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

Learn More →

Wetland mapping and evaluating the impacts on hydrology, using geospatial techniques: a case of Geba Watershed, Southwest Ethiopia

Wetland mapping and evaluating the impacts on hydrology, using geospatial techniques: a case of... GEOLOGY, ECOLOGY, AND LANDSCAPES INWASCON https://doi.org/10.1080/24749508.2021.1953744 RESEARCH ARTICLE Wetland mapping and evaluating the impacts on hydrology, using geospatial techniques: a case of Geba Watershed, Southwest Ethiopia Mintesnot Berhanu , Karuturi Venkata Suryabhagavan and Tesfaye Korme School of Earth Sciences, Addis Ababa University, Addis Ababa, Ethiopia ABSTRACT ARTICLE HISTORY Received 15 April 2021 Wetlands are one of the world’s most important ecosystems threatened by man. This investi- Accepted 4 July 2021 gation explores the use of Landsat TM and OLI imageries with SRTM DEM for mapping them. Mapping and monitoring of wetlands is done with 86.66% accuracy. As a result, a loss of KEYWORDS −1 21,400 ha yr could be noted. Due to this, differences were also found in the water quality and Agriculture; Climate; groundwater level between the degraded and un-degraded areas. As most rivers within the Wetland loss; Water quality; watershed punctuated from the wetlands, their existence was determined based on the Water level hydrological function. Wetland degradation occurred mainly due to climate and agricultural changes over time. Thus, geospatial techniques employed in the present study have proved very useful in simplification and visualization of wetland monitoring. Introduction cover 12 million ha in Tanzania and Kenya (Sakane Wetlands are among the most significant, multi- et al., 2011). The majority of wetlands lost historically functional, and productive ecosystems on the were drained or filled to create agricultural land. earth (Davidson et al., 2019; Mengesha, 2017). A comprehensive inventory of wetlands in Ethiopia Primarily, wetlands can provide essential environ- is not yet done, but wetlands are estimated to cover mental services, including storing floodwater, redu- about 2% of the country’s surface area (Amsalu & cing peak runoff, recharging groundwater, filtering Addisu, 2014; Mengesha, 2017; Terefe, 2017). impurities in water, carbon storage, and also eco- Wetlands are a common feature of the landscapes in logically serve as breeding grounds and critical the highlands of southwestern Ethiopia, particularly habitat for several species of plant communities, Western Wellega and Illubabor (Abebe & Geheb, invertebrates, fish, and wildlife (Chouari, 2021; 2003; Dixon et al., 2021). Despite recognizing their Kaplan & Avdan, 2017; Wang & Weng, 2014; Wu, many uses by people, their ecological services to 2018). Since the early 20th century, the global wet- humankind, and their environmental significance, land ecosystem has confronted significant chal- Ethiopian wetlands are under severe pressure and lenges, including rapid economic development and degradation. The loss of these wetlands is devastating habitat destruction. Wetlands cover approximately to several wetland-dependent endemic species 8% of the world’s land surface and contain 20% of (Bezabih & Mosissa, 2017). This is due to improper the global terrestrial carbon (Dixon et al., 2021; extraction and misconceptions forwarded to wetlands, Mitsch & Gosselink, 2007). However, despite the the health of the wetlands is continuously decreasing environmental degradation faced by wetlands, there from time to time which puts in doubt their existence is an increasing demand for ecosystem services they soon (Abebe & Geheb, 2003; Woldu & Yeshitela, provide (Suding, 2011). By 2050, global water 2003). demand is projected to increase by 55% (Terefe, In Ethiopia, wetland destruction and alteration 2017). To meet this growing demand, the services saw as an advanced development mode, even at the of wetlands must be valued appropriately, or water government level (Dixon & Wood, 2003). This security risks will rapidly increase. The need for indicates that wetlands and their value remain little information supporting wetland management is understood (Gebresllassie et al., 2014). Convention multi-scalar worldwide, and the challenge demands on Biological diversity of Ethiopia 4th report urgent and consistent wetland monitoring mechan- describe; Fogera marsh has been changed to the ism assessments to guide policymaking. rice field, Sululta marsh is distributed to investors, The wetlands are important in East Africa as they ELFORA PLC has transformed the Chefa wetland make up more than 80% of the total wetland area and in South Wello to farmland, and these are only CONTACT Karuturi Venkata Suryabhagavan drsuryabhagavan@gmail.com School of Earth Sciences, Addis Ababa University, 1176, Addis Ababa, Ethiopia © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON). This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 M. BERHANU ET AL. a few examples of wetland degradation in Ethiopia. Material and methods Lake Tana is loaded with silt and invasive water Study area hyacinth because the wetland vegetation in the surrounding catchments was destroyed and used The study area is located within the Geba River basin for agriculture. The wetlands used to stop silt and that constitutes the Baro River basin in Ethiopia at plant nutrition that is discharged to the lake have around 600 km from Addis Ababa. The area is been converted to a rice paddy. The recent total bounded by latitude 7°45ʹ00”−8°36ʹ00”N and longi- drying up of Lake Alemaya and the precarious tude 35°20ʹ00”−36°11ʹ00”E covering a total area of existence of Lake Abijata are clear evidence of the 7,125.35 km (Figure 1). The watershed geology com- looming danger on the wetland ecosystem (Amsalu prises Precambrian, metamorphosed volcanics, intru- & Addisu, 2014). sives and sediments in the north and southwest, In recent years, several studies have focused on Paleozoic and Mesozoic sediments in the center, and wetland mapping in Ethiopia’s southwest regions. localized quaternary formation along the valleys of the The complete drainage and cultivation of wetlands major rivers. The southwestern highland valley bot- have become common phenomena (Hailu, 1998; tom wetlands are developing through time to change Dixon & Wood, 2007). For instance, approximately external geological, geomorphological, and climatic one-third of the total valley bottom wetlands have conditions. The formation of ancient impermeable come under cultivation for growing food crops from quaternary and tertiary bedrock in the study area 1974 to 1983 (Hailu, 1998). More severely from this, could play a significant role in forming wetlands. The approximately 20% of the total wetlands in Illubabor elevation of the region ranges from 780 to 2661 m have been cultivated during 1986−1998, and this above mean sea level at the mountains areas. intensity increased to 35% in 1999 (Hailu, 2005). Climatic conditions in the study area are quite diverse The loss and degradation of these critical resources due to considerable differences in altitude and relief. (wetlands) need urgent mapping and monitoring of About 80% of the annual rainfall occurs in the Kirmet the resources and determining potential restoration (rainy) season from June to September, and 63% of the areas (Gerjevic, 2004). To better manage and con- annual rainfall is the peak recorded in July and serve wetland resources, we need to know the dis- August. The temperature varies from a minimum tribution and extent of wetlands and monitor their average of 6.5°C to 32°C. dynamic changes (Kaplan & Avdan, 2019; Wu, 2018). Remote sensing offers the opportunity to Data and methodology map and inventory wetlands rapidly and consis- tently, irrespective of the geographic location. The study used remote sensing data to map changing Combining remote sensing and geographic informa- wetland degradation trends in the Geba Watershed tion system approach integrated with in-situ mea- from 1985 to 2018. Landsat multi-temporal imageries surement provides an advanced tool in detecting of 1985, 2000, 2018, and Digital Elevation Model and identifying degraded wetland resources at regio- (DEM) were used for the study area (Table 1). Using nal and local scales. Wetland mapping involves most the TauDEM tool, the watershed area was generated often using satellite data and aerial photos due to automatically from the SRTM DEM, and two Landsat their remoteness and inaccessibility (Baker et al., scenes of 170/054 and 171/054 path and rows filled the 2006). Imageries from the Landsat, Aster, SPOT, study area. After radiometric and atmospheric nor- Sentinel-2, IRS, IKONOS, QuickBird, and malization, the ArcGIS mosaicking tool has been WorldView provide the necessary spatial and tem- used to generate a mosaic of two scenes covering the poral resolution to implementing effective wetland study area. The generated map was projected to World monitoring. The satellite data consisted of high reso- Geodetic System (WGS) 1984 Universal Transverse lution (1−4 m) and medium resolution (10−30 m) Mercator (UTM) zone 37 N. The images were multispectral imagery. The spatial and spectral reso- acquired through the USGS Earth explorer www. lution satellite data has significantly improved wet- usgs.gov. Different dated satellite images were of vary- land and habitat mapping (Jensen, 2007; ing pixel size, and resampling was done to obtain the Suryabhagavan, 2017). Therefore, this study aims to same pixel size in all the satellite imagery used. find how degraded wetland resources have been Meteorological data such as monthly temperature mapped. The spatial relationship between wetland and average monthly rainfall were acquired from degradation and its hydrological and related impacts Ethiopia’s National Meteorological Agency (NMA). on the surrounding environment was assessed using Based on the wetlands’ potential resources, three geospatial tools. A recent wetland degradation map kebeles of the study area were visited in February was produced for future mitigation and management and March. A field survey was conducted for ground- purposes to sustain the development at regional and truth data, land-cover, water samples were collected national levels. using handheld GPS (Global Positioning System) to GEOLOGY, ECOLOGY, AND LANDSCAPES 3 Figure 1. Location map of the study area. Table 1. Specifications of satellite data used. Path spectrally (supervised component) (Foody, 2002). In Satellite Sensor and Spatial this study, the Level 1 land-use and land-cover classi- ID type Year Acquisition date row resolution fication system was used and adapted from United Landsat 5 Thematic 1985 20 − 02 − 1985 170 30 m mapper 05 − 04 − 1988 and States Geological Survey (USGS) (James & Randolph, (TM) 171/ 2011). Most pixel-based classifications tend to utilize spectral information at individual pixels and poten- Landsat 5 Thematic 2000 27 − 02 − 2000 170 30 m mapper 17 − 01 − 2000 and tially textural information extracted from neighboring (TM) 171/ pixels. Pixel-based classification can highlight noise, Landsat 8 OLI_TIRS 2018 27 − 01 − 2018 170 30 m salt-and-pepper effects and ignore important topolo- 18 − 01 − 2018 and gical and semantic information in the images 171/ (Blaschke, 2010). Based on the maximum likelihood classification algorithm, six different classes were iden- tified; agricultural land, forest, shrubland, wetland, Gumro tea plantation, and settlement. enhance satellite image and wetland mapping classifi - cation. The flow chart of the methodology used is Vegetation indices NDVI and NDWI given in Figure 2. This study’s method holds pixel- based supervised classification, index-based classifica - The Normalized Difference Vegetation Index (NDVI) tion (NDVI and NDWI), SRTM wetness index, and and Normalized Difference Water Index (NDWI) are image enhancement methods algorithm was used to selected from this study’s satellite data image proces- identify and delineate wetland extent. sing. Several researchers have used both NDVI and NDWI for wetlands change detection (Das, 2017; Dehm et al., 2019; Li et al., 2019; Nsubuga et al., Pixel-based classification of wetland 2017; Xu, 2006); and they all found that there are Thematic mapping from satellite data can be defined significant changes of both indices values during as grouping together cases (pixels) by their relative their study periods. Several developed spectral vegeta- spectral similarity (unsupervised component) to allo- tion and water indices can highlight wetlands and cate instances based on their similarity to a set of extract water bodies while efficiently suppressing predefined classes that have been characterized (Xu, 2006). They can be calculated using the following 4 M. BERHANU ET AL. Figure 2. Methodological flow of the process involved in wetland mapping. Equations (1) and (2), respectively (Jones, 2015; Lane depends on the topographic conditions of a region. et al., 2014). The TWI was directly derived from Equation (3) (Beven & Kirkby, 1979). ðNIR RedÞ NDVI ¼ (1) � � ðNIRþ RedÞ a TWI ¼ 1n (3) tanðbÞ where NIR represents the Near Infrared Band and Red represents the Red band. where a is the local up-slope contributing area drain- ing a point per unit contour length, b is the local slope ðGreen NIRÞ NDWI ¼ (2) in radians. ðGreenþ NIRÞ Accuracy assessment Topography Wetness Index An accuracy assessment for the supervised wetland Wetlands are easily confused with other upland habi- classification was done for the 1985 image using tat cover types, such as forests because these classes ERDAS Imagine @ 15. From the classifier, 80, 75, show overlapping spectral signatures (Ozesmi & and 95 points were generated randomly for 1985, Bauer, 2002). Thematic maps of both primary and 2000, and 2018 supervised images, respectively. Each secondary key hydrological parameters attribute of point had a specific color tone and pixel value recog- slope, drainage, flow direction, flow accumulation, nized by the software itself when the data sets were stream order, Topographic Wetness Index (TWI) trained during supervised wetland classification. were derived from Shuttle Radar Topography These values were considered as reference values. All Mission (SRTM) 90 m DEM using d-infinity algo- the randomly generated points were then identified by rithm in TauDEM hydrological tools as described by the user and assigned to different classes. This process (Islam et al., 2008; Tarboton, 1997). TWI were used to was done for the three supervised classification images reduce the error of commission with upland land (i.e., 1985, 2000, and 2018). The correctly identified covers since the wetlands’ occurrence strongly points were considered as classified values. An Error GEOLOGY, ECOLOGY, AND LANDSCAPES 5 matrix and Kappa statistics were also generated from Results and discussion this reference and classified data from the report sec- Wetland classification tion of ERDAS Imagine @ 15 software. Accuracy assessment was conducted by collecting 250 in situ Initially, the NDVI and NDWI were considered for ground truth points (GTP), which were systematically water detection, as these indices have already been distributed throughout the study areas in the accessi- proven suitable for this purpose in previous studies ble parts. Overall accuracy was calculated from the (Das, 2017; Dehm et al., 2019). Supervised classifica - error matrix by dividing the sum of the entries that tions were performed for separating the valley wet- make major diagonal by the total number of examined lands from the other land-cover types. The pixel-based pixels. Kappa coefficient of the agreement was also classification was conducted on the study images’ mul- calculated by using the following Equations (4), (5), tiple segmentation levels using the maximum likeli- and (6), respectively (Kulawardhana et al., 2007). hood classification. The results were compared with corresponding pixel-based classifications to delineate Overall accuracyð%Þ ¼ � � the wetlands’ extent and boundaries in the study area 0 0 0 0 Total no: of GT points of class X that falling on class X � 100 (4) (Figure 3). However, wetland spectral mixing with Total no: of GT points of classX other land-cover regions was observed. Error of omissionð%Þ ¼ � � 0 0 0 0 Total no: of GT points of class X not falling on class X � 100 (5) Total no: of GT points of classX Vegetation indices NDVI and NDWI The finest of these indices (NDVI and NDWI) pro- Error of commissionð%Þ ¼ vided the only accuracy of less than 19%, with high Total no: of GT points of other classes falling on class X ð Þ� 100 (6) levels of errors of omissions and commissions in the 0 0 Total no: of GT points of class X present study. With several iterations, threshold values of NDVI and NDWI were done to extract wetland Kappa coefficient ¼ from other land covers. This index combination’s gen- Observed accuracy Change agreement (7) eral applicability was not proved to be very high since 1 Chance agreement the minimum water probability value was kept at 19% for all areas in the watershed. Adding the NDVI to one where observed accuracy is determined by diagonal in of these water indices reduced detection rates over error matrix, and chance agreement incorporates off- water bodies that contained some vegetation. Map of diagonal (sum of the product of row and column totals NDVI and NDWI based on threshold values of −0.25 for each class). Figure 3. Wetland cover maps for the years 1985, 2000, and 2018. 6 M. BERHANU ET AL. Figure 4. Map of NDVI and NDWI for the period of 1985, 2000, and 2018. < NDVI > 0.10 and −0.15 < NDWI > 0 are presented considered as a wetland (Islam et al., 2008). The in Figure 4. Findings of the study by Kulawardhana SRTM DEM slope of less than 1% also helped deline- et al. (2007) showed the best of these NDVI indices ate higher-order wetlands rapidly and accurately. provided the only accuracy of less than 30% with high Based on the higher accuracy of threshold values, levels of omissions and commissions. A primary cause a 5% slope threshold was used for the study area and for this is rugged topography, except few areas in produce potential maps for higher-order wetland different parts of the watershed. Some studies showed boundaries (Figure 5). that topography significantly affects VIs in a rugged The TWI was used as an additional method to map mountainous area (Veraverbeke et al., 2010; Verbyla wetlands in the study area. TWI was calculated for the et al., 2008; Wang et al., 2012). Deng et al. (2007) whole watershed (Figure 6). The result showed that observed that the NDVI and the Normalized the TWI performs better than NDVI and NDWI to Difference Infrared Index (NDII) showed detect wetlands in the study area. A larger wetland a significant correlation (r ) (p = 0.001) with topogra- area was identified with better accuracy, while smaller phy variables such as slope and the cosine of the wetlands were invisible. TWI maps with their respec- aspect. The highest amount of rainfall in the study tive wetland threshold were considered as a detected area (mono-modal rainfall pattern) exerts seasonal wetland. Thus, the percentage of pixels that were well control on vegetation greenness, leading to similar predicted was compared to the total of pixels of the spectral reflectance of wetlands with other land-cover potential wetlands mapped using supervised classifica - types. tion. Bisrat and Berhanu (2018) stated TWI was used to represent the spatial distribution of water flow and water stagnating across the study area. According to Wetlands derived from slope and topographic Wu (2018), high-resolution LiDAR-based DEMs have wetness index been used to derive TWI and facilitate forested wet- The results of delineation of wetlands using SRTM land mapping. The enhanced FD8 TWI provided DEM. Different threshold values were used to identify a good prediction of wetland location but could not wetlands within various studies based on topography, predict the periodicity of inundation. data type, and landscape nature. The overwhelming proportions of the wetlands in any landscape are along Wetland mapping with the drainage system, with drainage forming its centre. For example (Ozesmi & Bauer, 2002; Zhang The combination of automated and pixel-based et al., 2016), a threshold value of <5 degrees was classification methods consisted of slope and TWI GEOLOGY, ECOLOGY, AND LANDSCAPES 7 Figure 5. Reclassified slope map. separate wetlands from other upland land areas. Enago, chebere, Tulube wetland, Hamuma, and TWI was overlaid with a Wetland map to identify Wangegnye wetlands were some of the wetlands topographically wet areas. The main challenge for in the watershed. Most of the study area, wetlands detecting wetland within densely vegetated areas is are called by the name of standing kebele names of to differentiate dry upland forests from forested the wetlands. Enago and Wangegnye wetlands were wetlands. However, using TWI, upland forests commonly dominated by grasses cyperus latiolious were eliminated, and detection of the forested wet- wetland vegetation type. Vegetation types of Enago land was enhanced. But TWI did not eliminate and Wangenye wetlands are shown in Figure 8. upland areas from the forested wetland. As a result, digitization was done by enhancing wet- Mapping and quantification of wetland lands using false color Infrared color combinations. degradation TWI, slope, and classification maps were then com- bined according to the rule that pixels with a slope Wetlands within the study area have been experien- <5%, wet in classification map, and pixel that falls cing huge losses over the last 33 years. Detailed wet- under TWI threshold value were considered wet- land losses have been consecutively captured between lands, and the other pixels were coded to upland. 1985, 2000, and 2018. Spatial changes in surface area The land cover classification map was used to gen- and shape over a long period were considered wetland erate the wetland boundaries. For example, pixels degradation. Wetland disturbance through cultivation classified as forest in the classification map were re- and vegetation clearance alters wetlands functionality coded to forest wetland if they overlapped with leading to their degradation. The total wetland area in a wetland threshold with TWI and slope. Those the watershed accounted for 21,400 ha, 16,000 ha, and pixels coinciding with upland were coded as non- 12,700 ha for 1985, 2000, and 2018, respectively. The wetland. The final wetland classification map was detailed degradation map of wetland classes within the validated based on field data and Google Earth study area shows a similar pattern (Figure 9). Wetland data. Combining the information from Landsat degradation mapping from multi-temporal image TM/OLI, TWI, and slope, effectively map wetlands analysis revealed a significant loss of wetlands area in the study area (Figure 7). Most of the valley during 1985−2018 in the area. Figure 10 shows that wetland land-covers vegetation, locally known as 21,400 ha of the area was covered by wetland in cheffe (Cyperus latifolious), grasses, and forested the year 1985. Though, this coverage was reduced to wetlands were the dominant types of wetland vege- 16,000 ha by the year 2000. The total wetland cover tation types. A potential hotspot location of the degraded during 2000 and 2018 amounts to 3,300 ha. wetland was identified, and some of them are The wetland area was decreased by 59.34% between 8 M. BERHANU ET AL. Figure 6. Topographic wetness index map. the years 1985 and 2018. From the part of 21,400 ha in natural ecosystems (Leykun, 2003). Major rivers within 1985, about 8,700 ha was degraded by the year 2018. the watershed, which are Birbir, Geba, Dabena, Sor, and Table 2 shows the overall classification accuracy Keber rivers, are tributary rivers of the Baro riverine assessment and Kappa statistics of the results for the basin. Most of these rivers are punctuated by numerous study years 1985, 2000, and 2018. An accuracy assess- valley bottom wetlands that occur in their upper and ment for all of the used methods has been made by lower courses. The primary sources of all rivers in the comparing the results with high-resolution images watershed start from the wetland ecosystem (Figure 12). from Google Earth and field-collected data. The The hydrological function of wetlands determines the ground sample points (Figure 11) were overlaid on existence of rivers within the watershed and the sur- wetland maps to determine the classification accura- rounding area. As the wetland within a watershed cies and errors of each class. The overall accuracy of degrades, it will significantly impact the hydrology of the three aggregated wetlands (1985, 2000, and 2018) the watershed. Dixon (2002) had monitored the water in the study area was 86.66%, with reasonable errors of table wetlands of Illubabor. Monitoring was conducted omissions (7.54%) and low errors of commissions on 10–12 deep wells within each wetland every week. (13.33%). This showed that all the maps meet the The hydrologic analysis of the well data showed that recommended minimum 87% accuracy and there is lowering the wetland water table was observed in the a strong agreement between the reference data and the degraded and cultivated wetlands and reducing the rate classified habitat classes. of water movement through the wetlands. Analysis of degraded or cultivated wetlands versus undrained wet- land observed temporal variability and height change in Impact of wetland degradation in hydrology the weekly wetland water table (Figures 13,14). Dixon Major river and lake systems, together with their asso- (2002) indicated environmental degradation on wet- ciated wetlands, are fundamental parts of life interwo- lands unable to provide their full range of function, ven into the structure and welfare of societies and which has implications for food security in the study GEOLOGY, ECOLOGY, AND LANDSCAPES 9 Figure 7. Wetland map of Geba watershed dated 1985, 2000, and 2018. Figure 8. Vegetation types of Enago and Hamuma wetland. 10 M. BERHANU ET AL. Figure 9. Wetland degradation map of year 1985, 2000, and 2018. Figure 10. Wetland degradation graph for the period of 1985, 2000, and 2018. Table 2. Accuracy assessment for the Wetland system. Error Class GTP Commission (%) Omission (%) Overall accuracy (%) Wetland 1985 100 16 11.3 84 Wetland 2000 100 12.66 6.66 87.33 Wetland 2018 100 11.33 4.66 88.66 Aggregated 300 13.33 7.54 86.66 area and water availability to local communities, both Illubabor Zones, southwest Ethiopia has led to several around wetlands themselves and downstream. ecological and economic problems. Some of these are There is a significant difference between wetland immediate and linked to drainage, such as the scarcity water levels of cultivated/degraded wetlands and of thatching reeds, vegetation change, lowered water undrained wetlands. Hence, over-cultivation and tables, reduced accessibility, and provides unsafe water draining of wetlands within the watershed directly (Wood & Dixon, 2000). impact the water level of the wetlands. Additionally, alterations of the hydrological regime of wetlands have Driving forces and implications significant physical, chemical, and biological effects that can have significant ecological and socio- As the classification of Landsat 8 of 2018, which has economic implications at a broader scale (Bezabih & 30 m spatial resolution images, showed that in the Mosissa, 2017). The complete drainage of wetlands in study period of 2018, Forest and agricultural land GEOLOGY, ECOLOGY, AND LANDSCAPES 11 Figure 11. Verification of delineated wetland point projected from Google Earth Pro (1985, 2000, and 2018). Figure 12. Wetland interaction with drainage network. were the dominant land-use and land-cover types. According to the wetland change matrix obtained Forest and agricultural LU/LC types together from the land-use and land-cover map of 2018, the accounted for 616,700 ha (87.07%) of the study area’s majority of wetland area 3,111.5 ha (14.53%) con- total area in 2018 (Table 3). Land-use and land-cover verted typically to agricultural land from 1985 to maps quantified land-cover area is shown in Figure 15. 2000. Similarly, in the same fashion, most wetlands Wetland land-cover was cover only 12,700 ha in 2018. 17.515 ha (10.9%) converted typically to agricultural 12 M. BERHANU ET AL. Figure 13. Mean weekly water table elevation in the undrained wetlands (August 1997− July 1998). Figure 14. Mean weekly water table elevation in the drained and degraded wetlands (August 1997−July 1998). maize, teff, and sugarcane were more common among Table 3. Land-use and land-cover area for the year 2018. rural ranchers. Land-use and land-cover Area (ha) Area (%) The cultivation of wetlands is still going on in the Urban 2413.67 0.35 study area. The highlands of southwest Ethiopia Tea plant 2088.46 0.29 (Illubabur) and swamps of Awash valley are good Wetland 12,700 2.39 Forest 364,734 51.49 examples of where the farmers are engaged in produ- Shrub land 700.83 9.89 cing more than seeing sustainable use of the resources Agricultural land 70,083.6 35.59 (Dixon, 2002; Dixon & Wood, 2007). Field photo cap- ture of wetland cultivation and draining (Figure 16). Food insecurity due to pests and crop storage pro- blems, land shortages for cultivation and grazing due land (Table 4). Accuracy assessment of land-cover to coffee planting on uplands giving more people access classification of 2018 is shown in Table 5. Wetland to wetlands, and encouraging use listed as the main cultivation and degradation practiced in the watershed drivers of wetland degradation in the Illubabor zone. dominated by the cultivation of different vegetables, A previous study report of the wetlands policy briefing GEOLOGY, ECOLOGY, AND LANDSCAPES 13 Figure 15. Land-use/land-cover map of the year 2018. workshop done by Wood and Dixon (2000) described provides recreational value but also substantial eco- that wetland cultivation had been practiced in the nomic value. Because of the magnitude and ubiquity Illubabor zone for about eight decades and about 70% of Geba watershed wetland cover change effects, we are of the study farmers area have cultivated wetlands at compelled to reflect on past policy, management, and least once a year. Also, a significant correlation between other decision-making processes and improve them for the rate of rainfall and the change in wetland area were the future. The policy will need to consider actions that analyzed, a linear regression model was set, change in help the wetland ecosystem accommodate changes rainfall and wetland degradation positively correlated adaptively to improve the wetland ecosystem’s capacity with r = 0.81 for (1985–2018). The rate of change in to return to desired states after disturbance and mea- rainfall and the change rate of the wetland was posi- sures that reduce anthropogenic influences on wetland tively correlated (Figure 17). components. Wetlands as protected areas Hydro-Chemistry of Wetland water quality assessment Some wetlands were found inside protected areas in the study area. Some organizations within the watershed Wetland water contains many chemical species in saved and improved wetland management by including the dissolved state. Hamuma degraded and these areas within protected areas. For example, Bishari cultivated wetlands have lots of organic matter. Park in Mettu town results from the Rehabilitation of Based on the hydro-chemistry analysis results, wetlands by Mettu town Beshari prison center the effluent from agricultural land goes through (Figure 18). Hence, the park provides endless popular a wetland. A moderate difference in qualities of recreational activities, such as boating, birdwatching, water compared to undrained/or uncultivated wet- and many fascinating life forms, making the wetlands lands was observed (Tables 6,7 and 8). The surface especially enjoyable. Bishari Park not only contains or wetland average values of pH for the Enago, 14 M. BERHANU ET AL. Table 4. Wetland change matrix vs other land-use and land-cover areas. Change in hectare Wetland Urban Tea plantation Shrubland Agricultural land Forest Area (ha) (%) (ha) (%) (ha) (%) (ha) (%) (ha) (%) (ha) (%) Wetland conversion 16,000 74.76 42 0.19 191 0.89 258 1.20 3111 14.53 1797 8.39 1985−2000 Wetland conversion 2000−2018 12,700 79.04 152 0.94 280 1.74 149 0.92 1751 10.90 1033 6.42 Hamuma, and bake Chora was ranged from 7.45 grazing activities by the local community leads to the to 7.55, 7.15 to 8.35, and 7.15 to 8.01, respectively. hydrologic and water quality functions of wetlands in As a result of cultivation on wetlands, the rise of question. There is a significant difference in the water PH values to 7.98 of degraded Hamuma wetland quality of degraded and un-degraded wetlands. Many was received at a maximum limit of WHO water of the key issues for wetlands are related to water use. quality standard. However, partially undrained Demand for water in the study area will likely increase Enago wetland results in an average pH value of due to the growing human population (CSA, 2007), 7.3. An excess of agricultural fertilizers/ pesticides agricultural expansion, and climatic changes. Climate near a wetland and cultivated wetland increases change will also reduce water availability. Geba the value of leached chemical pollutants to watershed wetlands or watersheds cross national or Hamuma Degraded wetland represented by the regional borders, presenting both challenges and nitrogenous pollutant chemicals. The average opportunities for management. Make the science- values of Enago and Bake Chora partially policy arena more interactive, with scientists and poli- undrained wetland was ranged from 0.22 and ticians working closely together to mutual benefit 0.12 mg/N, respectively. However, the average (Toomey et al., 2017). Our horizon scan highlighted values of drained Hamuma wetlands were over perceived deficiencies in the governance of wetlands, at ranged values of nitrate nitrogen and nitrite nitro- both national scales, that are likely to continue into the gen was verified. future. Unified wetland ecosystem evaluation indica- Additionally, in terms of color and turbidity, tors are helpful for providing a guide for integrating a significant difference was recorded between wetland ecosystem evaluation data in different geogra- degraded/or cultivated wetlands and undrained wet- phical regions to improve the reliability of horizontal lands within the watershed. The average watercolor studies (Janse et al., 2019). Studies of similar geogra- values for the undrained Hamuma and Bake Chora phical regions in which the same wetland ecosystem were 19 and 38.33 mg/l pt was verified, which is accep- are examined could be evaluated to develop standar- table water quality based on WHO standards. But, dized indexes for wetland ecosystem across different a very significant average value of color was observed geographical regions. A long-term monitoring pro- in cultivated degraded wetland of 228.6 mg/l pt. The gram on wetland with field observations and remote turbidity values for uncultivated Enago were recorded, sensing will assist in gathering big data for evaluation, and 4NTU and Bake Chora Surface wetland water analysis, and management of wetland ecosystem. samples were obtained for four values that are accep- Exploring trade-offs among ecosystem service and link- table based on WHO water quality guideline standard. ing them with stakeholders can help to determine the But, again, a massive difference in turbidity between potential losers and winners of wetland management drained and undrained wetlands was observed. Based (Guida et al., 2016). This analysis will support research, on the result, average turbidity values of 18NTU were policy and practice related to environmental conserva- verified in cultivated and drained Hamuma wetlands. tion and sustainable development in the Geba Generally, wetland degradation has significant impact watershed, and provides a model for similar analyses on the surface wetland water qualities. Wetland dis- elsewhere in the world. traction resulting from draining, cultivating, and Conclusion Table 5. Accuracy assessment of land-use and land-cover The geospatial techniques provide an accurate, fast, and classification. economical way for wetland degradation in the Geba User Producer Overall watershed, southwest of Ethiopia. It has solved the Land-use type accuracy accuracy accuracy problems faced using old traditional techniques that Urban 93.75 76 84.84 Wetland 88.77 86.45 were difficult to undertake and consume a lot of time. Forest 88.54 85 Global climate change and anthropogenic impact Shrub land 85.56 84.69 Agricultural 81.91 84.61 degrade wetlands, which creates a severe problem in land identifying and quantifying wetland areas. The main Tea plantation 86.31 88.17 objective was to map the degradation of wetlands GEOLOGY, ECOLOGY, AND LANDSCAPES 15 Figure 16. Wetland cultivation and draining in the watershed. Figure 17. Spatial distribution of tendency variation of precipitation from January to December. Figure 18. Mettu Beshari Recreation Wetland Park. 16 M. BERHANU ET AL. Table 6. Water quality for Enago wetland. which have been under several pressure from both S/ anthropogenic and natural driving factors for three N Parameters Sample 1 Sample 2 Sample 3 Average consequent periods of years. Wetland change produces 1 PH 7.55 7.45 7.15 7.38 a significant impact on the surrounding environment, 2 Color 11 mg/L pt over range 14 mg/L pt 12.5 3 Turbidity 7 FTU 5 FTU 0 FTU 4 mainly influenced hydrological variation. The changes 4 Chlorine 0.23 mg/L 0.15 mg/L 2.561 mg/L 0.98 in the groundwater table of the study area are attrib- Cl Cl Cl 2 2 2 5 Chlorine total 0.28 mg/L 0.90 mg/L 0.060 mg/L 0.41 uted to wetland area change. Based on the groundwater Cl Cl Cl 2 2 2 table’s weekly measurement inside the watershed, 6 Fluoride Under 1.06 mg/ 0.75 mg/L F 0.90 a significant difference in groundwater table level was range L F 7 Manganese 0.003 mg/L 0.016 mg/L 0.0081 mg/L 0.009 recorded on degraded wetlands. The degraded wetland 8 Nitrate 0.028 mg/ 0.038 mg/ 0.605 0.22 maps that have been generated in the present study nitrogen L N L N 9 Nitrite 0.019 mg/l 0.205 mg/ 0.1375 mg/ 0.12 advance our understanding of current use, transforma- nitrogen N LN LN tion dynamics in wetlands and may provide the quan- 11 Potassium 4.9 mg/L K 12.0 mg/L 15.45 mg/L 10.78 k k titative basis needed to guide and predict future 12 Transmittance 9.6 17.3 13.76 13.55 wetland uses and their impacts on surrounding natural 13 Concentrations 1.1 0.8 2.35 1.41 14 Absorbance 1.016 0.76 0.472 0.74 resources. A balanced regulation or law needs to be put forward for agricultural development and wetland eco- system sustainability. Therefore, the study suggests urgent attention of decision-makers on conservation of the wetland’s remaining resources, taking necessary Table 7. Water quality for Hamuma wetland. measures to reduce environmental risk and new tech- S/ niques and different data fusion for exploring the N Parameters Sample 1 Sample 2 Sample 3 Average potential of geospatial data in wetland monitoring sup- 1 Ph 7.45 8.35 7.15 7.98 2 Color 210 mg/l 280 mg/l 196 mg/l pt 228.6 ported with field measurements. pt pt 3 Turbidity 10 ftu 30 ftu 15 ftu 18.33 4 Chlorine 0.77 mg/l 4.00 mg/l 0.077 mg/l 1.61 cl cl cl 2 2 2 Acknowledgments 5 Chlorine total 0.75 mg/l 3.90 mg/l 2.99 mg/l 2.54 cl cl cl 2 2 2 We are thankful to the head and staff of the School of Earth 6 Fluoride 0.60 Under Under 0.6 Sciences, College of Natural and Computational Sciences, range range 7 Manganese Over range Over range Over range Over Addis Ababa University, for providing all kinds of necessary range facilities and support during the present study. The authors 8 Nitrate 0.014 mg/l Over range Over range 0.01 would like to acknowledge the earth explorer (USGS) to access nitrogen n Landsat series images free of charge and Ethiopia – National 9 Nitrite 0.081 mg/l Over range Over range 0.08 nitrogen n Meteorology Agency, Metrological Data, respectively. We are 11 Potassium 2.8 mg/lk 0.97 mg/lk 5.0 mg/lk 2.92 also indebted to the editor Geology, Ecology and Landscapes, 12 Transmittance 19.6 9.8 14.52 14.64 and anonymous reviewers for their constructive review that 13 Concentrations 2.058 1.45 2.89 2.13 helped to improve the structure and quality of the paper. 14 Absorbance 0.27 0.22 0.97 0.59 Disclosure statement No potential conflict of interest was reported by the author(s). Table 8. Water quality for Bake Chora wetland. S/ N Parameters Sample 1 Sample 2 Sample 3 Average 1 PH 7.15 7.25 7.15 7.18 Funding 2 Color 50 mg/L pt 50 mg/L pt 15 mg/L pt 38.33 3 Turbidity 3 FTU 7 FTU 4 FTU 4.6 The authors received no direct funding for this research. 4 Chlorine 0.35 mg/L 0.68 mg/L 0.35 mg/L 0.46 Cl Cl Cl 2 2 2 5 Chlorine total 0.31 mg/L 0.75 mg/L 0.31 mg/L 0.45 Cl Cl Cl 2 2 2 ORCID 6 Fluoride 0.070 Under 0.01 0.04 Range Mintesnot Berhanu http://orcid.org/0000-0002-9547- 7 Manganese 0.05 mg/L 0.01 mg/L 0.43 mg/L 0.16 8 Nitrate 0.081 mg/ Over range 0.081 mg/l 0.08 Karuturi Venkata Suryabhagavan http://orcid.org/0000- nitrogen L N N 9 Nitrite nitrogen 0.014 mg/ Over range 0.014 mg/ 0.01 0003-2528-9106 L N L N 11 Potassium 2.8 mg/L K 0.97 mg/ 5.07 mg/ 2.94 L K L K References 12 Transmittance 8.7 7.7 18.85 11.75 13 Concentrations 3.5 0.25 0.35 1.36 Abebe, Y. D., & Geheb, K. (2003). Wetlands of Ethiopia. 14 Absorbance 1.77 0.05 1.675 1.16 Proceedings of a seminar on the resources and status of GEOLOGY, ECOLOGY, AND LANDSCAPES 17 Ethiopia’s wetlands 116. IUCN-Eastern Africa Regional Dixon, A.B., & Wood, A.P. (2007). Local institutions for office Nairobi, Kenya. wetland management in Ethiopia: sustainability and state Amsalu, T., & Addisu, S. (2014). A review of Wetland intervention. In: van Koppen B, Giordano M, Conservation and Management Policy in Ethiopian. Butterworth J (eds). Community -based water law and International Journal of Scientific and Research water resources management reform in developing coun- Publications, 4(9), 1–6. http://citeseerx.ist.psu.edu/view tries, Comprehensive Assessment of Water Management in doc/download?doi=10.1.1.654.2323&rep=rep1&type=pdf Agriculture Series 5. CABI International, Wallingford, Baker, C., Lawrence, R., Montagne, C., & Patten, D. (2006). 130–145. Mapping wetlands and riparian areas using Landsat ETM Foody, G. M. (2002). Status of land cover classification + imagery and decision-tree-based models. Wetlands, 26 accuracy assessment. Remote Sensing of Environment, 80 (2), 465–474. https://doi.org/10.1672/0277-5212(2006)26 (1), 185–201. https://doi.org/10.1016/S0034-4257(01) [465:MWARAU]2.0.CO;2 00295-4 Beven, K. J., & Kirkby, M. J. (1979). A physically based, Gebresllassie, H., Gashaw, T., & Mehari, A. (2014). Wetland variable contributing area model of basin hydrology/Un Degradation in Ethiopia: Causes, Consequences, and modèle à base physique de zone d’appel variable de l’hy- Remedies. Journal of Environment and Earth Science, 4 drologie du bassin versant. Hydrological Sciences Bulletin, (11), 40–48. 24 (1), 43–69. https://doi.org/10.1080/ Gerjevic, J. H. (2004). Using Remote Sensing Data to Study 02626667909491834 Wetland. In Iowa: University of Northern Iowa. Bezabih, B., & Mosissa, T. (2017). Review on distribution, Guida, R. J., Remo, J. W. F., & Secchi, S. (2016). Tradeoffs of importance, threats and consequences of wetland degra- strategically reconnecting rivers to their floodplains: The dation in Ethiopia. International Journal of Water case of the Lower Illinois River (USA). Sci. Total Environ, Resources Environmental Engineering, 9(3), 64–71. 572,43−55. doi:10.1016/j.scitotenv.2016.07.190 https://doi.org/10.5897/IJWREE2016.0697 Hailu, A. (1998). An overview of wetland use in Illubabur Bisrat, E., & Berhanu, B. (2018). Identification of Surface Zone, South-western Ethiopia. In A Woreda survey. Water Storing Sites Using Topographic Wetness Index Ethiopian Wetlands Research Project. (TWI) and Normalized Difference Vegetation Index Hailu, A. (2005). Ethiopian Wetlands Distribution, Benefits (NDVI). Journal of Natural Resources and Development, and Threats, Ethio-Wetlands and Natural Resources 08, 91–100. https://doi.org/10.5027/jnrd.v8i0.09 Association. In In: The Proceedings of the second Blaschke, T. (2010). Object based image analysis for remote Awareness creation Workshop on Wetlands in the sensing. ISPRS Journal of Photogrammetry and Remote Amhara Region (pp. 3–17). Addis Ababa. Sensing, 65(1), 2–16. https://doi.org/10.1016/j.isprsjprs. Islam, M. A., Thenkabail, P. S., Kulawardhana, R. W., 2009.06.004 Alankara, R., Gunasinghe, S., Edussriya, C., & Chouari, W. (2021). Wetland land cover change detection using Gunawardana, A. (2008). Semi-automated methods for multitemporal Landsat data: A case study of the Al-Asfar mapping wetlands using Landsat ETM+ and SRTM data. wetland, Kingdom of Saudi Arabia. Arab J Geosci, 14(6), 523. International Journal of Remote Sensing, 29(24), https://doi.org/10.1007/s12517-021-06815-y 7077–7106. https://doi.org/10.1080/01431160802235878 Das, K. (2017). NDVI and NDWI based change detection James, B. C., & Randolph, H. W. (2011). Introduction to analysis of Borfoibam Beelmukh Wetlandscape, Assam Remote Sensing (Fifth Edition ed.). The Guilford Press, using IRS LISS III data. ADBU-Journal of Engineering New York. Technology, 6(2), 17–21. Janse, J. H., Dam, A. A., Hes, E. M. A., Klein, J. J. M., Davidson, N. C., Van Dam, A. A., Finlayson, C. M., & Finlayson, C. M., Janssen, A. B. G., Wijk, D., McInnes, R. J. (2019). Worth of wetlands: Revised global Mooij, W. M., & Verhoeven, J. T. A. (2019). Towards monetary values of coastal and inland wetland ecosystem a global model for wetlands ecosystem services Curr. services. Mar Freshw Res, 70(8), 1189–1194. https://doi. Opin. Environ. Sustain, 36, 11–19. https://doi.org/10. org/10.1071/MF18391 1016/j.cosust.2018.09.002 Dehm, D., Becker, R., & Godre, A. (2019). SUAS based Jensen, J. R. (2007). Remote Sensing of the Environment: An multispectral imagery for monitoring wetland inundation Earth Resource Perspective. In Pearson Prentice Hall, and vegetation. Environmental Sciences. https://doi.org/ Upper Saddle River (2nd Edition ed.). New Jersey: 10.20944/preprints201911.0326.v1 Prentice Hall. Deng, Y., Chen, X., Chuvieco, E., Warner, T., & Wilson, J. P. Jones, J. W. (2015). Efficient Wetland Surface Water (2007). Multi-scale linkages between topographic attri- Detection and Monitoring via Landsat: Comparison butes and vegetation indices in a mountainous landscape. with in situ Data from the Everglades Depth Estimation Remote Sens. Environ, 111(1), 122–134. https://doi.org/ Network. Remote Sensing, 7(9), 12503–12538. https://doi. 10.1016/j.rse.2007.03.016 org/10.3390/rs70912503 Dixon, A. (2002). The hydrological impacts and sustainabil- Kaplan, G., & Avdan, U. (2017). Mapping and monitoring ity of wetlands drainage cultivation in Illubabur, Ethiopia. wetlands using sentinel-2 satellite imagery. ISPRS Annals Land Degradation and Development, 13(1), 17–31. of the Photogrammetry. Remote Sensing and Spatial https://doi.org/10.1002/ldr.479 Information Sciences, 4(4), 271–277. https://doi.org/10. Dixon, A., Wood, A., & Hailu, A. (2021). Wetlands in 5194/isprs-annals-IV-4-W4-271-2017 Ethiopia: Lessons From 20 years of Research, Policy and Kaplan, G., & Avdan, U. (2019). Evaluating the utilization of Practice. Wetlands, 41(2), 20. https://doi.org/10.1007/ the red edge and radar bands from Sentinel sensors for s13157-021-01420-x wetland classification. Catena, 178,109–119. doi:10.1016/ Dixon, A. B., . &., & Wood, A. P. (2003). Wetland cultiva- j.catena.2019.03.011 tion and hydrological man-agreement in East Africa: Kulawardhana, R. W., Thenkabail, P. S., Vithanage, J., Matching community and hydrological needs through Biradar, C., Md.A, I., Gunasinghe, S., & Alankara, R. sustainable wetland use. Natural Resources Forum, 27 (2007). Evaluation of the Wetland Mapping Methods (2), 117–129. https://doi.org/10.1111/1477-8947.00047 using Landsat ETM+ and SRTM Data. Journal of 18 M. BERHANU ET AL. Spatial Hydrology, 7(2), 62–96. https://hdl.handle.net/ Terefe, D. (2017). Ethiopia: Why Conservation of Wetlands 10568/21603 Makes Sense. Water Journalists Africa. (https://waterjour Lane, C. R., Liu, H., Autrey, B. C., Anenkhonov, O. A., nalistsafrica.com/2017/02/ethiopia-why-conservation- Chepinoga, V. V., & Wu, Q. (2014). Improved ofwetlands-makes-sense/#respond ) Wetland Classification Using Eight-Band High Toomey, A. H., Knight, A. T., & Barlow, J. (2017). Resolution Satellite Imagery and a Hybrid Approach. Navigating the space between research and implementa- Remote Sensing, 6(12), 12187–12216. https://doi.org/ tion in conservation: Research-implementation spaces. 10.3390/rs61212187 Conserv Lett, 10(5), 619–625. https://doi.org/10.1111/ Leykun, A. (2003). The distribution and status of Ethiopian conl.12315 wetlands: An overview. In Y. D. Abebe & K. Geheb (Eds.), Veraverbeke, S., Verstraeten, W. W., Lhermitte, S., & In Wetlands of Ethiopia, proceedings of a seminar on the Goossens, R. (2010). Illumination effects on the differ - resources and status of Ethiopia’s wetlands. IUCN. enced Normalized Burn Ratio’s optimality for assessing Leykun, A. (2003). The distribution and status of Ethiopian fire severity. International Journal of Applied Earth wetlands: An overview. Proceedings of a seminar on the Observation and Geoinformation, 12(1), 60–70. https:// resources and status of Ethiopia’s wetlands (pp. 12. vi + doi.org/10.1016/j.jag.2009.10.004 116). IUCN Eastern Africa Regional Office. Kenya. Verbyla, D. L., Kasischke, E. S., & Hoy, E. E. (2008). Seasonal Li, N., Li, L., Lu, D., Zhang, Y., & Wu, M. (2019). Detection and topographic effects on estimating fire severity from of coastal wetland change in China: A case study in Landsat TM/ETM+ data. International Journal of hangzhou bay. Wetlands Ecology and Management, 27 Wildland Fire, 17(4), 527–534. https://doi.org/10.1071/ (1), 103–124. https://doi.org/10.1007/s11273-018-9646-3 WF08038 Mengesha, T. A. (2017). Review on the natural conditions Wang, G., & Weng, Q. (2014). Remote Sensing of Natural and anthropogenic threats of Wetlands in Ethiopia. Resources. In Taylor & Francis Series in Remote Sensing Global Journal of Ecology, 2(1), 006–014. https://doi.org/ Applications. Indiana State University Terre Haute, 461. 10.17352/gje.000004 Gondar Indiana, U.S.A.: CRC Press Taylor & Francis Group. Mitsch, W. J., & Gosselink, J. G. (2007). Wetlands (4th https://doi.org/10.1201/b15159 Edition ed.). John Wiley & Sons, Inc.. Wang, Y., Hou, X., Wang, M., Wang, M., Wu, L., Ying, L., & Nsubuga, F. W. N., Botai, J. O., . &., Olwoch, J. M., Feng, Y. (2012). Topographic controls on vegetation Rautenbach, C. J. D., Kalumba, A. M., Tsela, P., index in a hilly landscape: A case study in the Jiaodong Adeola, A. M., Sentongo, A. A., & Mearns, K. F.. (2017). Peninsula, eastern China. Environmental Earth Sciences, Detecting changes in surface water area of Lake Kyoga 70(2), 625–634. https://doi.org/10.1007/s12665-012- sub-basin using remotely sensed imagery in a changing 2146-5 climate. Theoretical and Applied Climatology, 127(1–2), Woldu, Z., & Yeshitela, K. (2003). Wetland plants in 327–337. https://doi.org/10.1007/s00704-015-1637-1 Ethiopia with examples from Illubabor, south-western Ozesmi, S. L., & Bauer, M. E. (2002). Satellite remote sensing Ethiopia in: Yilma D. Abebe and Kim Geheb (2003) of wetlands. Wetlands Ecology and Management, 10, 381– Wetlands of Ethiopia: Proceedings of a seminar on the 402. https://doi.org/10.1023/A:1020908432489 resources and status of Ethiopia’s wetlands. IUCN. Ozesmi, S. L., & Bauer, M. E. (2002). Satellite remote sensing Wood, A. P., & Dixon, A. B. (2000). Sustainable wetland of wetlands. Wetlands Ecology and Management, 10(5), management in Illubabor Zone: Research report summa- 381–402. https://doi.org/10.1023/A:1020908432489 ries. Unpublished Report, The University of Huddersfield . Sakane, N., Alvarez, M., Becker, M., Böhme, B., Handa, C., Available via: https://www.wetlandaction.org/wp-content Kamiri, H. W., Langensiepen, M., Menz, G., Misana, S., /uploads/EWRP-Objective-Summaries.pdf . Mogha, N., Möseler, B., Mwita, E., Oyieke, A. H., & van Wu, Q.2018 GIS and Remote Sensing Applications in Wijk, M. (2011). Classification, characterization, and use Wetland Mapping and Monitoring.B. Huang,Ed.. of small wetlands in East Africa. Wetlands, 31(6), Comprehensive Geographic Information 1103–1116. https://doi.org/10.1007/s13157-011-0221-4 Systems.2 140–157.Oxford: Elsevier.https://doi.org/10. Suding, K. N. (2011). Toward an Era of Restoration in Ecology: 1016/B978-0-12-409548-9.10460-9 Successes, Failures, and Opportunities Ahead. Annual Review Xu, H. (2006). Modification of normalised difference water of Ecology, Evolution, and Systematics, 42(1), 465–487. https:// index (NDWI) to enhance open water features in remo- doi.org/10.1146/annurev-ecolsys-102710-145115 tely sensed imagery. International Journal of Remote Suryabhagavan, K. V. (2017). GIS-based climate variability Sensing, 27(14), 3025–3033. https://doi.org/10.1080/ and drought characterization in Ethiopia over three 01431160600589179 decades. Weather and Climate Extremes, 15, 11–23. Zhang, Z., Zimmermann, N., Kaplan, J., & Poulter, B. https://doi.org/10.1016/j.wace.2016.11.005 (2016). Modeling spatiotemporal dynamics of global wet- Tarboton, D. G. (1997). A New Method for the Determination lands: Comprehensive evaluation of a new sub-grid of Flow Directions and Contributing Areas in Grid Digital TOPMODEL parameterization and uncertainties. Elevation Models. Water Resources Research, 33(2), 309–319. Biogeosciences, 13(5), 1387–1408. https://doi.org/10. https://doi.org/10.1029/96WR03137 5194/bg-13-1387-2016 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geology Ecology and Landscapes Taylor & Francis

Wetland mapping and evaluating the impacts on hydrology, using geospatial techniques: a case of Geba Watershed, Southwest Ethiopia

Download PDF
18 pages

Loading next page...
 
/lp/taylor-francis/wetland-mapping-and-evaluating-the-impacts-on-hydrology-using-C0BVJT3o1a

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Taylor & Francis
Copyright
© 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON).
ISSN
2474-9508
DOI
10.1080/24749508.2021.1953744
Publisher site
See Article on Publisher Site

Abstract

GEOLOGY, ECOLOGY, AND LANDSCAPES INWASCON https://doi.org/10.1080/24749508.2021.1953744 RESEARCH ARTICLE Wetland mapping and evaluating the impacts on hydrology, using geospatial techniques: a case of Geba Watershed, Southwest Ethiopia Mintesnot Berhanu , Karuturi Venkata Suryabhagavan and Tesfaye Korme School of Earth Sciences, Addis Ababa University, Addis Ababa, Ethiopia ABSTRACT ARTICLE HISTORY Received 15 April 2021 Wetlands are one of the world’s most important ecosystems threatened by man. This investi- Accepted 4 July 2021 gation explores the use of Landsat TM and OLI imageries with SRTM DEM for mapping them. Mapping and monitoring of wetlands is done with 86.66% accuracy. As a result, a loss of KEYWORDS −1 21,400 ha yr could be noted. Due to this, differences were also found in the water quality and Agriculture; Climate; groundwater level between the degraded and un-degraded areas. As most rivers within the Wetland loss; Water quality; watershed punctuated from the wetlands, their existence was determined based on the Water level hydrological function. Wetland degradation occurred mainly due to climate and agricultural changes over time. Thus, geospatial techniques employed in the present study have proved very useful in simplification and visualization of wetland monitoring. Introduction cover 12 million ha in Tanzania and Kenya (Sakane Wetlands are among the most significant, multi- et al., 2011). The majority of wetlands lost historically functional, and productive ecosystems on the were drained or filled to create agricultural land. earth (Davidson et al., 2019; Mengesha, 2017). A comprehensive inventory of wetlands in Ethiopia Primarily, wetlands can provide essential environ- is not yet done, but wetlands are estimated to cover mental services, including storing floodwater, redu- about 2% of the country’s surface area (Amsalu & cing peak runoff, recharging groundwater, filtering Addisu, 2014; Mengesha, 2017; Terefe, 2017). impurities in water, carbon storage, and also eco- Wetlands are a common feature of the landscapes in logically serve as breeding grounds and critical the highlands of southwestern Ethiopia, particularly habitat for several species of plant communities, Western Wellega and Illubabor (Abebe & Geheb, invertebrates, fish, and wildlife (Chouari, 2021; 2003; Dixon et al., 2021). Despite recognizing their Kaplan & Avdan, 2017; Wang & Weng, 2014; Wu, many uses by people, their ecological services to 2018). Since the early 20th century, the global wet- humankind, and their environmental significance, land ecosystem has confronted significant chal- Ethiopian wetlands are under severe pressure and lenges, including rapid economic development and degradation. The loss of these wetlands is devastating habitat destruction. Wetlands cover approximately to several wetland-dependent endemic species 8% of the world’s land surface and contain 20% of (Bezabih & Mosissa, 2017). This is due to improper the global terrestrial carbon (Dixon et al., 2021; extraction and misconceptions forwarded to wetlands, Mitsch & Gosselink, 2007). However, despite the the health of the wetlands is continuously decreasing environmental degradation faced by wetlands, there from time to time which puts in doubt their existence is an increasing demand for ecosystem services they soon (Abebe & Geheb, 2003; Woldu & Yeshitela, provide (Suding, 2011). By 2050, global water 2003). demand is projected to increase by 55% (Terefe, In Ethiopia, wetland destruction and alteration 2017). To meet this growing demand, the services saw as an advanced development mode, even at the of wetlands must be valued appropriately, or water government level (Dixon & Wood, 2003). This security risks will rapidly increase. The need for indicates that wetlands and their value remain little information supporting wetland management is understood (Gebresllassie et al., 2014). Convention multi-scalar worldwide, and the challenge demands on Biological diversity of Ethiopia 4th report urgent and consistent wetland monitoring mechan- describe; Fogera marsh has been changed to the ism assessments to guide policymaking. rice field, Sululta marsh is distributed to investors, The wetlands are important in East Africa as they ELFORA PLC has transformed the Chefa wetland make up more than 80% of the total wetland area and in South Wello to farmland, and these are only CONTACT Karuturi Venkata Suryabhagavan drsuryabhagavan@gmail.com School of Earth Sciences, Addis Ababa University, 1176, Addis Ababa, Ethiopia © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON). This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 M. BERHANU ET AL. a few examples of wetland degradation in Ethiopia. Material and methods Lake Tana is loaded with silt and invasive water Study area hyacinth because the wetland vegetation in the surrounding catchments was destroyed and used The study area is located within the Geba River basin for agriculture. The wetlands used to stop silt and that constitutes the Baro River basin in Ethiopia at plant nutrition that is discharged to the lake have around 600 km from Addis Ababa. The area is been converted to a rice paddy. The recent total bounded by latitude 7°45ʹ00”−8°36ʹ00”N and longi- drying up of Lake Alemaya and the precarious tude 35°20ʹ00”−36°11ʹ00”E covering a total area of existence of Lake Abijata are clear evidence of the 7,125.35 km (Figure 1). The watershed geology com- looming danger on the wetland ecosystem (Amsalu prises Precambrian, metamorphosed volcanics, intru- & Addisu, 2014). sives and sediments in the north and southwest, In recent years, several studies have focused on Paleozoic and Mesozoic sediments in the center, and wetland mapping in Ethiopia’s southwest regions. localized quaternary formation along the valleys of the The complete drainage and cultivation of wetlands major rivers. The southwestern highland valley bot- have become common phenomena (Hailu, 1998; tom wetlands are developing through time to change Dixon & Wood, 2007). For instance, approximately external geological, geomorphological, and climatic one-third of the total valley bottom wetlands have conditions. The formation of ancient impermeable come under cultivation for growing food crops from quaternary and tertiary bedrock in the study area 1974 to 1983 (Hailu, 1998). More severely from this, could play a significant role in forming wetlands. The approximately 20% of the total wetlands in Illubabor elevation of the region ranges from 780 to 2661 m have been cultivated during 1986−1998, and this above mean sea level at the mountains areas. intensity increased to 35% in 1999 (Hailu, 2005). Climatic conditions in the study area are quite diverse The loss and degradation of these critical resources due to considerable differences in altitude and relief. (wetlands) need urgent mapping and monitoring of About 80% of the annual rainfall occurs in the Kirmet the resources and determining potential restoration (rainy) season from June to September, and 63% of the areas (Gerjevic, 2004). To better manage and con- annual rainfall is the peak recorded in July and serve wetland resources, we need to know the dis- August. The temperature varies from a minimum tribution and extent of wetlands and monitor their average of 6.5°C to 32°C. dynamic changes (Kaplan & Avdan, 2019; Wu, 2018). Remote sensing offers the opportunity to Data and methodology map and inventory wetlands rapidly and consis- tently, irrespective of the geographic location. The study used remote sensing data to map changing Combining remote sensing and geographic informa- wetland degradation trends in the Geba Watershed tion system approach integrated with in-situ mea- from 1985 to 2018. Landsat multi-temporal imageries surement provides an advanced tool in detecting of 1985, 2000, 2018, and Digital Elevation Model and identifying degraded wetland resources at regio- (DEM) were used for the study area (Table 1). Using nal and local scales. Wetland mapping involves most the TauDEM tool, the watershed area was generated often using satellite data and aerial photos due to automatically from the SRTM DEM, and two Landsat their remoteness and inaccessibility (Baker et al., scenes of 170/054 and 171/054 path and rows filled the 2006). Imageries from the Landsat, Aster, SPOT, study area. After radiometric and atmospheric nor- Sentinel-2, IRS, IKONOS, QuickBird, and malization, the ArcGIS mosaicking tool has been WorldView provide the necessary spatial and tem- used to generate a mosaic of two scenes covering the poral resolution to implementing effective wetland study area. The generated map was projected to World monitoring. The satellite data consisted of high reso- Geodetic System (WGS) 1984 Universal Transverse lution (1−4 m) and medium resolution (10−30 m) Mercator (UTM) zone 37 N. The images were multispectral imagery. The spatial and spectral reso- acquired through the USGS Earth explorer www. lution satellite data has significantly improved wet- usgs.gov. Different dated satellite images were of vary- land and habitat mapping (Jensen, 2007; ing pixel size, and resampling was done to obtain the Suryabhagavan, 2017). Therefore, this study aims to same pixel size in all the satellite imagery used. find how degraded wetland resources have been Meteorological data such as monthly temperature mapped. The spatial relationship between wetland and average monthly rainfall were acquired from degradation and its hydrological and related impacts Ethiopia’s National Meteorological Agency (NMA). on the surrounding environment was assessed using Based on the wetlands’ potential resources, three geospatial tools. A recent wetland degradation map kebeles of the study area were visited in February was produced for future mitigation and management and March. A field survey was conducted for ground- purposes to sustain the development at regional and truth data, land-cover, water samples were collected national levels. using handheld GPS (Global Positioning System) to GEOLOGY, ECOLOGY, AND LANDSCAPES 3 Figure 1. Location map of the study area. Table 1. Specifications of satellite data used. Path spectrally (supervised component) (Foody, 2002). In Satellite Sensor and Spatial this study, the Level 1 land-use and land-cover classi- ID type Year Acquisition date row resolution fication system was used and adapted from United Landsat 5 Thematic 1985 20 − 02 − 1985 170 30 m mapper 05 − 04 − 1988 and States Geological Survey (USGS) (James & Randolph, (TM) 171/ 2011). Most pixel-based classifications tend to utilize spectral information at individual pixels and poten- Landsat 5 Thematic 2000 27 − 02 − 2000 170 30 m mapper 17 − 01 − 2000 and tially textural information extracted from neighboring (TM) 171/ pixels. Pixel-based classification can highlight noise, Landsat 8 OLI_TIRS 2018 27 − 01 − 2018 170 30 m salt-and-pepper effects and ignore important topolo- 18 − 01 − 2018 and gical and semantic information in the images 171/ (Blaschke, 2010). Based on the maximum likelihood classification algorithm, six different classes were iden- tified; agricultural land, forest, shrubland, wetland, Gumro tea plantation, and settlement. enhance satellite image and wetland mapping classifi - cation. The flow chart of the methodology used is Vegetation indices NDVI and NDWI given in Figure 2. This study’s method holds pixel- based supervised classification, index-based classifica - The Normalized Difference Vegetation Index (NDVI) tion (NDVI and NDWI), SRTM wetness index, and and Normalized Difference Water Index (NDWI) are image enhancement methods algorithm was used to selected from this study’s satellite data image proces- identify and delineate wetland extent. sing. Several researchers have used both NDVI and NDWI for wetlands change detection (Das, 2017; Dehm et al., 2019; Li et al., 2019; Nsubuga et al., Pixel-based classification of wetland 2017; Xu, 2006); and they all found that there are Thematic mapping from satellite data can be defined significant changes of both indices values during as grouping together cases (pixels) by their relative their study periods. Several developed spectral vegeta- spectral similarity (unsupervised component) to allo- tion and water indices can highlight wetlands and cate instances based on their similarity to a set of extract water bodies while efficiently suppressing predefined classes that have been characterized (Xu, 2006). They can be calculated using the following 4 M. BERHANU ET AL. Figure 2. Methodological flow of the process involved in wetland mapping. Equations (1) and (2), respectively (Jones, 2015; Lane depends on the topographic conditions of a region. et al., 2014). The TWI was directly derived from Equation (3) (Beven & Kirkby, 1979). ðNIR RedÞ NDVI ¼ (1) � � ðNIRþ RedÞ a TWI ¼ 1n (3) tanðbÞ where NIR represents the Near Infrared Band and Red represents the Red band. where a is the local up-slope contributing area drain- ing a point per unit contour length, b is the local slope ðGreen NIRÞ NDWI ¼ (2) in radians. ðGreenþ NIRÞ Accuracy assessment Topography Wetness Index An accuracy assessment for the supervised wetland Wetlands are easily confused with other upland habi- classification was done for the 1985 image using tat cover types, such as forests because these classes ERDAS Imagine @ 15. From the classifier, 80, 75, show overlapping spectral signatures (Ozesmi & and 95 points were generated randomly for 1985, Bauer, 2002). Thematic maps of both primary and 2000, and 2018 supervised images, respectively. Each secondary key hydrological parameters attribute of point had a specific color tone and pixel value recog- slope, drainage, flow direction, flow accumulation, nized by the software itself when the data sets were stream order, Topographic Wetness Index (TWI) trained during supervised wetland classification. were derived from Shuttle Radar Topography These values were considered as reference values. All Mission (SRTM) 90 m DEM using d-infinity algo- the randomly generated points were then identified by rithm in TauDEM hydrological tools as described by the user and assigned to different classes. This process (Islam et al., 2008; Tarboton, 1997). TWI were used to was done for the three supervised classification images reduce the error of commission with upland land (i.e., 1985, 2000, and 2018). The correctly identified covers since the wetlands’ occurrence strongly points were considered as classified values. An Error GEOLOGY, ECOLOGY, AND LANDSCAPES 5 matrix and Kappa statistics were also generated from Results and discussion this reference and classified data from the report sec- Wetland classification tion of ERDAS Imagine @ 15 software. Accuracy assessment was conducted by collecting 250 in situ Initially, the NDVI and NDWI were considered for ground truth points (GTP), which were systematically water detection, as these indices have already been distributed throughout the study areas in the accessi- proven suitable for this purpose in previous studies ble parts. Overall accuracy was calculated from the (Das, 2017; Dehm et al., 2019). Supervised classifica - error matrix by dividing the sum of the entries that tions were performed for separating the valley wet- make major diagonal by the total number of examined lands from the other land-cover types. The pixel-based pixels. Kappa coefficient of the agreement was also classification was conducted on the study images’ mul- calculated by using the following Equations (4), (5), tiple segmentation levels using the maximum likeli- and (6), respectively (Kulawardhana et al., 2007). hood classification. The results were compared with corresponding pixel-based classifications to delineate Overall accuracyð%Þ ¼ � � the wetlands’ extent and boundaries in the study area 0 0 0 0 Total no: of GT points of class X that falling on class X � 100 (4) (Figure 3). However, wetland spectral mixing with Total no: of GT points of classX other land-cover regions was observed. Error of omissionð%Þ ¼ � � 0 0 0 0 Total no: of GT points of class X not falling on class X � 100 (5) Total no: of GT points of classX Vegetation indices NDVI and NDWI The finest of these indices (NDVI and NDWI) pro- Error of commissionð%Þ ¼ vided the only accuracy of less than 19%, with high Total no: of GT points of other classes falling on class X ð Þ� 100 (6) levels of errors of omissions and commissions in the 0 0 Total no: of GT points of class X present study. With several iterations, threshold values of NDVI and NDWI were done to extract wetland Kappa coefficient ¼ from other land covers. This index combination’s gen- Observed accuracy Change agreement (7) eral applicability was not proved to be very high since 1 Chance agreement the minimum water probability value was kept at 19% for all areas in the watershed. Adding the NDVI to one where observed accuracy is determined by diagonal in of these water indices reduced detection rates over error matrix, and chance agreement incorporates off- water bodies that contained some vegetation. Map of diagonal (sum of the product of row and column totals NDVI and NDWI based on threshold values of −0.25 for each class). Figure 3. Wetland cover maps for the years 1985, 2000, and 2018. 6 M. BERHANU ET AL. Figure 4. Map of NDVI and NDWI for the period of 1985, 2000, and 2018. < NDVI > 0.10 and −0.15 < NDWI > 0 are presented considered as a wetland (Islam et al., 2008). The in Figure 4. Findings of the study by Kulawardhana SRTM DEM slope of less than 1% also helped deline- et al. (2007) showed the best of these NDVI indices ate higher-order wetlands rapidly and accurately. provided the only accuracy of less than 30% with high Based on the higher accuracy of threshold values, levels of omissions and commissions. A primary cause a 5% slope threshold was used for the study area and for this is rugged topography, except few areas in produce potential maps for higher-order wetland different parts of the watershed. Some studies showed boundaries (Figure 5). that topography significantly affects VIs in a rugged The TWI was used as an additional method to map mountainous area (Veraverbeke et al., 2010; Verbyla wetlands in the study area. TWI was calculated for the et al., 2008; Wang et al., 2012). Deng et al. (2007) whole watershed (Figure 6). The result showed that observed that the NDVI and the Normalized the TWI performs better than NDVI and NDWI to Difference Infrared Index (NDII) showed detect wetlands in the study area. A larger wetland a significant correlation (r ) (p = 0.001) with topogra- area was identified with better accuracy, while smaller phy variables such as slope and the cosine of the wetlands were invisible. TWI maps with their respec- aspect. The highest amount of rainfall in the study tive wetland threshold were considered as a detected area (mono-modal rainfall pattern) exerts seasonal wetland. Thus, the percentage of pixels that were well control on vegetation greenness, leading to similar predicted was compared to the total of pixels of the spectral reflectance of wetlands with other land-cover potential wetlands mapped using supervised classifica - types. tion. Bisrat and Berhanu (2018) stated TWI was used to represent the spatial distribution of water flow and water stagnating across the study area. According to Wetlands derived from slope and topographic Wu (2018), high-resolution LiDAR-based DEMs have wetness index been used to derive TWI and facilitate forested wet- The results of delineation of wetlands using SRTM land mapping. The enhanced FD8 TWI provided DEM. Different threshold values were used to identify a good prediction of wetland location but could not wetlands within various studies based on topography, predict the periodicity of inundation. data type, and landscape nature. The overwhelming proportions of the wetlands in any landscape are along Wetland mapping with the drainage system, with drainage forming its centre. For example (Ozesmi & Bauer, 2002; Zhang The combination of automated and pixel-based et al., 2016), a threshold value of <5 degrees was classification methods consisted of slope and TWI GEOLOGY, ECOLOGY, AND LANDSCAPES 7 Figure 5. Reclassified slope map. separate wetlands from other upland land areas. Enago, chebere, Tulube wetland, Hamuma, and TWI was overlaid with a Wetland map to identify Wangegnye wetlands were some of the wetlands topographically wet areas. The main challenge for in the watershed. Most of the study area, wetlands detecting wetland within densely vegetated areas is are called by the name of standing kebele names of to differentiate dry upland forests from forested the wetlands. Enago and Wangegnye wetlands were wetlands. However, using TWI, upland forests commonly dominated by grasses cyperus latiolious were eliminated, and detection of the forested wet- wetland vegetation type. Vegetation types of Enago land was enhanced. But TWI did not eliminate and Wangenye wetlands are shown in Figure 8. upland areas from the forested wetland. As a result, digitization was done by enhancing wet- Mapping and quantification of wetland lands using false color Infrared color combinations. degradation TWI, slope, and classification maps were then com- bined according to the rule that pixels with a slope Wetlands within the study area have been experien- <5%, wet in classification map, and pixel that falls cing huge losses over the last 33 years. Detailed wet- under TWI threshold value were considered wet- land losses have been consecutively captured between lands, and the other pixels were coded to upland. 1985, 2000, and 2018. Spatial changes in surface area The land cover classification map was used to gen- and shape over a long period were considered wetland erate the wetland boundaries. For example, pixels degradation. Wetland disturbance through cultivation classified as forest in the classification map were re- and vegetation clearance alters wetlands functionality coded to forest wetland if they overlapped with leading to their degradation. The total wetland area in a wetland threshold with TWI and slope. Those the watershed accounted for 21,400 ha, 16,000 ha, and pixels coinciding with upland were coded as non- 12,700 ha for 1985, 2000, and 2018, respectively. The wetland. The final wetland classification map was detailed degradation map of wetland classes within the validated based on field data and Google Earth study area shows a similar pattern (Figure 9). Wetland data. Combining the information from Landsat degradation mapping from multi-temporal image TM/OLI, TWI, and slope, effectively map wetlands analysis revealed a significant loss of wetlands area in the study area (Figure 7). Most of the valley during 1985−2018 in the area. Figure 10 shows that wetland land-covers vegetation, locally known as 21,400 ha of the area was covered by wetland in cheffe (Cyperus latifolious), grasses, and forested the year 1985. Though, this coverage was reduced to wetlands were the dominant types of wetland vege- 16,000 ha by the year 2000. The total wetland cover tation types. A potential hotspot location of the degraded during 2000 and 2018 amounts to 3,300 ha. wetland was identified, and some of them are The wetland area was decreased by 59.34% between 8 M. BERHANU ET AL. Figure 6. Topographic wetness index map. the years 1985 and 2018. From the part of 21,400 ha in natural ecosystems (Leykun, 2003). Major rivers within 1985, about 8,700 ha was degraded by the year 2018. the watershed, which are Birbir, Geba, Dabena, Sor, and Table 2 shows the overall classification accuracy Keber rivers, are tributary rivers of the Baro riverine assessment and Kappa statistics of the results for the basin. Most of these rivers are punctuated by numerous study years 1985, 2000, and 2018. An accuracy assess- valley bottom wetlands that occur in their upper and ment for all of the used methods has been made by lower courses. The primary sources of all rivers in the comparing the results with high-resolution images watershed start from the wetland ecosystem (Figure 12). from Google Earth and field-collected data. The The hydrological function of wetlands determines the ground sample points (Figure 11) were overlaid on existence of rivers within the watershed and the sur- wetland maps to determine the classification accura- rounding area. As the wetland within a watershed cies and errors of each class. The overall accuracy of degrades, it will significantly impact the hydrology of the three aggregated wetlands (1985, 2000, and 2018) the watershed. Dixon (2002) had monitored the water in the study area was 86.66%, with reasonable errors of table wetlands of Illubabor. Monitoring was conducted omissions (7.54%) and low errors of commissions on 10–12 deep wells within each wetland every week. (13.33%). This showed that all the maps meet the The hydrologic analysis of the well data showed that recommended minimum 87% accuracy and there is lowering the wetland water table was observed in the a strong agreement between the reference data and the degraded and cultivated wetlands and reducing the rate classified habitat classes. of water movement through the wetlands. Analysis of degraded or cultivated wetlands versus undrained wet- land observed temporal variability and height change in Impact of wetland degradation in hydrology the weekly wetland water table (Figures 13,14). Dixon Major river and lake systems, together with their asso- (2002) indicated environmental degradation on wet- ciated wetlands, are fundamental parts of life interwo- lands unable to provide their full range of function, ven into the structure and welfare of societies and which has implications for food security in the study GEOLOGY, ECOLOGY, AND LANDSCAPES 9 Figure 7. Wetland map of Geba watershed dated 1985, 2000, and 2018. Figure 8. Vegetation types of Enago and Hamuma wetland. 10 M. BERHANU ET AL. Figure 9. Wetland degradation map of year 1985, 2000, and 2018. Figure 10. Wetland degradation graph for the period of 1985, 2000, and 2018. Table 2. Accuracy assessment for the Wetland system. Error Class GTP Commission (%) Omission (%) Overall accuracy (%) Wetland 1985 100 16 11.3 84 Wetland 2000 100 12.66 6.66 87.33 Wetland 2018 100 11.33 4.66 88.66 Aggregated 300 13.33 7.54 86.66 area and water availability to local communities, both Illubabor Zones, southwest Ethiopia has led to several around wetlands themselves and downstream. ecological and economic problems. Some of these are There is a significant difference between wetland immediate and linked to drainage, such as the scarcity water levels of cultivated/degraded wetlands and of thatching reeds, vegetation change, lowered water undrained wetlands. Hence, over-cultivation and tables, reduced accessibility, and provides unsafe water draining of wetlands within the watershed directly (Wood & Dixon, 2000). impact the water level of the wetlands. Additionally, alterations of the hydrological regime of wetlands have Driving forces and implications significant physical, chemical, and biological effects that can have significant ecological and socio- As the classification of Landsat 8 of 2018, which has economic implications at a broader scale (Bezabih & 30 m spatial resolution images, showed that in the Mosissa, 2017). The complete drainage of wetlands in study period of 2018, Forest and agricultural land GEOLOGY, ECOLOGY, AND LANDSCAPES 11 Figure 11. Verification of delineated wetland point projected from Google Earth Pro (1985, 2000, and 2018). Figure 12. Wetland interaction with drainage network. were the dominant land-use and land-cover types. According to the wetland change matrix obtained Forest and agricultural LU/LC types together from the land-use and land-cover map of 2018, the accounted for 616,700 ha (87.07%) of the study area’s majority of wetland area 3,111.5 ha (14.53%) con- total area in 2018 (Table 3). Land-use and land-cover verted typically to agricultural land from 1985 to maps quantified land-cover area is shown in Figure 15. 2000. Similarly, in the same fashion, most wetlands Wetland land-cover was cover only 12,700 ha in 2018. 17.515 ha (10.9%) converted typically to agricultural 12 M. BERHANU ET AL. Figure 13. Mean weekly water table elevation in the undrained wetlands (August 1997− July 1998). Figure 14. Mean weekly water table elevation in the drained and degraded wetlands (August 1997−July 1998). maize, teff, and sugarcane were more common among Table 3. Land-use and land-cover area for the year 2018. rural ranchers. Land-use and land-cover Area (ha) Area (%) The cultivation of wetlands is still going on in the Urban 2413.67 0.35 study area. The highlands of southwest Ethiopia Tea plant 2088.46 0.29 (Illubabur) and swamps of Awash valley are good Wetland 12,700 2.39 Forest 364,734 51.49 examples of where the farmers are engaged in produ- Shrub land 700.83 9.89 cing more than seeing sustainable use of the resources Agricultural land 70,083.6 35.59 (Dixon, 2002; Dixon & Wood, 2007). Field photo cap- ture of wetland cultivation and draining (Figure 16). Food insecurity due to pests and crop storage pro- blems, land shortages for cultivation and grazing due land (Table 4). Accuracy assessment of land-cover to coffee planting on uplands giving more people access classification of 2018 is shown in Table 5. Wetland to wetlands, and encouraging use listed as the main cultivation and degradation practiced in the watershed drivers of wetland degradation in the Illubabor zone. dominated by the cultivation of different vegetables, A previous study report of the wetlands policy briefing GEOLOGY, ECOLOGY, AND LANDSCAPES 13 Figure 15. Land-use/land-cover map of the year 2018. workshop done by Wood and Dixon (2000) described provides recreational value but also substantial eco- that wetland cultivation had been practiced in the nomic value. Because of the magnitude and ubiquity Illubabor zone for about eight decades and about 70% of Geba watershed wetland cover change effects, we are of the study farmers area have cultivated wetlands at compelled to reflect on past policy, management, and least once a year. Also, a significant correlation between other decision-making processes and improve them for the rate of rainfall and the change in wetland area were the future. The policy will need to consider actions that analyzed, a linear regression model was set, change in help the wetland ecosystem accommodate changes rainfall and wetland degradation positively correlated adaptively to improve the wetland ecosystem’s capacity with r = 0.81 for (1985–2018). The rate of change in to return to desired states after disturbance and mea- rainfall and the change rate of the wetland was posi- sures that reduce anthropogenic influences on wetland tively correlated (Figure 17). components. Wetlands as protected areas Hydro-Chemistry of Wetland water quality assessment Some wetlands were found inside protected areas in the study area. Some organizations within the watershed Wetland water contains many chemical species in saved and improved wetland management by including the dissolved state. Hamuma degraded and these areas within protected areas. For example, Bishari cultivated wetlands have lots of organic matter. Park in Mettu town results from the Rehabilitation of Based on the hydro-chemistry analysis results, wetlands by Mettu town Beshari prison center the effluent from agricultural land goes through (Figure 18). Hence, the park provides endless popular a wetland. A moderate difference in qualities of recreational activities, such as boating, birdwatching, water compared to undrained/or uncultivated wet- and many fascinating life forms, making the wetlands lands was observed (Tables 6,7 and 8). The surface especially enjoyable. Bishari Park not only contains or wetland average values of pH for the Enago, 14 M. BERHANU ET AL. Table 4. Wetland change matrix vs other land-use and land-cover areas. Change in hectare Wetland Urban Tea plantation Shrubland Agricultural land Forest Area (ha) (%) (ha) (%) (ha) (%) (ha) (%) (ha) (%) (ha) (%) Wetland conversion 16,000 74.76 42 0.19 191 0.89 258 1.20 3111 14.53 1797 8.39 1985−2000 Wetland conversion 2000−2018 12,700 79.04 152 0.94 280 1.74 149 0.92 1751 10.90 1033 6.42 Hamuma, and bake Chora was ranged from 7.45 grazing activities by the local community leads to the to 7.55, 7.15 to 8.35, and 7.15 to 8.01, respectively. hydrologic and water quality functions of wetlands in As a result of cultivation on wetlands, the rise of question. There is a significant difference in the water PH values to 7.98 of degraded Hamuma wetland quality of degraded and un-degraded wetlands. Many was received at a maximum limit of WHO water of the key issues for wetlands are related to water use. quality standard. However, partially undrained Demand for water in the study area will likely increase Enago wetland results in an average pH value of due to the growing human population (CSA, 2007), 7.3. An excess of agricultural fertilizers/ pesticides agricultural expansion, and climatic changes. Climate near a wetland and cultivated wetland increases change will also reduce water availability. Geba the value of leached chemical pollutants to watershed wetlands or watersheds cross national or Hamuma Degraded wetland represented by the regional borders, presenting both challenges and nitrogenous pollutant chemicals. The average opportunities for management. Make the science- values of Enago and Bake Chora partially policy arena more interactive, with scientists and poli- undrained wetland was ranged from 0.22 and ticians working closely together to mutual benefit 0.12 mg/N, respectively. However, the average (Toomey et al., 2017). Our horizon scan highlighted values of drained Hamuma wetlands were over perceived deficiencies in the governance of wetlands, at ranged values of nitrate nitrogen and nitrite nitro- both national scales, that are likely to continue into the gen was verified. future. Unified wetland ecosystem evaluation indica- Additionally, in terms of color and turbidity, tors are helpful for providing a guide for integrating a significant difference was recorded between wetland ecosystem evaluation data in different geogra- degraded/or cultivated wetlands and undrained wet- phical regions to improve the reliability of horizontal lands within the watershed. The average watercolor studies (Janse et al., 2019). Studies of similar geogra- values for the undrained Hamuma and Bake Chora phical regions in which the same wetland ecosystem were 19 and 38.33 mg/l pt was verified, which is accep- are examined could be evaluated to develop standar- table water quality based on WHO standards. But, dized indexes for wetland ecosystem across different a very significant average value of color was observed geographical regions. A long-term monitoring pro- in cultivated degraded wetland of 228.6 mg/l pt. The gram on wetland with field observations and remote turbidity values for uncultivated Enago were recorded, sensing will assist in gathering big data for evaluation, and 4NTU and Bake Chora Surface wetland water analysis, and management of wetland ecosystem. samples were obtained for four values that are accep- Exploring trade-offs among ecosystem service and link- table based on WHO water quality guideline standard. ing them with stakeholders can help to determine the But, again, a massive difference in turbidity between potential losers and winners of wetland management drained and undrained wetlands was observed. Based (Guida et al., 2016). This analysis will support research, on the result, average turbidity values of 18NTU were policy and practice related to environmental conserva- verified in cultivated and drained Hamuma wetlands. tion and sustainable development in the Geba Generally, wetland degradation has significant impact watershed, and provides a model for similar analyses on the surface wetland water qualities. Wetland dis- elsewhere in the world. traction resulting from draining, cultivating, and Conclusion Table 5. Accuracy assessment of land-use and land-cover The geospatial techniques provide an accurate, fast, and classification. economical way for wetland degradation in the Geba User Producer Overall watershed, southwest of Ethiopia. It has solved the Land-use type accuracy accuracy accuracy problems faced using old traditional techniques that Urban 93.75 76 84.84 Wetland 88.77 86.45 were difficult to undertake and consume a lot of time. Forest 88.54 85 Global climate change and anthropogenic impact Shrub land 85.56 84.69 Agricultural 81.91 84.61 degrade wetlands, which creates a severe problem in land identifying and quantifying wetland areas. The main Tea plantation 86.31 88.17 objective was to map the degradation of wetlands GEOLOGY, ECOLOGY, AND LANDSCAPES 15 Figure 16. Wetland cultivation and draining in the watershed. Figure 17. Spatial distribution of tendency variation of precipitation from January to December. Figure 18. Mettu Beshari Recreation Wetland Park. 16 M. BERHANU ET AL. Table 6. Water quality for Enago wetland. which have been under several pressure from both S/ anthropogenic and natural driving factors for three N Parameters Sample 1 Sample 2 Sample 3 Average consequent periods of years. Wetland change produces 1 PH 7.55 7.45 7.15 7.38 a significant impact on the surrounding environment, 2 Color 11 mg/L pt over range 14 mg/L pt 12.5 3 Turbidity 7 FTU 5 FTU 0 FTU 4 mainly influenced hydrological variation. The changes 4 Chlorine 0.23 mg/L 0.15 mg/L 2.561 mg/L 0.98 in the groundwater table of the study area are attrib- Cl Cl Cl 2 2 2 5 Chlorine total 0.28 mg/L 0.90 mg/L 0.060 mg/L 0.41 uted to wetland area change. Based on the groundwater Cl Cl Cl 2 2 2 table’s weekly measurement inside the watershed, 6 Fluoride Under 1.06 mg/ 0.75 mg/L F 0.90 a significant difference in groundwater table level was range L F 7 Manganese 0.003 mg/L 0.016 mg/L 0.0081 mg/L 0.009 recorded on degraded wetlands. The degraded wetland 8 Nitrate 0.028 mg/ 0.038 mg/ 0.605 0.22 maps that have been generated in the present study nitrogen L N L N 9 Nitrite 0.019 mg/l 0.205 mg/ 0.1375 mg/ 0.12 advance our understanding of current use, transforma- nitrogen N LN LN tion dynamics in wetlands and may provide the quan- 11 Potassium 4.9 mg/L K 12.0 mg/L 15.45 mg/L 10.78 k k titative basis needed to guide and predict future 12 Transmittance 9.6 17.3 13.76 13.55 wetland uses and their impacts on surrounding natural 13 Concentrations 1.1 0.8 2.35 1.41 14 Absorbance 1.016 0.76 0.472 0.74 resources. A balanced regulation or law needs to be put forward for agricultural development and wetland eco- system sustainability. Therefore, the study suggests urgent attention of decision-makers on conservation of the wetland’s remaining resources, taking necessary Table 7. Water quality for Hamuma wetland. measures to reduce environmental risk and new tech- S/ niques and different data fusion for exploring the N Parameters Sample 1 Sample 2 Sample 3 Average potential of geospatial data in wetland monitoring sup- 1 Ph 7.45 8.35 7.15 7.98 2 Color 210 mg/l 280 mg/l 196 mg/l pt 228.6 ported with field measurements. pt pt 3 Turbidity 10 ftu 30 ftu 15 ftu 18.33 4 Chlorine 0.77 mg/l 4.00 mg/l 0.077 mg/l 1.61 cl cl cl 2 2 2 Acknowledgments 5 Chlorine total 0.75 mg/l 3.90 mg/l 2.99 mg/l 2.54 cl cl cl 2 2 2 We are thankful to the head and staff of the School of Earth 6 Fluoride 0.60 Under Under 0.6 Sciences, College of Natural and Computational Sciences, range range 7 Manganese Over range Over range Over range Over Addis Ababa University, for providing all kinds of necessary range facilities and support during the present study. The authors 8 Nitrate 0.014 mg/l Over range Over range 0.01 would like to acknowledge the earth explorer (USGS) to access nitrogen n Landsat series images free of charge and Ethiopia – National 9 Nitrite 0.081 mg/l Over range Over range 0.08 nitrogen n Meteorology Agency, Metrological Data, respectively. We are 11 Potassium 2.8 mg/lk 0.97 mg/lk 5.0 mg/lk 2.92 also indebted to the editor Geology, Ecology and Landscapes, 12 Transmittance 19.6 9.8 14.52 14.64 and anonymous reviewers for their constructive review that 13 Concentrations 2.058 1.45 2.89 2.13 helped to improve the structure and quality of the paper. 14 Absorbance 0.27 0.22 0.97 0.59 Disclosure statement No potential conflict of interest was reported by the author(s). Table 8. Water quality for Bake Chora wetland. S/ N Parameters Sample 1 Sample 2 Sample 3 Average 1 PH 7.15 7.25 7.15 7.18 Funding 2 Color 50 mg/L pt 50 mg/L pt 15 mg/L pt 38.33 3 Turbidity 3 FTU 7 FTU 4 FTU 4.6 The authors received no direct funding for this research. 4 Chlorine 0.35 mg/L 0.68 mg/L 0.35 mg/L 0.46 Cl Cl Cl 2 2 2 5 Chlorine total 0.31 mg/L 0.75 mg/L 0.31 mg/L 0.45 Cl Cl Cl 2 2 2 ORCID 6 Fluoride 0.070 Under 0.01 0.04 Range Mintesnot Berhanu http://orcid.org/0000-0002-9547- 7 Manganese 0.05 mg/L 0.01 mg/L 0.43 mg/L 0.16 8 Nitrate 0.081 mg/ Over range 0.081 mg/l 0.08 Karuturi Venkata Suryabhagavan http://orcid.org/0000- nitrogen L N N 9 Nitrite nitrogen 0.014 mg/ Over range 0.014 mg/ 0.01 0003-2528-9106 L N L N 11 Potassium 2.8 mg/L K 0.97 mg/ 5.07 mg/ 2.94 L K L K References 12 Transmittance 8.7 7.7 18.85 11.75 13 Concentrations 3.5 0.25 0.35 1.36 Abebe, Y. D., & Geheb, K. (2003). Wetlands of Ethiopia. 14 Absorbance 1.77 0.05 1.675 1.16 Proceedings of a seminar on the resources and status of GEOLOGY, ECOLOGY, AND LANDSCAPES 17 Ethiopia’s wetlands 116. IUCN-Eastern Africa Regional Dixon, A.B., & Wood, A.P. (2007). Local institutions for office Nairobi, Kenya. wetland management in Ethiopia: sustainability and state Amsalu, T., & Addisu, S. (2014). A review of Wetland intervention. In: van Koppen B, Giordano M, Conservation and Management Policy in Ethiopian. Butterworth J (eds). Community -based water law and International Journal of Scientific and Research water resources management reform in developing coun- Publications, 4(9), 1–6. http://citeseerx.ist.psu.edu/view tries, Comprehensive Assessment of Water Management in doc/download?doi=10.1.1.654.2323&rep=rep1&type=pdf Agriculture Series 5. CABI International, Wallingford, Baker, C., Lawrence, R., Montagne, C., & Patten, D. (2006). 130–145. Mapping wetlands and riparian areas using Landsat ETM Foody, G. M. (2002). Status of land cover classification + imagery and decision-tree-based models. Wetlands, 26 accuracy assessment. Remote Sensing of Environment, 80 (2), 465–474. https://doi.org/10.1672/0277-5212(2006)26 (1), 185–201. https://doi.org/10.1016/S0034-4257(01) [465:MWARAU]2.0.CO;2 00295-4 Beven, K. J., & Kirkby, M. J. (1979). A physically based, Gebresllassie, H., Gashaw, T., & Mehari, A. (2014). Wetland variable contributing area model of basin hydrology/Un Degradation in Ethiopia: Causes, Consequences, and modèle à base physique de zone d’appel variable de l’hy- Remedies. Journal of Environment and Earth Science, 4 drologie du bassin versant. Hydrological Sciences Bulletin, (11), 40–48. 24 (1), 43–69. https://doi.org/10.1080/ Gerjevic, J. H. (2004). Using Remote Sensing Data to Study 02626667909491834 Wetland. In Iowa: University of Northern Iowa. Bezabih, B., & Mosissa, T. (2017). Review on distribution, Guida, R. J., Remo, J. W. F., & Secchi, S. (2016). Tradeoffs of importance, threats and consequences of wetland degra- strategically reconnecting rivers to their floodplains: The dation in Ethiopia. International Journal of Water case of the Lower Illinois River (USA). Sci. Total Environ, Resources Environmental Engineering, 9(3), 64–71. 572,43−55. doi:10.1016/j.scitotenv.2016.07.190 https://doi.org/10.5897/IJWREE2016.0697 Hailu, A. (1998). An overview of wetland use in Illubabur Bisrat, E., & Berhanu, B. (2018). Identification of Surface Zone, South-western Ethiopia. In A Woreda survey. Water Storing Sites Using Topographic Wetness Index Ethiopian Wetlands Research Project. (TWI) and Normalized Difference Vegetation Index Hailu, A. (2005). Ethiopian Wetlands Distribution, Benefits (NDVI). Journal of Natural Resources and Development, and Threats, Ethio-Wetlands and Natural Resources 08, 91–100. https://doi.org/10.5027/jnrd.v8i0.09 Association. In In: The Proceedings of the second Blaschke, T. (2010). Object based image analysis for remote Awareness creation Workshop on Wetlands in the sensing. ISPRS Journal of Photogrammetry and Remote Amhara Region (pp. 3–17). Addis Ababa. Sensing, 65(1), 2–16. https://doi.org/10.1016/j.isprsjprs. Islam, M. A., Thenkabail, P. S., Kulawardhana, R. W., 2009.06.004 Alankara, R., Gunasinghe, S., Edussriya, C., & Chouari, W. (2021). Wetland land cover change detection using Gunawardana, A. (2008). Semi-automated methods for multitemporal Landsat data: A case study of the Al-Asfar mapping wetlands using Landsat ETM+ and SRTM data. wetland, Kingdom of Saudi Arabia. Arab J Geosci, 14(6), 523. International Journal of Remote Sensing, 29(24), https://doi.org/10.1007/s12517-021-06815-y 7077–7106. https://doi.org/10.1080/01431160802235878 Das, K. (2017). NDVI and NDWI based change detection James, B. C., & Randolph, H. W. (2011). Introduction to analysis of Borfoibam Beelmukh Wetlandscape, Assam Remote Sensing (Fifth Edition ed.). The Guilford Press, using IRS LISS III data. ADBU-Journal of Engineering New York. Technology, 6(2), 17–21. Janse, J. H., Dam, A. A., Hes, E. M. A., Klein, J. J. M., Davidson, N. C., Van Dam, A. A., Finlayson, C. M., & Finlayson, C. M., Janssen, A. B. G., Wijk, D., McInnes, R. J. (2019). Worth of wetlands: Revised global Mooij, W. M., & Verhoeven, J. T. A. (2019). Towards monetary values of coastal and inland wetland ecosystem a global model for wetlands ecosystem services Curr. services. Mar Freshw Res, 70(8), 1189–1194. https://doi. Opin. Environ. Sustain, 36, 11–19. https://doi.org/10. org/10.1071/MF18391 1016/j.cosust.2018.09.002 Dehm, D., Becker, R., & Godre, A. (2019). SUAS based Jensen, J. R. (2007). Remote Sensing of the Environment: An multispectral imagery for monitoring wetland inundation Earth Resource Perspective. In Pearson Prentice Hall, and vegetation. Environmental Sciences. https://doi.org/ Upper Saddle River (2nd Edition ed.). New Jersey: 10.20944/preprints201911.0326.v1 Prentice Hall. Deng, Y., Chen, X., Chuvieco, E., Warner, T., & Wilson, J. P. Jones, J. W. (2015). Efficient Wetland Surface Water (2007). Multi-scale linkages between topographic attri- Detection and Monitoring via Landsat: Comparison butes and vegetation indices in a mountainous landscape. with in situ Data from the Everglades Depth Estimation Remote Sens. Environ, 111(1), 122–134. https://doi.org/ Network. Remote Sensing, 7(9), 12503–12538. https://doi. 10.1016/j.rse.2007.03.016 org/10.3390/rs70912503 Dixon, A. (2002). The hydrological impacts and sustainabil- Kaplan, G., & Avdan, U. (2017). Mapping and monitoring ity of wetlands drainage cultivation in Illubabur, Ethiopia. wetlands using sentinel-2 satellite imagery. ISPRS Annals Land Degradation and Development, 13(1), 17–31. of the Photogrammetry. Remote Sensing and Spatial https://doi.org/10.1002/ldr.479 Information Sciences, 4(4), 271–277. https://doi.org/10. Dixon, A., Wood, A., & Hailu, A. (2021). Wetlands in 5194/isprs-annals-IV-4-W4-271-2017 Ethiopia: Lessons From 20 years of Research, Policy and Kaplan, G., & Avdan, U. (2019). Evaluating the utilization of Practice. Wetlands, 41(2), 20. https://doi.org/10.1007/ the red edge and radar bands from Sentinel sensors for s13157-021-01420-x wetland classification. Catena, 178,109–119. doi:10.1016/ Dixon, A. B., . &., & Wood, A. P. (2003). Wetland cultiva- j.catena.2019.03.011 tion and hydrological man-agreement in East Africa: Kulawardhana, R. W., Thenkabail, P. S., Vithanage, J., Matching community and hydrological needs through Biradar, C., Md.A, I., Gunasinghe, S., & Alankara, R. sustainable wetland use. Natural Resources Forum, 27 (2007). Evaluation of the Wetland Mapping Methods (2), 117–129. https://doi.org/10.1111/1477-8947.00047 using Landsat ETM+ and SRTM Data. Journal of 18 M. BERHANU ET AL. Spatial Hydrology, 7(2), 62–96. https://hdl.handle.net/ Terefe, D. (2017). Ethiopia: Why Conservation of Wetlands 10568/21603 Makes Sense. Water Journalists Africa. (https://waterjour Lane, C. R., Liu, H., Autrey, B. C., Anenkhonov, O. A., nalistsafrica.com/2017/02/ethiopia-why-conservation- Chepinoga, V. V., & Wu, Q. (2014). Improved ofwetlands-makes-sense/#respond ) Wetland Classification Using Eight-Band High Toomey, A. H., Knight, A. T., & Barlow, J. (2017). Resolution Satellite Imagery and a Hybrid Approach. Navigating the space between research and implementa- Remote Sensing, 6(12), 12187–12216. https://doi.org/ tion in conservation: Research-implementation spaces. 10.3390/rs61212187 Conserv Lett, 10(5), 619–625. https://doi.org/10.1111/ Leykun, A. (2003). The distribution and status of Ethiopian conl.12315 wetlands: An overview. In Y. D. Abebe & K. Geheb (Eds.), Veraverbeke, S., Verstraeten, W. W., Lhermitte, S., & In Wetlands of Ethiopia, proceedings of a seminar on the Goossens, R. (2010). Illumination effects on the differ - resources and status of Ethiopia’s wetlands. IUCN. enced Normalized Burn Ratio’s optimality for assessing Leykun, A. (2003). The distribution and status of Ethiopian fire severity. International Journal of Applied Earth wetlands: An overview. Proceedings of a seminar on the Observation and Geoinformation, 12(1), 60–70. https:// resources and status of Ethiopia’s wetlands (pp. 12. vi + doi.org/10.1016/j.jag.2009.10.004 116). IUCN Eastern Africa Regional Office. Kenya. Verbyla, D. L., Kasischke, E. S., & Hoy, E. E. (2008). Seasonal Li, N., Li, L., Lu, D., Zhang, Y., & Wu, M. (2019). Detection and topographic effects on estimating fire severity from of coastal wetland change in China: A case study in Landsat TM/ETM+ data. International Journal of hangzhou bay. Wetlands Ecology and Management, 27 Wildland Fire, 17(4), 527–534. https://doi.org/10.1071/ (1), 103–124. https://doi.org/10.1007/s11273-018-9646-3 WF08038 Mengesha, T. A. (2017). Review on the natural conditions Wang, G., & Weng, Q. (2014). Remote Sensing of Natural and anthropogenic threats of Wetlands in Ethiopia. Resources. In Taylor & Francis Series in Remote Sensing Global Journal of Ecology, 2(1), 006–014. https://doi.org/ Applications. Indiana State University Terre Haute, 461. 10.17352/gje.000004 Gondar Indiana, U.S.A.: CRC Press Taylor & Francis Group. Mitsch, W. J., & Gosselink, J. G. (2007). Wetlands (4th https://doi.org/10.1201/b15159 Edition ed.). John Wiley & Sons, Inc.. Wang, Y., Hou, X., Wang, M., Wang, M., Wu, L., Ying, L., & Nsubuga, F. W. N., Botai, J. O., . &., Olwoch, J. M., Feng, Y. (2012). Topographic controls on vegetation Rautenbach, C. J. D., Kalumba, A. M., Tsela, P., index in a hilly landscape: A case study in the Jiaodong Adeola, A. M., Sentongo, A. A., & Mearns, K. F.. (2017). Peninsula, eastern China. Environmental Earth Sciences, Detecting changes in surface water area of Lake Kyoga 70(2), 625–634. https://doi.org/10.1007/s12665-012- sub-basin using remotely sensed imagery in a changing 2146-5 climate. Theoretical and Applied Climatology, 127(1–2), Woldu, Z., & Yeshitela, K. (2003). Wetland plants in 327–337. https://doi.org/10.1007/s00704-015-1637-1 Ethiopia with examples from Illubabor, south-western Ozesmi, S. L., & Bauer, M. E. (2002). Satellite remote sensing Ethiopia in: Yilma D. Abebe and Kim Geheb (2003) of wetlands. Wetlands Ecology and Management, 10, 381– Wetlands of Ethiopia: Proceedings of a seminar on the 402. https://doi.org/10.1023/A:1020908432489 resources and status of Ethiopia’s wetlands. IUCN. Ozesmi, S. L., & Bauer, M. E. (2002). Satellite remote sensing Wood, A. P., & Dixon, A. B. (2000). Sustainable wetland of wetlands. Wetlands Ecology and Management, 10(5), management in Illubabor Zone: Research report summa- 381–402. https://doi.org/10.1023/A:1020908432489 ries. Unpublished Report, The University of Huddersfield . Sakane, N., Alvarez, M., Becker, M., Böhme, B., Handa, C., Available via: https://www.wetlandaction.org/wp-content Kamiri, H. W., Langensiepen, M., Menz, G., Misana, S., /uploads/EWRP-Objective-Summaries.pdf . Mogha, N., Möseler, B., Mwita, E., Oyieke, A. H., & van Wu, Q.2018 GIS and Remote Sensing Applications in Wijk, M. (2011). Classification, characterization, and use Wetland Mapping and Monitoring.B. Huang,Ed.. of small wetlands in East Africa. Wetlands, 31(6), Comprehensive Geographic Information 1103–1116. https://doi.org/10.1007/s13157-011-0221-4 Systems.2 140–157.Oxford: Elsevier.https://doi.org/10. Suding, K. N. (2011). Toward an Era of Restoration in Ecology: 1016/B978-0-12-409548-9.10460-9 Successes, Failures, and Opportunities Ahead. Annual Review Xu, H. (2006). Modification of normalised difference water of Ecology, Evolution, and Systematics, 42(1), 465–487. https:// index (NDWI) to enhance open water features in remo- doi.org/10.1146/annurev-ecolsys-102710-145115 tely sensed imagery. International Journal of Remote Suryabhagavan, K. V. (2017). GIS-based climate variability Sensing, 27(14), 3025–3033. https://doi.org/10.1080/ and drought characterization in Ethiopia over three 01431160600589179 decades. Weather and Climate Extremes, 15, 11–23. Zhang, Z., Zimmermann, N., Kaplan, J., & Poulter, B. https://doi.org/10.1016/j.wace.2016.11.005 (2016). Modeling spatiotemporal dynamics of global wet- Tarboton, D. G. (1997). A New Method for the Determination lands: Comprehensive evaluation of a new sub-grid of Flow Directions and Contributing Areas in Grid Digital TOPMODEL parameterization and uncertainties. Elevation Models. Water Resources Research, 33(2), 309–319. Biogeosciences, 13(5), 1387–1408. https://doi.org/10. https://doi.org/10.1029/96WR03137 5194/bg-13-1387-2016

Journal

Geology Ecology and Landscapes – Taylor & Francis

Published: Oct 2, 2023

Keywords: Agriculture; Climate; Wetland loss; Water quality; Water level

There are no references for this article.