 2013. The Varnes classification of landslide types, an update. Landslides 11(2): 167–194. doi:10.1007/s10346-013-0436-y. Authors’ contributions JICA. 2002. The Development study on the environmental conservation of Phewa KSR supervised this research work and helped to draft the manuscript by Lake. Kathmandu: Silt Consultants (P) Ltd. providing a writing assistance. SD participated in the field work campaign Keller, G., Sherar, J. 2003. Low volumes roads engineering: Best management and helped to acquire the data. LN furnished a technical help to vectorize practices field guide. USDA Forest Service/USAID. National Transportation the aerial photos and the satellite images. MJ, MHD and IP provided a Library. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 21 of 21 Koirala, A., L.N. Rimal, S.M. Sikrikar, U.B. Pradhananga, P.M. Pradhan, et al. 1998. Engineering and environmental geological map of Pokhara valley. Kathmandu: Department of Mines and Geology in Cooperation with Bundesanstalt für Geowissenschaften und Rohstoffe (Federal Institute for Geosciences and Natural Resources, Hannover, Germany). Laban, P. 1979. Landslide occurrence in Nepal. Kathmandu: Integrated watershed management torrent control and Land use development project, Ministry of Forest, Department of Soil and Water Conservation. MoHA. 2009. Nepal Disaster Report 2009, The Hazardscape and Vulnerability. Kathmandu: Ministry of Home Affairs, Government of Nepal. Papathome Koehle, M., and T. Glade. 2013. The role of vegetation cover change in landslide hazard and risk. In The Role of Ecosystems in Disaster Risk Reduction,ed. F. Renaud, K. Sudmeier Rieux, and M. Estrella, 293–320. Tokyo: UNU Press. Petley, D., G.J. Hearn, A. Hart, N. Rosser, S. Dunning, K. Oven, and W. Mitchell. 2007. Trends in landslide occurrence in Nepal. Natural Hazards 43: 23–44. Rimal, B., H. Baral, N. Stork, K. Paudyal, and S. Rijal. 2015. Growing City and Rapid Land Use Transition: Assessing Multiple Hazards and Risks in the Pokhara Valley, Nepal. Land 4: 957–978. Sharma, B.K., K. Timalsina, R. Rai, S.K. Maharjan, A. Joshi, and B. Rakhal. 2013. Biodiversity Resource Inventory, Ecosystem Assessment of Bhadaure Tamagi VDC, Kaski: An Ecosystem-based Adaptation in mountain Ecosystem in Nepal. Lalitpur: IUCN Nepal. Sharma, S., and M.L. Maskay. 2009. Community Participation and Environmental Protection in the Construction of Mountain Roads: Promotion of the “Green Road” Approach in Nepal. New York: Transportation and Communication bulletin for Asia and Pacific, No. 69, Participatory approach to transport infrastructure development, United Nation. Sikrikar, S.M., L.N. Rimal, M. Kerntke, S. Jäger, et al. 1998. Landslide hazard zonation map of Phewa Lake catchment area, Pokhara, Nepal. Kathmandu: Department of Mines and Geology in Cooperation with Bundesanstalt für Geowissenschaften und Rohstoffe (Federal Institute for Geosciences and Natural Resources, Hannover, Germany). Sthapit, K.M., and M.K. Baila. 1998. Sediment Monitoring of Phewa Lake. Watershed Management and Environmental Science. Kathmandu: Institute of Forestry. Sudmeier Rieux, K. 2011. On Landslide Risk, Resilience and Vulnerability of mountain Communities in Central-Eastern Nepal. Lausanne: Université de Lausanne, Faculté des Géosciences et de l’Environnement, Institut de Géomatique et d’Analyse de Risque. Validya, K.S. 1985. Accelerated erosion and landslides-prone zones in the Himalayan region. In Environmental Regeneration in Himalaya: Concepts and Strategies. Nainital: Central Himalayan Environmental Association. Validya, K.S. 1987. Environmental Geology. New Delhi: Indian Context, Tata-McGraw-Hill. World Bank. 2013. Nepal road sector assessment study: Main report. Kathmandu: World Bank. Submit your manuscript to a journal and benefi t from: 7 Convenient online submission 7 Rigorous peer review 7 Immediate publication on acceptance 7 Open access: articles freely available online 7 High visibility within the fi eld 7 Retaining the copyright to your article Submit your next manuscript at 7 springeropen.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geoenvironmental Disasters Springer Journals

Rural earthen roads impact assessment in Phewa watershed, Western region, Nepal

Loading next page...
 
/lp/springer-journals/rural-earthen-roads-impact-assessment-in-phewa-watershed-western-7aV0WPTmcy

References (44)

Publisher
Springer Journals
Copyright
Copyright © 2016 by The Author(s).
Subject
Environment; Environment, general; Earth Sciences, general; Geography, general; Geoecology/Natural Processes; Natural Hazards; Environmental Science and Engineering
eISSN
2197-8670
DOI
10.1186/s40677-016-0047-8
Publisher site
See Article on Publisher Site

Abstract

Background: This work describes current research being conducted in the Phewa watershed, near Pokhara in Nepal’s Middle hills, a moist sub-tropical zone with the highest amount of annual rainfall in Nepal (4,500–5,000 mm). The main purpose of this study is to focus on the increase and impact of rural earthen road construction in the Phewa watershed as part of land use changes over 30 years in one of Nepal’s most touristic regions. Research methods were interdisciplinary and based on a combination of remote sensing, field observations and discussions with community members. The study compared 30 year-old aerial photos with current high resolution satellite images to detect changes in the watershed road network. Secondly, 42 % of the watershed’s roads (138 km) were surveyed in order to inventory and quantify shallow landslide events. Using an erosion-characteristics grid, their main features were measured (location, size, type and dimensions of damaged areas, etc.) and a Geographic Information System data base was created. We then estimated economic impacts of these events in terms of direct agriculture lands losses and road maintenance. Results: Results of the remote sensing analysis demonstrate that the road network increase is following an exponential trend, which enables us to predict future watershed road network scenarios. Field work investigations have demonstrated that unplanned road excavations are producing mainly embankment shallow planar soil slides and/or gullying which primarily affect roads themselves, wiping them out and blocking vehicle circulation, and secondly, destroying or burying agriculture lands. Along the 138 km surveyed, we calculated an amount of soil material released of about 70,000 cubic meters, which amounted to 99 cubic meters per kilometer annually. Of 179 cases of roadside erosion processes sampled, about 85 % directly impact roads or agricultural lands. Conclusion: The current mode of road construction which is currently occurring in Nepal is largely related with erosion and shallow landslide processes. Considering the exponential growth of rural earthen road networks, we would expect an increase of sediments released by roads and serious consideration must be taken if roads continue to be made without more careful methods. Through simple technologies using low cost and local resources along the lines of ‘green road’ or ‘eco-safe road’ approaches, it may be possible to reduce the impacts of rural road construction. Keywords: Rural earthen roads, Erosion, Sedimentation, Shallow landslides, Nepal * Correspondence: geoffroy.leibundgut@gmail.com Institute of Earth Sciences, University of Lausanne, CH-1015 Lausanne, Switzerland Full list of author information is available at the end of the article © 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 2 of 21 Background rural road construction has become a priority for the main Nepal has the greatest altitude variation on earth which village and district authorities, in other words the Village ranges from 60 to 8,848 m above sea level. The high Development Committees (VDCs) and the District Devel- mountains are the result of the collision between the opment Committees (DDCs) (World Bank, 2013). Many Indian tectonic plate and the Eurasian tectonic plate communities are collecting their own funds to rent bull- (Avouac, 2015). The uplift is currently active and makes dozers and build rural roads without proper technical and the country prone to earthquakes. The 147,000 square geological expertise (Das Mulmi, 2009). The road con- kilometers of this landlocked mountainous country are struction boom and lack of enforcement of government divided into five physiographic units from north to regulations has led to a low quality Nepal rural network, south: the high Himalaya, high mountains, the middle with roads which often do not provide year round access hills region, the Churia or Siwalik range and the Terai re- (World Bank, 2013). Slope cut during rural earthen gion in the plains (Fig. 1). About 80 % of Nepal is moun- roads construction certainly affect the frequency of tainous or hilly and 20 % are plains area of Terai, which shallow and in some cases larger landslides and local are located at the northern Ganga Basin (Agrawala et al., soil erosion/deposition processes, although the exact per- 2003). With about 200 km separating the plains area of centage of road induced shallow landslides is unknown Terai (south) from the high Himalayas (north), the coun- (Fort et al., 2010; Furniss et al., 1991). This human- try presents great topographic diversity. The country induced phenomenon directly impacts communities and therefore has high climate and vegetation diversity which infrastructure (Sudmeier-Rieux, 2011). varies from tropical in the Terai to High-mountains cli- In 1979, Laban demonstrated that 5 % of Nepal’s land- mate (snow and ice) in the north (Agrawala et al., 2003). slides (above 50 m ) were induced by roads or trails and Annual rainfall varies from 1,000 to 5,000 mm. As the formulated a serious warning about this high value since middle mountain range (around 2000 m.a.s.l) are the first the percentage of the land covered by roads network mountain barriers for the monsoon clouds moving north was at the time very small. The road and transportation from the Bay of Bengal, around 80 % of this precipitation network has clearly increased dramatically over the past occurs during this season, which lasts from June to three decades and we should expect a drastic increase of September (MoHA, 2009). Poor rock mass conditions road-induced landslides in Nepal since 1979 (Laban, makes Nepal’s slopes prone to fluvial and landslide 1979; Petley et al., 2007; Fort et al., 2010). Green roads, erosion. Moreover, geological context of the Middle or the use of simple engineering structures for drainage hills region is dominated by metasedimentary rock and slope stabilization, have been promoted in Nepal for with metamorphosed granites in upper part and, car- several decades (Das Mulmi, 2009). They are character- bonate and clastic sedimentary rocks, which are weak ized by the use of locally available deep-rooted grass spe- (Hashimoto et al. 1973). cies based on a participatory approach from planning to According to the World Bank (2013), the road net- implementation, and environment consideration. work has tripled in Nepal in the past decade. In 2013, The Phewa watershed is actually an area that has expe- the Strategic road network (SRN), which is managed at rienced an exponential increase of rural roads network the central level, was approximately about 11,000 km (as it will be developed farther in this paper) partly due to while the Local Road Network (LRN), which is managed the presence of the adjacent town of Pokhara and the at local level, was about 60,000 km (World Bank, 2013). presence of the Phewa Tal, one of the most prominent lake Roughly half of the SRN and more than 95 % of the in Nepal. Both have contributed to attract people for agri- LRN is unpaved, giving a paved network of 8,000 km culture and tourism. Therefore, we selected this watershed only or 11 % of the total road network (World Bank, as the study area to quantify roadside erosion events. 2013). The SRN road density increased from 3.22 km of The objective of this paper is to share observations on roads per 100 km in 1998 to 8.49 km of roads per the current state of rural earthen roads and their im- 100 km (DoR, 2015). According to the World Bank pacts in terms of an acceleration of erosion rates and re- (2013), the Nepal road density is actually high as com- lated costs for road maintenance and agriculture land pared to other mountainous countries and largely due to losses in the Phewa watershed. Sedimentation rate and the trend in the past decade of opening up new roads in costs results could provide a basis for further studies on Nepal. The government has placed a great emphasis on conventional versus green road construction. developing the roads and the transportation infrastruc- ture as a real means of development for the rural popu- Description of the study area lation (World Bank, 2013). Phewa watershed is located in the western part of the Moreover, according to Sudmeier-Rieux (2011), the re- Pokhara valley of Kaksi District in the Western Develop- sult of the 2008 Decentralization Act produced authority ment Region of Nepal. The watershed lies within latitude and budgets transfer to local governments. Consequently, of 28° 11’ 41.7” to 28° 17’ 26.0” north and longitude of Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 3 of 21 Fig. 1 Nepal’s physiographical regions, and Phewa watershed location and boundaries as considered in the study. Note on the bottom map the erosion areas measured along the surveyed roads. The surveyed roads network makes part of the total roads network (Source for the top map: Forest Resource Assessment Nepal, 2015; Source for the bottom map: Pleiades Satellite Imagery, 2013) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 4 of 21 83° 47’ 53.2” to 83° 58’ 04.3”east. It covers an area of lithology is mostly gritty phyllite (46 % of the total area) about 111 square kilometers (Fig. 1). and fine grain massive quartzite (25 % of the total area), The outlet of the Phewa Lake is situated at 784 m in the south-western part of the watershed (Harpan sub- above sea level (Sthapit and Baila, 1998) where the high- watershed). The watershed is also comprised of colluvial est point is marked by the Panchase ridge summit at in homogeneous deposits (silt, sand, gravel and boul- 2517 m.a.s.l. The confluence of two main tributaries ders) and debris flow deposits. The flood plain is mainly Sidhane and Adheri, form the Phewa river drains into gravel, sand, silt and clay deposits. the Phewa lake. These two sub-watersheds cover 24.5 The watershed climate is based on annual monsoon and 28.4 % respectively of the total watershed area events, which bring more than two thirds of the annual (Sthapit and Baila, 1998). According to our data, in rainfall between June to September. This season is char- 2013, Phewa lake covered 3.3 % of the total watershed acterized by intense rainfall; events of 300 mm in 24 h area. This figure is similar to recently published data, are not uncommon in this area (MoHA, 2009). Between which determined the lake’s surface to be 3.96 % in 2013 1982 and 2012 the mean annual rainfall at Lumle me- (Rimal et al., 2015). The capacity of the lake was estimated teorological station (located at North West of the water- as 42.18 million m in 1998, and the annual average sedi- shed, 1740 m.a.s.l.) was about 5,506 mm. At the Pokhara mentation rate in the lake was about 180,000 m (Sthapit Airport meteorological station (827 m.a.s.l.), it is about and Baila, 1998). If a constant rate is considered, 80 % of 3,875 mm. These extreme natural events contribute of the lake’s storage will be silted up in about 190 years ac- the natural degradation of steep terrain including the cording to the same authors. Moreover, Phewa Lake is triggering of landslides or flash flooding. also experiencing accelerated eutrophication, land en- croachment, and massive invasion of water hyacinth and Methods exotic carp fish species (FEED, 2014; JICA, 2002). Data available and field investigation According to the 2012 land use classification based on The present survey was conducted through data available a 2012 RapidEye satellite image (5 m resolution) under- and data field collected during November 2014 campaign. taken by UNEP (Sharma et al., 2013), land use in Phewa Available data: watershed was comprised of 41 % productive (agricul- ture/grassland terrain), 49 % forest (trees and bushes), ▪ Digitalized topographical map of 1996 from the and 5 % water bodies (lake, river and swamp area), 3 % Government of Nepal, NGIIP, Survey Department, built up area, 1 % sand area (near rivers and lake). The Ministry of Land Reform and Management. area was part of a watershed land use management pro- This digitalized topographical data set is directly usable gram, which started in the 1970s (Fleming and Puleston on a Geographic Information System (GIS) program. Fleming, 2009). As far as the difficulties to reduce ero- The data are compounded by Shapefiles of sion are concerned, the program focused on the conver- transportation and hydrographic networks, sion of “critical landscapes” (Fleming and Puleston administrative and built zone boundaries, 20 m Fleming, 2009: 38), such as degraded shrubs, grazing topographical contours lines and land cover areas. land and unmanaged forests, to managed community ▪ 1979 aerial photography and 2013 satellite images. forests or managed pasture. According to the authors, The 2013 images come from Pleiades Satellite Imagery watershed forest land increased from 28 % to 36 % be- and are of 2 m resolution for the 4 bands (multispectral) tween 1978 and 2006 while the terraced arable land and of 0.5 m resolution for the panchromatic. The 1979 remained constant. Forests managed by Community aerial photos come from the Department of survey of the Forest User Groups exceeded 60 % of total forests in the Government of Nepal. Both fully cover the watershed. watershed in 2006. Forest (and bush) cover steeper ▪ 1998 Geological map of Pokhara valley published by areas, possibly due to improved community forest devel- Department of Mines and Geology in cooperation with opment in the watershed, and could play a role in the Bundasanstalt für Geowissenschaften und Rohstoffe - protection of soil from mass movement and failures Federal Institute for Geosciences and Natural (Papathome Koehle and Glade 2013). Resources, Hannover, Germany (Koirala et al., 1998). The watershed lithology is compounded by intensively ▪ 2012 Land use base map prepared by UNEP for the weathered rocks and weak soils, highly prone to erosion Ecosystem-based Adaptation in Mountain Ecosystem and shallow landslides (Agrawala et al., 2003). According project in Nepal. The classification was undertaken on to the geological map made in 1998 by the Department a 2012 RapidEye satellite image (5 m resolution) using of Mines and Geology of Nepal in cooperation with the segmentation and object classification method on the Bundesanstalt für Geowissenschaften und Rohstoffe - eCognition software tool (Sharma et al., 2013). Forest Federal Institute for Geosciences and Natural Resources, areas, agriculture areas, and water body and sand areas Hannover, Germany (Sikrikar et al., 1998), the watershed were classified. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 5 of 21 During field work in November 2014, about 42 % of Nevertheless, not all the events directly affected the the total watershed roads network (138 km) was sur- surrounding environment, infrastructure or persons. veyed to detect and inventory erosion processes directly induced by road construction. We created an erosion- We then created an Excel sheet summarizing the field characteristics grid to characterize and classify the obser- work measurements. One important step of the inven- vations and measurements undertaken during field tory work was to measure relevant dimensions of each work. All these data were gathered in a Geographic failure event with a laser distance meter to propose volu- Information System (GIS) project using ArcMap 10.2 metric estimation using basic 3D geometric shapes (see (ArcGis program) in the projected geographic coordin- further for the volume estimation methodology). For ate system WGS 1984 UTM, Zone 44 North. each damaged area, the road deposit volume and the agriculture land lost surface were also characterized. Road network detection and comparison for the period Moreover, by representing event GPS location points 1979–2013 (as Shapefiles) in the GIS project, it was possible to rec- Three sources of data enabled to detect and compare ord some relevant information that could be directly in- changes over the period 1979–2013 in the Phewa water- tegrated in the database. For example, we noted the shed road network. The 1979 aerial photographs were 2012 land use and slope angle for each landslide event digitalized and geo-referenced by our care using the pro- point location. gram ENVI 5.0. Roads were manually detected respect- The main idea of creating an erosion-characteristic ively on the 1979 aerial photos and on the 2013 satellite database paired with a GIS project of the Phewa water- images. We define roads as the transportation ways for shed was to be able to generate statistics about simple vehicles in general (black top and earthen roads) but volume estimations of the road side events. earthen roads are clearly the major type of transporta- tion way in the watershed (both in 1979 and 2013). Sha- Classification details pefiles from the 1996 digitalized topographical map Fluvial and hill slope erosion processes were inventoried, provided a detailed overview of the entire road and path and differentiated considering their main conditioning network, which was re-classified to differentiate the two. factor (natural or human induced). The landforms iden- The main goal of this remote sensing work is to com- tified differ in type, shape, road location and potential pare the length of the road network over the three past damages involved. decades in order to characterize road network trends of The following list/sub-sections details each relevant the watershed. parameter of the erosion-characteristics grid: Road induced erosion and damaged area events 1. Road-induced erosion events inventory Event type: The mass balanced method of excavation About 138 km of the watershed’s roads were surveyed (‘cut-and-fill’) is commonly used for road construction over 3 weeks of field work to measure main features of on hill slope (Keller and Sherar, 2003;Cornforth, the erosion processes (shallow landslides, gullies) in- 2005). The ‘cut-and-fill’ design (Fig. 3) could be at duced by road construction (see Fig. 1). origin of landslides and road embankment failures: “an We directly separated this inventory work into three approximately equal cut-and-fill cross-section can (i) different general classes (Fig. 2): undermine the upper slope, causing it to fail, (ii) overload the downhill slope, causing it to fail, or (iii) 1) Road-induced erosion events: we measured their cause the entire slope to become an active landslide.” position along the road and elevation using a (Cornforth, 2005: 12). Following the definition of Geographic Positioning System (GPS), their Hungr et al. (2013), roadside erosion events observed dimensions using a laser distance meter; we also in Phewa watershed are commonly shallow and planar noted failure event activity/age, the material soil slides. Upper and lower roadside embankments involved, etc. can be affected. Moreover, gullies induced by runoff, 2) Classification of the road through observations and especially during monsoon, on the ‘cut-and-fill’ design discussions with community members. road embankments are also an erosion phenomenon 3) Potential damaged areas due to road-induced erosion observed in the watershed. Erosion events were events. Two damaged areas were considered: the classified as follows (Fig. 3): road itself that could be wiped out or blocked, and agriculture land that could be buried or destroyed. Each one represents an economic impact and, in some a. Embankment shallow planar soil slide or shallow cases, could injure persons living in the watershed. soil slip: Shallow soil volume sliding along an Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 6 of 21 Fig. 2 General architecture of the erosion-characteristics grid summarizing field’s measurements. The three main classes - erosion events, surveyed roads and damaged area - are inter-related and are characterized for each point measured inclined slip and planar surface. They are embankments and form gullies. Moreover, this characterized by small size and thickness, with phenomenon could be at origin of channelization in volumes up to few cubic meters (Crosta et al., 2003). the lower slopes; it is thus considered to be gully They are occurring directly on the upper or lower erosion induced by road construction. The road embankment. consequences are deposits produced by bedload b. Extended shallow planar soil slide: Corresponds to a and/or small debris-flows. larger shallow planar slide and doesn’t occur only on the roadside embankment but also on the Activity surrounding hillsides. They are characterized by This parameter is important and is mainly signaled by length up to 20 m and width up to 10 m, and the age of deposits. Numerous signs indicate activity, involve larger quantity of material. As far as is such as the rock deposit patina and weathering, if the difficult to be sure to consider road as the driver of deposit is re-vegetated, etc. this larger event, the few cases measured here were clearly triggered by the roads. ▪ Recent: less than 2 years; c. Gully: In some cases, road construction affects water ▪ Medium: between 2 and 5 years; drainage that could cut upper or lower road ▪ Old: more than 5 years. Fig. 3 Cut-and-fill design of road construction and related slides and failures events potentially occurring in Phewa watershed (modified from Keller & Sherar, 2003) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 7 of 21 1.5 Event volumes V = 0.05 × S (with S the landslide surface and V the The volume of material failures was estimated as the vol- landslide volume). ume of sediment removed assuming interpreted ‘pre ero- Slide surface could be easily approximated as an sional’ land surface and simple geometric 3D erosion ellipse depending on length and width measured shapes (Cossart & Fort, 2008; Campbell & Church, 2003). during field work. The erosion event type is, defined as a horizontal shape in two dimensions of the erosion area used as base Slope to calculate the volume. The formulae used to estimate Contour lines of the 1996 digitalized topographical map erosion volume are based on the basic dimensions mea- were used to create a raster DEM 30 m resolution sured in the field with a laser distance meter (length, through interpolation tools on ArcGis 10.2 (based on width and thickness). For example, shallow soil slide discretized thin plate spline techniques). The DEM has events are largely defined by a parallelepiped volume of enabled then to compute slope angle values. Due to the material of a given width (W), length and thickness slid- small DEM resolution, the slope layer represents only ing along a plane (L); in this case the 2D base shape is a the angle value of the regional slope of the watershed. parallelogram. Figure 4 illustrates the flowchart detailing We considered 4 slope classes: [0°; 13°], [13°; 27°], [27°; the volume calculation process. 39°] and [39°; 90°]. The list below details the volumes estimation method according to each event type. Basic dimensions and 2D Lithology base shapes are illustrated by Fig. 5. This parameter was processed by joining the 1998 Geo- logical Map, digitalized and vectorized to be available on ▪ Embankment shallow planar soil slide or shallow soil a GIS program, with event points shapefiles. slip: The 2D base shape could be of 2 types: semi-ellipse or parallelogram. Volume is calculated as the product Land use type of the base shape area (depending on length and width This parameter was processed by joining the 2012 Land parameters) by the thickness of the sediment volume Use layer with event points shapefiles. removed. ▪ Gully: The 2D base shape section is defined here as a Surveyed roads semi-ellipse (similar to the U-shape section of a gully) A large part of the watershed road was traced using a and volume estimation is the product of the surface Global Positioning System (GPS) during the field work and length of the gully. and then classified according to the observations and ▪ Extended shallow planar soil slide: When it was discussion with the community members. Each event difficult to measure the slide thickness on the field, we also contained road information according to its location used an empirical relation (Hovius et al., 1997): (as it is classified). VOLUME ESTIMATION PROCESS Erosion event FOR EROSION EVENTS Embankment shallow planar soil Extended shallow planar Gully Event type slide or shallow soil slip soil slide Ellipse Semi- Parallelogram U-shape 2D base shape ellipse Parallelogram surface × Empirical relation 1/2 ellipse surface × Semi-cylinder with Event volume estimation thickness thickness ellipse base section (Hovius et al., 1997) Equation Fig. 4 Volume estimation process for erosion events. L, W and T correspond respectively to length, width and thickness of the event and are illustrated in the Fig. 5 Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 8 of 21 Fig. 5 Basic dimensions as measured on field used to determine: (a) embankment shallow planar soil slide event volumes according to the two types of base shapes, (b) gully erosion event volumes Surface type: Volumes to remove or repair: The volumes that are removed or used to repair are estimated as ▪ Paved road: Black top surfacing. parallelepipeds of given length, width and thickness. ▪ Earthen road: Earth surfacing. They were measured during field work. In some cases, the full volume of the involved material is Road access: considered to be removed; this depends largely on the observations undertaken on field (Fig. 6). ▪ Drivable: By a 4 wheel drive car. This parameter is directly used to estimate the ▪ Not-drivable: Because of a bad road surfacing or direct cost of maintenance. deposit/failure cutting the road. B. Affected agriculture land Type of damage: Road maintenance: a. Failure: Failure of terrace areas destroying crop area due to a collapse/slip of the upper road ▪ Maintained: Shows numerous signs of maintenance embankment. (protection and stabilization infrastructure, drainage b. Deposit: Burying of crop areas by soil material system, etc.). However, this does not necessarily reflect coming from upper-sides. how well the road is maintained. Crop type: Type of crop cultivated before the ▪ Unmaintained: The road is left in its actual state. damages occurred. Dimension of agriculture surface lost: The productive Damaged areas crop surface lost is considered as parallelogram of a given Characterizing the damaged area was undertaken mainly length and width measured during field work (Fig. 6). to estimate direct costs of the road maintenance and the This parameter is directly used to estimate the agriculture land losses according to several scenarios. direct cost of agriculture lands losses. A. Affected roads The flowchart (Fig. 6) illustrates a simple way to esti- Type of damage: mate roadside volume to remove or repair and agriculture a. Deposit: Burying of road surface by up-slope soil surfaces directly affected by road induced erosion events. material. Maintenance will consist of removing deposits from the road. Results b. Cut in road shoulder: Road surface failure due to Slope angle distribution of the watershed a collapse/slip of the lower road embankment. We have divided the Phewa lake in four classes accord- Maintenance will consist of filling in the eroded ing the slope; each class has a different land use (Fig. 7): part with soil deposits. The first class [0–13 °] corresponds to the flood plain, Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 9 of 21 VOLUME AND SURFACE Damaged ESTIMATION PROCESS FOR area DAMAGED AREAS Agriculture Damaged area type Road land By material By material By cut in road By failure of deposition on deposition on Type of damage shoulder terrace road field Mainly Others event Mainly gully or Mainly shallow deposition Shallow soil slip Type of event types Shallow soil slip soil slip downslope of road Volume of the Surface of fallen Surface of deposit Volume of the Volume or surface Total volume of affecting the eroded cut terrace buried field the event road segment calculated as a calculated as a calculated as a estimation calculated as a parrallelogram parallelogram parallelepiped parallelepiped Fig. 6 Volume and surface estimation process for damaged area. Note that simple shapes, such as parallelepiped and parallelogram, are used to estimate volumes and surfaces fluvial terraces, distal parts of alluvial fans and the lake. than 39°; this area corresponds to rock slopes with low Most of the settlements, schools, and agriculture and vegetation and steep terrain covered by forest. grazing lands, occupy this area, accounting for 25 % of It is important to note that the 5 m resolution of the the total watershed and part of Pokhara city, which lies 2012 RapidEye satellite image allowed a rough classifica- in the watershed. The main reason for the land use of tion of the land use type of the watershed. Boundaries this sector is the soil conditions, better ability for terra- between two classes are not well defined. Also, an iso- cing, water drainage, and access to roads. The second lated element of a land use type could be not detected. class [13°–27°], also corresponds to low relief area. It is However, in this case, the 30 m DEM that has allowed primarily covered by agriculture terrains and settlements computing only the angle value of the regional slope of whereas the third class [27°; 39°] is mostly covered by the watershed, it is relevant to assess the slope angle dis- forest. Finally, 3 % of the total watershed area is steeper tribution in relation to this rough land use classification. Fig. 7 Slope classes distribution (see below for calculation) according to the 2012 land use type. It has been standardized to take into account the same proportion of surface covered by each slope class area in relation to the total watershed area; in that way, land use type area for a given slope classes based on the same surface calculation and so, we represent here the weight of the different land use types for each slope class in the same proportion (Source: 2012 Land use base map prepared by UNEP for the Ecosystem Based Adaptation in Mountain Ecosystem project in Nepal. The classification was undertaken on a 2012 RapidEye satellite image (5 m resolution) using segmentation and object classification method on the eCognition software tool (Sharma et al., 2013)) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 10 of 21 Road network increase and trends is physically limited and the road network length will In 1979, there were no black top roads in Phewa water- therefore tend to a maximum value. The model would be shed whereas in 1996 about 5 km of the road network therefore characterized by an exponential trend reaching was paved. A classification between earthen roads and an asymptote corresponding to the maximum value. black top roads for the year 2013 is based on the road Aderamo (2013) has also characterized growth of cities survey conducted during field work, where most of road networks (Ilorin, Nigeria) using aerial photos and im- black top roads of the Phewa watershed (about 10 km) ages and he showed that it conforms to a logistic curve. were surveyed. In addition, we included a section of the However, to make earthen roads length predictions for the Pokhara road network belonging to the Phewa water- next decade, we assume here that in the short run, the shed to reach a total black top road network length of growth trend will not have reached its maximum value. about 18 km for the year 2013. The total length of the Regarding the total road network growth according to 2013 earthen roads in the watershed was calculated as slope angle classes (Fig. 10), the curve fitting of the road the difference between the total number of transporta- construction indicates the largest growth coefficient tion ways detected on the 2013 satellite image and the (superscript of the exponential relation) for the range black top roads length calculated as explained previ- [27°; 39°], which correspond to medium steep areas. ously. Following Fig. 8, the watershed black top roads Then, follow the ranges [13°; 27°] and [0°; 13°], which length increase is more defined by a 2nd order poly- correspond respectively to the low steep areas and the nomial relation whereas the earthen roads length flood plain. Both show a quite similar road construction grew exponentially. We then focused only on the increase coefficient. Moreover, the largest increase quan- earthen roads network, as we will demonstrate further titatively occurs for the range [13°; 27°]. This means that earthen roads are the main trigger of erosion events the major road network length is located in low steep in the watershed. areas and in the flood plain, with respectively 54 and The watershed earthen roads network length has 36 % of the total watershed road network. This observa- considerably increased over the past three decades, tion is clearly related to the land use type in these slopes from about 23 km in 1979 to 310 km in 2013, due to classes; agriculture activity and settlements location have the new earthen roads boom in Nepal (Fig. 9). The played an important role in developing transportation increasing trend is clearly defined by an exponential infrastructure in that areas. If the road network growth curve and we can consider the trend line (in Fig. 8, of these three slope classes follows an exponential trend, plot (B)) as representative of the earthen roads net- we can note that results for the range [13°; 27°] are not work length vs year, as the correlation coefficient is really relevant as the correlation coefficient is low. Fi- close to 0.95. nally, road length growth in the range [39°; 90°] is better Nevertheless, the transportation network growth model defined by a linear relation and is little represented in would be better defined by a logistic curve as the territory the total watershed road network. ab Black top roads network increase in Earthen roads network increase in Phewa Phewa watershed over the past 3 decades watershed over the past 3 decades 20 350 y = 0,0135x + 0,0632x + 3E-14 0,0764x y = 19,399e R² = 0,949 0 0 010 20 30 40 010 20 30 40 Year (from 1979) Year (from 1979) Fig. 8 Increase of (a) black top roads network length and, (b) earthen roads network length in Phewa watershed over the past 3 decades. The horizontal scale starts in the year 1979 (0 corresponds to the year 1979). Note that for the black top roads increase case (plot (a)), the growth is defined by a 2nd order polynomial trend. Only 17 km of black top roads built more than 30 years illustrates well the few considerations in developing a paved road network in Nepal. For the earthen roads increase case (plot (b)), the growth is well defined by an exponential trend and clearly illustrates the earthen roads construction “boom” occurring in Nepal over the past decade Length (km) Length (km) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 11 of 21 Fig. 9 1979, 1996 and 2013 roads networks in Phewa watershed. Note that the watershed roads networks length has increased significantly over the period of 1979–2013 Erosion events and damaged areas: from inventory to partially or totally and in 21 cases, agriculture lands were statistics affected. Most of the shallow landslides identified locate Along the 129 km of earthen roads surveyed, 179 ero- in the northern slope of the watershed (Fig. 11), where sive landforms induced by road construction were iden- the road is cutting perpendicularly a steeper part of the tified. In 155 cases, the road network was washed out watershed. In the south, fewer roads were built up, but Road network length increase according slope degree classes [0° ; 13°] [13° ; 27°] [27° ; 39°] 0,0764x y = 10,334e 140 [39° ; 90°] R² = 0,908 0,0743x y = 8,8072e R² = 0,9947 0,1149x y = 0,6108e R² = 0,9933 y = 0,0377x R² = 0,7529 0 5 10 15 20 25 30 35 40 Year (from 1979) Fig. 10 Total roads length increase trends according to slope classes. The horizontal scale starts in the year 1979 (0 corresponds to the year 1979) Length (km) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 12 of 21 Fig. 11 General view of Phewa watershed mapped with erosion type (with illustrations) measured on roads surveyed during 2014 field work. Extended and embankment shallow planar soil slides and gully are the three type of road-induced erosion events observed in the watershed Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 13 of 21 also fewer surveys were carried out. The contrasting to failure processes but also due to land down-slope buried landslide occurrence between north and south reflects by deposits. Extended shallow planar soil slides are also the link between road construction and shallow landslide playing an important role in burying agriculture lands. occurrence. More erosion events were detected in the medium range Results focus on counting the number of landforms in activity than in the recent one, as 3 years are included in addition to estimating the volume relation between ero- medium range while 2 years are included in the recent sion events, damaged areas and surveyed roads, first for range. Nevertheless, total volume released for the medium the total area of the watershed and, secondly according and the recent ranges activity, respectively led back to 1 to zonal parameters: slope angle classes, lithology and year, give an amount of about 12,500 m /year for each land use type. one; this result traduces that the event activity parameter is relevant for the recent and medium ranges. However, Main statistical results due to high land use changes and quick vegetation growth, Most of the roads on the Phewa lake are earthen roads old range activity total volume is not relevant due to the (Fig. 12 and Table 1). Along them, landslides and gullies difficulty in detecting older events. Otherwise, the dam- developed in relationship with the cut of slopes during aged areas cases were mostly induced by recent events. road construction and when by diverting run off. More- Amounts have been estimated for the total watershed over, in the same way, the majority of cases measured roads network: along 310 km of earthen roads, 166,982 m occur on the unmaintained road segment, wiping it of material is released, 46,835 m need to be maintained them out and leaving the road undrivable in more than and 89,445 m of agriculture lands is lost. These values let two third of cases. Nevertheless, about the half of the assume than road-induced erosional conditions are well network surveyed clearly showed signs of maintenance. homogenous along watershed roads. As mentioned above, Given that the watershed research primarily targeted ac- the remaining watershed roads which were not been sur- cessible roads by jeep, the remaining watershed road veyed during the field work are in worse state than the sur- network would obviously being worse state and certainly veyed one: this observation could significantly increase unmaintained, that means that the roads surveyed dur- erosion events volume and damaged areas cases. ing field work may be not representative of the total Assuming homogenous erosional conditions along watershed roads network. earthen roads in the watershed, we can estimate kilomet- Table 2 and Fig. 13 summarize total results of erosion ric indexes for erosion events volume, roadside mainten- events volume, roadside maintenance volume and agri- ance volume and agriculture land lost surface. Therefore, culture land lost surface amounts. It also classifies these this human-induced phenomenon is responsible for re- values according to type and activity of erosion events. leasing 544 m of soil per kilometer of earthen road. To Along the 129 km of earthen road surveyed, an esti- take into account the time dimension, we can use totals mated 70,133 m were directly released along roadsides. estimated according to the event activity parameter: about 3 3 Of this total amount, 28 % or 19,671 m of material de- 198 m per kilometer of earthen road is released over posited need to be maintained and 37,567 m of agricul- 2years (total ‘recent’ events volume per 129 km), which ture land were buried or destroyed. gives an annual volumetric erosion rate per kilometer of 3 −1 −1 Extended shallow planar soil slide is the mechanism about 99 m .km .year .The ‘recent’ value was used to releasing the most of material volume even if the num- calculate the rate because it is considered as the most ex- ber of cases measured is lower in relation to embank- haustive events volume measurements in the watershed. ment shallow planar soil slide or shallow soil slip (13 The same assumptions was used to determine annual cases detected for extended shallow planar soil slide roadside maintenance volume per kilometer and annual against 150 cases for embankment shallow planar soil agriculture land lost surface per kilometer which amounts 3 −1 −1 2 −1 −1 slide or shallow soil slip). This observation could be ex- respectively to 39 m .km .year and 86 m .km .year . plained by the fact that extended shallow planar soil This Figure is comparable to the Figure given by Validya 3 −1 −1 slides are clearly a much larger phenomenon than the (1985, 1987) of 55 m .km .year of debris produced by fewer embankment slips defined as embankment shallow rural earthen roads (cited by Sharma and Maskay, 2009). planar soil slide or shallow soil slip and involve greater quantities of soil material. Moreover, 16 cases of signifi- Influence of various parameters on roadside erosion and cant gully erosion were detected and involved lower damaged areas quantities of soil material. Nevertheless, regarding the Additional analyses were undertaken on key watershed pa- damaged areas, embankment shallow planar soil slide or rameters related to erosion events and damaged areas shallow soil slip are clearly responsible for most of the (measured along 42 % of the watershed roads network). road maintenance needed and also represent a large part The study therefore focused on the lithology, on the 2012 of agriculture land losses, especially in terraces subject land use and on slope angle classes with the goal to Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 14 of 21 Fig. 12 Kilometric distribution of erosion events and damaged area cases along the surveyed roads. Note that to avoid the length factor, the number of cases is standardized in the distance unit, e. g. by kilometer of road. Results are clearly linked with the state of roads; erosion events (as defined in the methodology section) and induced damage areas (e.g. the road itself and the agriculture lands) cases occur most of time along unmaintained earthen roads calculate the total erosion events volume, the total roadside a. Influence of lithology maintenance volume and the total agriculture land lost sur- face per classes of slope, per classes of land use and per Erosion events occur in colluvial soil and in the fine classes of lithology. However, it is likely that greater vol- grained and massive quartzite formations (which are umes or surface value are to be found on a larger area of located in the in Harpan sub-watershed - south-west detection. Similarly, we standardized the erosion events and of Phewa watershed), constituting 40 % of the total maintenance volumes, the surface areas lost and the road volume released and 60 % of material volume affecting length by the surface of class of detection. Finally we plot- road network (Fig. 14, plot (a)). According to the ted the ratio volume/km or surface/km for each class in second revision of the geological and soil cover map order to represent only the influence of the parameter, not (Sikrikar et al., 1998), colluvial soils are indeed taking into account the surface value of detection. considered the most erosion prone of the geologic Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 15 of 21 Table 1 Erosion events and damaged area cases number measured along 42 % of the watershed roads network Surveyed road Erosion events Damaged areas Road Agriculture land Classification Length (km) Ratio Cases number Cases number Cases number Total road 138.2 100 % 179 155 21 Earthen road 128.9 93 % 179 155 21 Black top road 9.3 7 % 0 0 0 Drivable 124 90 % 144 122 12 Undrivable 14.2 10 % 38 33 9 Maintained 72.1 52 % 39 29 2 Unmaintained 66.1 48 % 140 126 19 Results are given also according to surveyed roads classification. Note that the majority of the watershed road network surveyed is earthen surfaced and drivable formations of the watershed. According to Sikrikar b. Influence of slope angle et al. (1998) and as observed in the south-west of the Phewa watershed, the quartzite formation is prone The erosion events detected and road networks to deep gully formation, rock fall, rock slides and surveyed up to a slope angle of 39° clearly make the wedge failures involving rather rough material (rocky higher slope class the most prone to erosion events deposits, as gravels and boulders). Some deep and and volumes of soil released and thus induced road large gullies, due to road induced modification of damaged cases (Fig. 14, plot (b)). The volume water drainage at the origin of channelization in the released per kilometer is clearly increasing with the steep lower-slope, were also observed in this area. slope angle, which is realistic considering that slope Moreover, Gritty phyllite formations, covering roughly angle is a main triggering mechanism for mass half of the total watershed area, is also susceptible to movement and slope failure. Nevertheless, seeing mass movement and slope failure and represents that the road network increased coefficient in the about 20 % of the total material volume released and range [39°; 90°] is very low (Fig. 10), we cannot 25 % of material volume affecting road network. expect a real significant increase of the material Agriculture land lost surface are mostly (almost 70 %) released in slope area up to 39° for the next decades. occurring in Gritty phyllite because most agriculture The real road impact will be more significant in the areas are located in this formation (south aspect of slope ranges [13°; 27°] and [0°; 13°]. Regarding the watershed). agriculture land lost surface, it appears theoretically Table 2 Total volume released on roads surveyed, total volume that need maintenance and total agriculture land lost surface amounts along 42 % and 100 % of the watershed roads network Inventory of soil volume losses Erosion events Damaged areas on roads surveyed Road Agriculture land Erosion events Volume Volume to remove Surface buried or 3 3 2 number release (m ) or repair (m ) destroyed (m ) TOTAL Along 42 % of the 179 70133 19671 37567 roads network Along 100 % of the 426 166982 46835 89445 roads network Event type (42 % of Shallow soil slip 150 22469 18576 20457 the roads network) Gully 16 8817 356 0 Extended shallow 13 38847 740 17110 planar soil slide Event activity (42 % of Recent (<2 years) 72 25580 9998 22,202 the roads network) Medium 78 37178 5767 15,320 (>2 to < 5 years) Old (>5 years) 29 7375 3907 45 Results are also classified according to event type and event activity (for measurements on 42 % of the watershed roads network) Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 16 of 21 Fig. 13 Distribution of erosion events volume released, roadside maintenance volumes and agriculture land lost surface according to event parameters along 42 % of the watershed’s road network that it occurs on range [13°; 27°], area mostly Road length, roadside erosion volume and direct costs covered by agriculture land. forecasts In addition to the volume of erosion created by road con- c. Influence of the 2012 land use struction, this study also calculated direct economic losses Erosion events and roadside slope failures are mainly for rural earthen roads due to roadside maintenance and occurring in forest lands (60 %) and agriculture direct agriculture land lost surface. This analysis was lands (40 %) because these two land use type are undertaken considering community-based road mainten- covering the main part of low to highly steep areas ance scenarios using three different types of maintenance: (Fig. 14, plot (c)). A map has been prepared to put labor based, equipment based and mixed technology into evidence this observation: most of the roads (combination between labor-based and equipment-based network where erosion events are occurring is maintenance). A more detailed description of the method- located in these two main land use classes (Fig. 15). ology used is published in Additional file 1. Even if forests can be considered as protection against Table 3 summarizes data for a medium excavator erosion and landsliding, in this case, slope steepness is (Daewoo Solar 130LC – V) and incorporates the cost of the main factor explaining why more failures are land losses for the three scenarios. Results are given also occurring in the forest area than in agriculture area. according the event age (recent, medium and old). Con- Slope angle appears to be the most relevant parameter sidering the use of ‘mixed technology’ for road mainten- for explaining the erosion phenomenon induced by ance, economic impacts due to rural earthen road earthen roads in the watershed. Note finally that the construction in Phewa watershed amount to 49,260 USD agriculture surface lost found in the forest land use for 42 % of the network. Assuming homogenous ero- type is not relevant and could be translated by a lack sional condition along roadside in the watershed, the of precision of the land use classification undertaken total cost for 100 % of earthen roads will rise to 117,287 on a 5 m resolution satellite image in relation to the USD and the annual earthen roads economic impact per −1 −1 field work measurements. Earthen roads are of about kilometer would be 116 USD.km .year . 3 m maximum of wideness and measured GPS points Furthermore, road length, erosion and cost values could easily be taken into account into the bad land were forecasted in Table 4 for the period 2014–2030. It use class due to the rough resolution of the provides forecasts assuming constant rural roads growth classification. and homogenous erosional conditions in the watershed. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 17 of 21 Fig. 14 Kilometric distribution of the erosion events volume released, roadside maintenance volume and agriculture land lost surface according to: (a) lithology, (b) slope angle classes and, (c) the 2012 land use type. Note that all erosion and failure events occurred in the three main types of lithology of high susceptibility for landslides. Volume released and maintenance volume clearly growth with the slope angle, whereas land lost surface keep on the range [13°; 27°] as it mainly covered by agriculture land For instance, in 2030, 955 km of rural earthen road in impacts. Results demonstrate that road-induced erosion the watershed will produce a total erosion volume of in the watershed is mostly occurring on roadside em- about 231,000 m , which represent an economic impact bankments, which produce small and shallow soil fail- of about 192,000 USD. Even if erosion and cost indexes ures which often make the road drivable for no more were characterized for 42 % of the watershed road net- than one year until the next monsoon season, which was work, they are applied for the total road network. Note the main trigger of all failures observed. that the surveyed roads network may be not representa- We note that physical conditions present in the water- tive of the total network and the remaining roads may shed naturally favor the occurrence of erosion: geology be in worse state: this observation could significantly in- with low soil cohesion, high rates of weathering, steep crease erosion and cost indexes and, therefore the fore- slopes and one of the highest rates of rainfall in Nepal. casts for the future. These conditions are amplified through human activities in the watershed, often attributed to deforestation, agri- Discussion culture or construction. However we note that the forest This study is the result of field work conducted in 2014 cover has actually increased over the past two decades, which resulted in an exhaustive inventory of the road agriculture is in decline, whereas the rural earthen road network of Phewa Watershed (42 % of the total water- network has expanded exponentially. Unplanned road shed road network) and road-related landslides. It is construction without proper drainage tends to accumu- based on observations and measurements, which were at late water and channel it in ways that lead to gullying, times roughly estimated (e.g. such as the age of an ero- greater release of soil volumes, roadside failures and sion event), yet supported by a statistical analysis. The shallow landslides, especially in areas with such intense study evaluated the direct and forecasted impacts of rainfall events, such as Phewa watershed. This situation road construction as measured by volumes of soil re- leads to immediate problems: a high number of road leased by road construction and related direct economic failures, higher road maintenance costs and road cuts Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 18 of 21 Fig. 15 Surveyed roads network in relation to the 2012 land use classification in Phewa watershed. Note that the road-induced erosion areas have been placed on the map and events occur mostly in the forest and agriculture land use types leading to reduced mobility and access to employment, scale of Phewa watershed during field work in 2014 with health care, education, etc. an update of the situation after the extreme rainfall This field work was conducted before the 2015 Gorkha event of July 29, 2015. As this study demonstrated, the earthquake and 2015 monsoon season, with one follow- current mode of road construction which is currently up field visit in September 2015. Although little was ob- occurring in Nepal is largely related with erosion and served due to the 2015 earthquake, the 2015 monsoon shallow landslide processes. Considering the exponential produced exceptionally high amounts of rainfall, includ- growth of rural earthen road networks, we would expect ing one extreme rainfall event on July 29, 2015 an increase of sediments released by roads, These road- (150.4 mm, in 24 h recorded at Panchase metrological induced sediments contribute to the Phewa watershed station and 135.2 mm at Gharelu the weather station sediment budget, however identifying other erosion established by UNIL in 2014). The event created five sources (erosion from terraces, other natural events, debris flows in the Harpan sub-watershed (Simpani vil- etc.) was beyond the scope of this study. Thus, we can- lage in Bhadaure VDC) which caused 9 casualties, the not estimate which percentage of sedimentation is due destruction of at least 10 houses and numerous fields. A to roads as compared to other sources of erosion. As a first inventory reveals at least 50 roadside failures, land- real ‘boom’ of rural road construction has occurred in slides and debris flows with a large number of roads Nepal during the past 15 years, erosion and shallow completely or partially destroyed. A more detailed inven- landslides have obviously increased around the country tory of events is being established. and, as Laban (1979) already warned 30 years ago, ser- ious consideration must be taken if the road network Conclusions continues to grow without more careful construction This study characterized the effect of road construction, methods (Fort et al., 2010; Petley et al., 2007). It would a human-induced environmental phenomenon, at the be clearly more than the 5 % of all landslides due to Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 19 of 21 Table 3 Synthetic results for the cost estimation of road maintenance and agriculture land losses due to road induced erosion phenomenon considering a medium capacity machine, Daewoo Solar 130LC – V Scenarios Part of the watershed roads Event activity network considered Recent (<2 years) Medium (>2 to < 5 years) Old (>5 years) Total (USD) Road maintenance costs 42 % 37186 21577 14520 73283 (USD) LB technology 100 % 88538 51375 34571 174484 Road maintenance costs 42 % 29391 11506 7370 48267 (USD) Mixed technology 100 % 69978 27395 17548 114921 Road maintenance costs 42 % 12615 7407 4948 24970 (USD) EB technology 100 % 30035 17635 11782 59452 Agriculture land 42 % 587 405 1 994 losses costs (USD) 100 % 1398 965 3 2366 TOTAL Scenario LB (USD) 42 % 37773 21983 14521 74277 100 % 89936 52340 34574 176850 Total Scenario Mixed (USD) 42 % 29978 11911 7371 49260 100 % 71376 28360 17551 117287 TOTAL Scenario EB (USD) 42 % 13202 7812 4950 25963 100 % 31433 18599 11785 61818 Results are based on the measurements undertaken along 42 % of the watershed roads network. They are also estimated for 100 % of the network assuming homogenous roadside erosional conditions Table 4 Forecasts of erosion volumes, roadside maintenance volumes and damaged agricultural surfaces, and direct costs for the total road network of the watershed assuming constant annual indexes and exponential roads increase trend Year Total watershed Cumulative value for Damaged areas earthen road network erosion event volume Cumulative value for Cumulative value for Cumulative value for cost length (km) release (m ) volume to remove agriculture land lost of damaged areas with 3 2 or repair (m ) surface (m ) Mixed maintenance (USD) 2014 310 166982 46835 89445 117287 2015 318 167732 47130 90096 118165 2016 328 168732 47524 90965 119337 2017 354 171307 48539 93202 122355 2018 382 174087 49634 95617 125612 2019 412 177088 50816 98224 129128 2020 445 180327 52092 101038 132924 2021 480 183823 53469 104075 137020 2022 518 187597 54956 107353 141442 2023 559 191670 56561 110891 146215 2024 604 196067 58293 114711 151366 2025 652 200813 60162 118833 156927 2026 703 205935 62180 123283 162929 2027 759 211465 64358 128086 169408 2028 820 217433 66710 133271 176401 2029 885 223875 69247 138867 183949 2030 955 230829 71987 144908 192097 The 2014 values were calculated for the total watershed area using the amounts characterized on 42 % of roads network considering homogenous erosional conditions in the watershed. Following year values are cumulative Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 20 of 21 roads as observed by Laban in 1979 for all of Nepal, technical help and general support in this research by helping in the interpretation of results. All authors read and approved the final manuscript. given that the total volume of sediments released by roads today is exponentially higher as compared to the Funding watershed area. This research was conducted as part of the Ecosystems Protecting To minimize the amount of damage caused by rural Infrastructure and Communities project through the International Union for Conservation of Nature, made possible through funding from the road construction, existing road construction policies International Climate Initiative, supported by the German Federal Ministry for should be enforced by providing better technical support the Environment, Nature Conservation, Building and Nuclear Safety (BMUB). to communities, which could take into account environ- Author details ment and intrinsic parameters of the area for construc- Institute of Earth Sciences, University of Lausanne, CH-1015 Lausanne, tion, such as slope angle and geological setting. 2 Switzerland. Institute of Engineering, Department of Civil Engineering, Moreover, simple technologies using low cost and local Tribhuvan University, Patandhoka Road, 44700 Patan, Lalitpur, Nepal. Norwegian Geological Survey, Postal Box 6315Sluppen, NO-7491 Trondheim, resources along the lines of ‘green road’ or ‘eco-safe Norway. road’ approaches to reduce the impacts of rural road construction. Such techniques include roadside drainage Received: 23 November 2015 Accepted: 26 May 2016 to control run-off to avoid high erosion occurring at road interfaces (especially gullies formation) due to lack References of proper water drainage. Furthermore, bio-engineering Aderamo, A.J. 2013. Monitoring of road network growth in developing countries: techniques for slope and soil protection (e.g., planting a case of Ilorin, Nigeria. European International Journal of Science and Technology 2(7): 98–105. local deep-rooted species on bare roadside embank- Agrawala, S., V. Raksakulthai, M. Van Aalst, P. Larsen, J. Smith, and J. Reynolds. ments to reduce soil erosion and stabilize slopes) 2003. Development and Climate Change in Nepal: Focus on Water Resources (Howell, 1999) are well-known by government officials and Hydropower. Paris: OECD. Avouac, J.P. 2015. Mountain Building: From Earthquakes to Geologic but not systematically implemented for new roads. Fi- Deformation. In Treatise on Geophysics, 2nd ed, ed. G. Schubert, 381–432. nally, as current road construction methods require a Oxford: Oxford UNiv. Press. large amount of maintenance, any improved construc- Campbell, D., Church, M. 2003. Reconnaissance sediment budgets for the Lynn Valley, British Columbia: Holocene and contemporary time scales. Canadian tion methods including bio-engineered roads with still Journal of Earth Sciences 40: 701–713. require maintenance yet to a lesser extent (Howell, Cornforth, D.H. 2005. Landslides in practice: Investigation, Analysis, and Remedial/ 1999). Bio-engineering constructions and roadside drain- Preventative options in soils (1st edition). Hoboken, New Jersey: Wiley. Cossart, E., and M. Fort. 2008. Consequences of landslide dams on alpine river ages will obviously be more efficient if routine and pre- valleys: Examples and typology from French Southern Alps. Norsk Geografisk ventive maintenance are planned with the community Tidsskrift-Norwegian Journal of Geography 62: 75–88. (e.g., cleaning drains, cutting and caring for plants, etc.). Crosta, G.B., P. Dal Negro, and P. Frattini. 2003. Soil slips and debris flows on terraced slopes. Natural Hazards and Earth System Sciences 3: 31–42. This type of routine work can be directly carried out by DoR 2015. Road Network Data. Government of Nepal, Department of Roads. hand and will clearly be less costly than the current large http://www.dor.gov.np/road_statistic_2008/Report%20Pages/tables/1.pdf. maintenance requirements, environmental damages and Accessed 20 June 2015. Das Mulmi, A. 2009. Green Road Approach in Rural Road Construction for the losses caused by road inaccessibility. The nuisance and Sustainable Development of Nepal. Journal of Sustainable Development 2(3): damages caused by road failures in Phewa watershed are 149–165. a clear hinder to population mobility and development FEED. 2014. Development of Ecosystem based Sediment Control Technique & Design of Siltation Dam to Protect Phewa Lake: Herpan Khola Watershed Kaski. which are the result of the lack of proper initial road de- Kathmandu: FEED (P) Ltd. sign and maintenance programs. Fleming, B., and J. Puleston Fleming. 2009. A watershed conservation success story in Nepal: Land use changes over 30 years. Waterlines 28(1): 29–46. doi: 10.3362/1756-3488.2009.004. Endnotes Fort, M., E. Cossart, and G. Arnaud-Fassetta. 2010. Hillslope-channel coupling in In total 138 kilometers were surveyed, of which 9 km the Nepal Himalayas and threat to man-made structures: The middle Kali Gandaki valley. Geomorphology 124(3–4): 178–199. were paved roads. Furniss, M.J., Roelofs, T.D., Yee, C.S. 1991. Road construction and Maintenance. Influence of Forest and Rangeland Management on Salmonid Fishes and Additional file Their Habitats. American Fisheries Society, Special Publication 19: 297–323 Hashimoto, S., Y. Ohta, and C. Akiba. 1973. Geology of Nepal Himalayas. Sappdra: Himalayan Committee of Hokkaido University. Additional file 1: Economic estimation of the impact of rural earthen Hovius, N., Stark, C. P., Allen, P. A. 1997. Sediment flux from a mountain belt road construction in Phewa watershed. (DOCX 20 kb) derived by landslide mapping, Geology, 25: 231–234. Howell, J. 1999. Roadside Bio-engineering. Kathmandu: HMG Nepal, Department of Competing interests Roads, Babar Mahal. The authors declare that they have no competing interests. Hungr, O., S. Leroueil, and L. Picarelli. 2013. The Varnes classification of landslide types, an update. Landslides 11(2): 167–194. doi:10.1007/s10346-013-0436-y. Authors’ contributions JICA. 2002. The Development study on the environmental conservation of Phewa KSR supervised this research work and helped to draft the manuscript by Lake. Kathmandu: Silt Consultants (P) Ltd. providing a writing assistance. SD participated in the field work campaign Keller, G., Sherar, J. 2003. Low volumes roads engineering: Best management and helped to acquire the data. LN furnished a technical help to vectorize practices field guide. USDA Forest Service/USAID. National Transportation the aerial photos and the satellite images. MJ, MHD and IP provided a Library. Leibundgut et al. Geoenvironmental Disasters (2016) 3:13 Page 21 of 21 Koirala, A., L.N. Rimal, S.M. Sikrikar, U.B. Pradhananga, P.M. Pradhan, et al. 1998. Engineering and environmental geological map of Pokhara valley. Kathmandu: Department of Mines and Geology in Cooperation with Bundesanstalt für Geowissenschaften und Rohstoffe (Federal Institute for Geosciences and Natural Resources, Hannover, Germany). Laban, P. 1979. Landslide occurrence in Nepal. Kathmandu: Integrated watershed management torrent control and Land use development project, Ministry of Forest, Department of Soil and Water Conservation. MoHA. 2009. Nepal Disaster Report 2009, The Hazardscape and Vulnerability. Kathmandu: Ministry of Home Affairs, Government of Nepal. Papathome Koehle, M., and T. Glade. 2013. The role of vegetation cover change in landslide hazard and risk. In The Role of Ecosystems in Disaster Risk Reduction,ed. F. Renaud, K. Sudmeier Rieux, and M. Estrella, 293–320. Tokyo: UNU Press. Petley, D., G.J. Hearn, A. Hart, N. Rosser, S. Dunning, K. Oven, and W. Mitchell. 2007. Trends in landslide occurrence in Nepal. Natural Hazards 43: 23–44. Rimal, B., H. Baral, N. Stork, K. Paudyal, and S. Rijal. 2015. Growing City and Rapid Land Use Transition: Assessing Multiple Hazards and Risks in the Pokhara Valley, Nepal. Land 4: 957–978. Sharma, B.K., K. Timalsina, R. Rai, S.K. Maharjan, A. Joshi, and B. Rakhal. 2013. Biodiversity Resource Inventory, Ecosystem Assessment of Bhadaure Tamagi VDC, Kaski: An Ecosystem-based Adaptation in mountain Ecosystem in Nepal. Lalitpur: IUCN Nepal. Sharma, S., and M.L. Maskay. 2009. Community Participation and Environmental Protection in the Construction of Mountain Roads: Promotion of the “Green Road” Approach in Nepal. New York: Transportation and Communication bulletin for Asia and Pacific, No. 69, Participatory approach to transport infrastructure development, United Nation. Sikrikar, S.M., L.N. Rimal, M. Kerntke, S. Jäger, et al. 1998. Landslide hazard zonation map of Phewa Lake catchment area, Pokhara, Nepal. Kathmandu: Department of Mines and Geology in Cooperation with Bundesanstalt für Geowissenschaften und Rohstoffe (Federal Institute for Geosciences and Natural Resources, Hannover, Germany). Sthapit, K.M., and M.K. Baila. 1998. Sediment Monitoring of Phewa Lake. Watershed Management and Environmental Science. Kathmandu: Institute of Forestry. Sudmeier Rieux, K. 2011. On Landslide Risk, Resilience and Vulnerability of mountain Communities in Central-Eastern Nepal. Lausanne: Université de Lausanne, Faculté des Géosciences et de l’Environnement, Institut de Géomatique et d’Analyse de Risque. Validya, K.S. 1985. Accelerated erosion and landslides-prone zones in the Himalayan region. In Environmental Regeneration in Himalaya: Concepts and Strategies. Nainital: Central Himalayan Environmental Association. Validya, K.S. 1987. Environmental Geology. New Delhi: Indian Context, Tata-McGraw-Hill. World Bank. 2013. Nepal road sector assessment study: Main report. Kathmandu: World Bank. Submit your manuscript to a journal and benefi t from: 7 Convenient online submission 7 Rigorous peer review 7 Immediate publication on acceptance 7 Open access: articles freely available online 7 High visibility within the fi eld 7 Retaining the copyright to your article Submit your next manuscript at 7 springeropen.com

Journal

Geoenvironmental DisastersSpringer Journals

Published: Jun 6, 2016

There are no references for this article.