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

Learn More →

Comparative evaluation of models to estimate direct runoff volume from an agricultural watershed

Comparative evaluation of models to estimate direct runoff volume from an agricultural watershed GEOLOGY, ECOLOGY, AND LANDSCAPES 2021, VOL. 5, NO. 2, 94–108 INWASCON https://doi.org/10.1080/24749508.2020.1833629 RESEARCH ARTICLE Comparative evaluation of models to estimate direct runoff volume from an agricultural watershed a b c Kishan Singh Rawat , Sudhir Kumar Singh and Szabo Szilard a b Geo-Informatics, Civil Engineering Department, Graphic Era (Deemed to Be University), Dehradun, India; K. Banerjee Centre of Atmospheric Ocean Studies, IIDS, Nehru Science Centre, University of Allahabad, Allahabad, India; Professor, Head of Department, Department of Physical Geography and Geoinformation Systems, University of Debrecen, Debrecen, Hungary ABSTRACT ARTICLE HISTORY Received 20 April 2020 Generally, runoff records are the most important input data in water resource management; Accepted 4 October 2020 however, their availability is very limited especially in developing country as compared to rainfall records, especially under medium and small-scale catchments. In our study, we esti- KEYWORDS mated runoff from ungauged agricultural watershed with the curve number method and Curve number; runoff; RMSE; empirical mathematical models were compared with SCS-CN. Empirical mathematical models soil erosion; ungauged (Inglis and De Souza Formula (IDS), Turc relationship (TR), Indian Irrigation Department (DII) watershed model, Coutagine relationship (CR), Khosla method (KH), Justin Equation (JE), Lacey relation- ship (LR), and Indian Council of Agricultural Research (ICAR)) model were used to estimate annual runoff (in cm). It was found that IDS model has capability to simulate annual runoff as very close to Soil Conservation Service Curve Number (SCS-CN) model and has lowest Root Mean Square Error (RMSE) value as 7.75, and ranking of this model (based on K factor (value of st 0.001) analysis) was topmost (or 1 ) in comparison to other eight models. This study suggests that empirical mathematical model has potential for annual runoff estimation from ungauged watershed. 1. Introduction (2015), a hydrological model is a simplification of a real-world system, used mainly for the prediction of Soil erosion is one the most important process that hydrological processes based on rainfall, drainage area endangers the soil quality and, therefore, the agricul- (topography), soil properties, vegetation cover, and tural production (Pásztor et al., 2016; Waltner et al., runoff model. It is defined as a set of equations that 2018). Runoff has strong relationship with the rate of enable the estimation of runoff as a function of various erosion, as it is the consequence of the precipitation parameters used for describing watershed (duration, & intensity; Mohamadi & Kavian, 2015), characteristics. the slope characteristics (steepness, length, & shape), The Universal Soil Loss Equation (USLE) is an soil characteristics (infiltration capacity, depth of empirical equation. The Revised Universal Soil Loss humus layer, particle size& initial water content; Equation (RUSLE) is a modification of USLE, espe- Centeri et al., 2015; Szabó et al., 2015), and the vegeta- cially for more complex situations of rill and inter-rill tion/land use (management, density, leaf-area index, erosion in conservation planning and land uses. Both arboreal, or herbaceous; Jakab et al., 2013; Tadesse et erosion-prone models calculate detachment capacity al., 2017). and soil loss. RUSLE model predicts soil degradation Simple empirical equation relates catchment char- and sediment concentrations better using another soil acteristics and complicated physical models are avail- erodibility factor (F-soil factor, based on soil texture). able to estimate the catchment runoff . The application The soil conservation service-curve number (SCS-CN) of conceptual hydrological model to generate runoff method has been used widely (Bérod et al., 1999; from ungauged watershed with limited data have been Pandey & Dabral, 2004; Vaze et al., 2011). The SCS- studied by researchers in past (Kaleris et al., 2015). CN method is simple, predictable, stable, and relies on Regional scale model can explain the variation of the only one parameter, namely the CN. The land use/ model parameters with physiographic factors. These land cover (LULC) class can be integrated with the models did not fully capture the local scale process and hydrologic soil groups (HSG) of the sub basin in GIS, variations. However, the certainty of the calibrated and the weighted CN can be estimated. These esti- model parameters is high enough to simulate the mated weighted CN for the entire area can be used to hydrologic response of ungauged watershed . compute runoff. Moore and Clarke (1981) showed According to Wheater et al. (2008) and Devia et al. CONTACT Sudhir Kumar Singh sudhirinjnu@gmail.com K. Banerjee Centre of Atmospheric Ocean Studies, IIDS, Nehru Science Centre, University of Allahabad, Allahabad, U.P, India © 2020 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. GEOLOGY, ECOLOGY, AND LANDSCAPES 95 that a variety of distributions that can be easily incor- to SCS-CN for generating annual runoff of ungauged porated into this type of model structure and they Jhagrabaria watershed using satellite data. derive analytical equations for the response of differ - ent distributions. Hosking and Clarke (1990) extended 2. Study area and data the work of Moore and Clarke (1981), and reported that the model can be used to derive a relationship The Jhagrabaria watershed is located in the Allahabad between the frequencies of storm rainfall and flow district of state Uttar Pradesh, India (Figure 1). peak magnitude in an analytical form. The UK Geologically the area consists of Upper Vindhayan Institute of Hydrology has shown the model for long formation consist of mainly sandstone and shale. The runs to derive flood frequencies (Dorum et al., 2010; elevation is ranging from 85 to 192 m above mean sea Lamb, 1999). Recently many studies have applied level with nearly flat to gently undulated topography machine learning and soft computing approaches to and small occasional hillocks. The upland area is cov- study the soil properties and erosion (Jahani et al., ered with loam, except in the south-western part of 2016; Singh et al., 2020; Mosaffaei et al., 2020; tehsil Karchhana, where the soil is a mixture of clay Rahmati et al., 2020). and marsh. The area has semi-tropical climate as The earth observation datasets integration within summer and the winter. The area receives about 91% Geographical Information System (GIS) make of the total annual rainfall due to southwest monsoon watershed modeling easy and accurate (Balázs et al. from June to September. The relative humidity is high 2018). The capabilities of these technologies have been during the monsoon month (Rawat & Singh, 2017). successfully utilized by many researchers in rainfall- runoff modeling . Earth-observing satellite provides more reliable input parameters for hydrological mod- 3. Materials and methods eling (Rawat & Singh, 2017; Maliqi & Singh, 2019). 3.1. Datasets used However, GIS processing has become a critical step in hydrologic modeling (Thakur et al., 2017), since it LANDSAT7 ETM+ (path/row: 231/67) was acquired contributes to generate model parameter distribution on June 27, 2006 (Table 1). The image was converted in spatial manner. to apparent reflectance through an image-based calibra- Although researchers delineated several models, the tion method. Atmospheric correction was performed satellite-based inputs in these models were not com- using Fast Line-of-sight Atmospheric Analysis of paratively used and limited model performance was Hypercubes (FLAASH) algorithm. Image was geome- evaluated. Hence, objectives of work are as follows: (i) trically rectified using ground control points collected to estimate daily runoff using SCS-CN and (ii) to find from Survey of India (SOI) topographic sheets using an optimal empirical mathematical model with respect nearest-neighborhood resampling technique and a Figure 1. Location of the study area (Jhagrabaria agriculture watershed), U.P., India. 96 K. S. RAWAT ET AL. Table 1. Specifications of LANDSAT (TM and ETM ) sensors used in the present study. Spatial resolution (meters) Spectral resolution (µm) Spectral bands TM ETM+ TM ETM+ 1 (Blue) 30 30 0.45–0.52 0.45–0.52 2 (Green) 30 30 0.52–0.60 0.53–0.61 3 (Red) 30 30 0.63–0.69 0.63–0.69 4 (Near IR) 30 30 0.76–0.90 0.78–0.90 5 Shortwave Infrared (SWIR) 1 30 30 1.55–1.75 1.55–1.75 6 (Thermal IR)* 120* (30) 60 * (30) 10.4–12.5 10.4–12.5 7 Shortwave Infrared (SWIR) 2 30 30 2.08–2.35 2.09–2.35 8 (Panchromatic)** 15 0.52–0.90 * TM Band 6 was acquired at 120-meter resolution, but products are resampled to 30-meter pixels. * ETM+ Band 6 is acquired at 60-meter resolution, but products are resampled to 30-meter pixels. root-mean-square error with less than one pixel was clipped with the administrative boundary of obtained during the geometric rectification. Land use/ Allahabad district’s study area, LST data were land cover (LULC), viz. barren land, fallow land, vege- extracted. tation, and water bodies/wetlands were identified in the The soil map of the Shankargarh block was col- field and their coordinates were recorded with a hand- lected from Soil Survey Department, Allahabad, U.P., held global positioning system (GPS) device (Garmin India. The map was scanned and then registered with eTrexH). The maximum likelihood classifier is a simple the help of geo-referenced Survey of India (SOI) topo- and easy to use classification algorithm, in which a pixel graphical sheet no. 63 G/11 and 63 G/12, respectively. with the maximum likelihood is classified into the The registered soil maps were digitized and different corresponding class (Singh et al., 2017; Lu et al., soil attributes were assigned to the different soil 2004). Afterward window 3 × 3 size majority filter was groups in digital format. In present study, CN map is applied to remove the “salt and pepper” noise from generated with help of LULC and HSG map, CNII is classified image. the CN for normal condition, CNI is the CN for dry The need of satellite-estimated precipita - condition, CNIII is the CN for wet condition and CN tion arises because of the non-availability or is assigned based on Section 2C-5 – Iowa Storm water poorly distributed ground rainfall data. For Management 2C-5 Manual (2C-5 NRCS TR-55 the work, the daily precipitation data were Methodology) (2008). downloaded from ftp://ftpprd.ncep.noaa. gov/pub/cpc/fews/S.Asia/. Resolution of 3.2. LST role in models rainfall estimates are of 0.1 × 0.1 degree and inputs include Global LST data sets are important because five models (TR, Telecommunication System (GTS) station CR, KH, JE, and ICAR) out of eight (KH, IDS, DII, TR, data, as well as GOES Precipitation Index CR, ICAR, JE, and LR) models required LST as input (GPI) infrared cloud top temperature fields data, to predict runoff. Average function was applied derived from Meteosat and polar-orbiting to calculate monthly and annual LST. In TR tempera- satellite precipitation estimate data from ture is part of denominator, and it is also under square Special Sensor Microwave/Imager (SSM/I) root function therefore its effective yield will be small, on board Defense Meteorological Satellite over all denominator will be a small quantity which Program and Advanced Microwave gives a little fraction of annual rainfall, net result will Sounding Unit (AMSU-B) on board come as high runoff from TR. In CR temperature is NOAA15, 16 and 17. also part of denominator, and it does not has any Land surface temperature (LST) is an important constrain (like square root function); thus, a good parameter in study of water resources. Data available yield will apply in denominator which gives small over tile (1100 km x 1100 km) of 2003–2014; the total fraction of annual rainfall, resulting in the overestima- 495 (per year 45) files, emissivity and quality control tion of runoff. KH model reveals a low annual runoff (QC) files were downloaded from http://glovis.usgs. because a major part is subtracted from annual rainfall gov. LST values were retrieved based on the Split (T/3.74 (in °C)), and will be a big quantity). From JE Window algorithm. Hierarchical Data Format (HDF) model, in denominator temperature has multiple fac- files of 8-day LST were stacked of each year and study tor of 1.8 additional 32 which will give large number at area was subset from tiles and from converted sinu- dominator; therefore, a small yield in JE; Thus, this soidal (SIN) to Universal Transverse Mercator (UTM, model have better result from other models being WGS84) projection in EVNI software. Average 8 days temperature-dependent (TR, CR, KH, and ICAR). images (spatial and temporal) were used to generate ICAR model reveals that temperature is part of monthly LST. The 8-days India’s LST data for 12 years denominator and it is also multiplied by another factor period (2003–2014) were downloaded and later it were which gives a big yield in denominator; therefore, a GEOLOGY, ECOLOGY, AND LANDSCAPES 97 less net annual runoff from ICAR. ICAR may be good overestimation or underestimation of the statistics. for a regional area because it is directly dependent on Mean Absolute Error (MAE, Rawat et al., 2020) is a area, slope and other factors that dominate at regional measure of how models are varied from SCS-CN. scale. Hence, annual runoff fluctuates if annual mean MAE is a more natural measure of average error and surface temperature slightly varies because all equa- is unambiguous. Percentage Bias (PBIAS; error index tions are directly linked to surface temperature. for model, Rawat et al., 2019) measures the average tendency of the simulated data to be larger or smaller than their observed from SCS-CN model. Mean 3.3. Runoff estimation Difference (BIAS, Rawat et al., 2020) is difference between model’s value and value from SCS-CN, if The SCS-CN method was developed to estimate sur- difference is zero, it is called unbiased otherwise face runoff from small agricultural watersheds biased. A low Mean Bias Error (MBE, Rawat et al., (USDA-SCS, 1967). The soils have been classified 2020) is desired; ideally a zero value of MBE should be into four hydrologic groups namely A, B, C, and D obtained. A positive value of MBE shows an over (USDA, 1986), based on infiltration, soil classification, estimate with respect to SCS-CN surface runoff and other criteria (soil’s surface condition (infiltration whereas a negative value show an under-estimate rate) and its horizon (transmission rate). Land use and with respect to SCS-CN surface runoff. management types have been used in the preparation of hydrological soil-cover complex, which has been utilized in estimating direct runoff. Antecedent 3.5 Ranking of empirical mathematical models Moisture Condition (AMC) is an indicator of Factor K was estimated to provide proper weight watershed wetness and availability of soil moisture (Rawat and Singh, 2018) to selected statistical index storage prior to a storm (Rawat & Singh, 2017). SCS (all used statistical test) as: has developed a guide for adjusting CN according to " , # AMC based on total rainfall in the 5 day period pre- � � ceding a storm. Three levels of AMC as: AMC-I (dry), K ¼ 1 (10) AMC-II (normal), and AMC-III (wet) conditions. The n¼1 seasonal rainfall limits for these three antecedent W ¼ and moisture conditions (Table 2). Many hydrologists have discussed relationships of precipitation and 1 ¼ W annual surface runoff with the assumption that physi- cal characteristics of the watershed are constant (Castiglioni et al., 2010). The brief information about n Rating ¼ ðW i Þ (11) the empirical models applied in this study is presented n n n¼1 in Tables 3–4. th where, K is factor, i is n statistical index and W is weight for particular statistical index. Lowest rating 3.4 Performance evaluation model will be on first rank and vice-versa. Model's performance was evaluated using statistical parameters. The reference values were taken of SCS- 4. Results and discussions CN model. Coefficient of Determination (R , Rawat et al., 2020) describes the dispersion of models vs. SCS- 4.1 Land use/land cover (LULC) CN model. The Root Mean Square Error (RMSE, LULC affects the infiltration, erosion, and evapo- Rawat et al., 2020) values define how models over- transpiration hence, it is an important character- estimate or underestimate the measurements with istic of runoff process. Overall 90% accuracy of respect to SCS-CN model. Relative Root Mean classified LULC map was achieved. The area of Square Error (R-RMSE; Rawat et al., 2020) is a stan- 2 2 barren land (36.91 km ), fallow land (36.62 km ), dardization of RMSE. R-RMSE value is expressed in and vegetation (74.71 km ) (Figure 2). The fallow percent and represents the standard variation of the and barren land together have the highest area as model. The R-RMSE assigns equal weight to any 48.99% whereas vegetation area is 47.81%. The area exposed for erosion offer high rate of water Table 2. Classification of antecedent moisture conditions erosion. Several studies have demonstrated the (AMC). role of LULC in hydrologic modeling and runoff Total 5 days Antecedent Rainfall (mm) estimation (Adham et al., 2014; Tedela et al., 2012; AMC Dormant Season Growing Season Kumar et al., 2018). I < 12.7 < 35.6 II 12.7 − 27.9 35.6–53.3 III > 27.9 > 53.3 98 K. S. RAWAT ET AL. Table 3. Specification of empirical mathematical models (EMM), the purpose, mathematical expression and references. Sr. no. EMM Purpose/reason Mathematical Expression Reference 1 SCS-CN method Developed to estimate surface runoff from small agricultural watersheds ðP 0:2SÞ SCS-CN model Q ¼ ðP< 0:2SÞ (1) Pþ0:8S in USDA, Q = is direct runoff (mm), P total precipitation (mm), S is watershed storage (1967) ðP 17:8Þ� P 2 Inglis and De Souza Plains of Maharashtra region of India Mutereja, R ¼ (2) (IDS) (1986) Where, P is annual precipitation (cm), and R is annual runoff (cm) Bavishi and Bhagat (2017) 0:86 3 Indian Irrigation Indian Irrigation Department uses the relationship equation between Rainfall and Praveen Kumar R ¼ P 1:17� P (3) Department (DII) Runoff et al. (2016) Where, P is annual precipitation (cm), and R is annual runoff (cm) rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4 Turc relationship (TR) Relationship for watersheds with the area less than 300 km based on achieved results Khosravi et al. � � R ¼ P D (4) D ¼ P LT ¼ 300þ 25� T þ 0:05� T Where, P is annual from doing a study on 254 watersheds in various climatic and weather conditions (2013) 0:9þ = LT precipitation (cm), R is annual runoff (cm), T is mean annual temperature (°C) and D is annual flow shortage. 5 Coutagine relationship Presented a general relationship Alizade, (1989) R ¼ P λ� P (5) (CR) Sadati et al. Where, P is annual precipitation (cm), R is annual runoff (cm), is coefficient and recommended 1 (2014) as:-λ ¼ = Where, T is mean annual temperature (°C). ð0:8þ 0:14� TÞ 6 Khosla method (KH) Amount of mean annual runoff Golshan and R ¼ P = (6) 3:74 Ebrahimi Where, P is annual precipitation (cm), R is annual runoff (cm) and T is mean annual temperature (° (2014) C). Khosravi et al. (2013) 0:155 2 H H 0:284� S � P max min pffiffi 7 Justin Equation (JE) Estimate runoff with considering parameters such as mean temperature, slope and R ¼ , s ¼ (7) Golshan and 1:8� Tþ32 precipitation in watershed Where, R is annual runoff (cm), P is mean annual precipitation (cm), T is Mean annual temperature Ebrahimi (°C), S is mean slope of watershed, H is elevations of watershed, A is the watershed area (2014) 1:44 1:115� P 8 Indian Council of Based on 17 watershed annual runoff in Nilgiri region that was conducted by Indian R ¼ (8) 1:34 0:0613 T � A Agricultural Council of Agricultural Research Sadati et al. Where, P is annual precipitation (cm), R is annual runoff (cm), A is the watershed area, T is mean Research (ICAR) (2014) annual temperature (°C). 9 Lacey relationship Lacey an Indian scientist investigated several indian watersheds to prepare a Lacey Bavishi and R ¼ (9) 304:8ðF Þ 1þ (LR) equation to estimate annual Bhagat R is mean annual runoff (cm), P is Mean annual precipitation (cm), F is Parameter of rainfall Runoff (2017) duration and physiographic properties, Values of Fz coefficient are given in Table 4. Golshan and Ebrahimi (2014) GEOLOGY, ECOLOGY, AND LANDSCAPES 99 Table 4. Values of F coefficient. observed. This low rainfall in those years leads severe Duration of rainfall to moderate drought. After year 2004 each year one or Catchment area Long Average Short more than one rainfall event (~80 mm) exists except Includes shelf, flat plains with deep soils and 6 4 2 year 2008 to 2011. It reveals that maximum rainfall was vegetation appropriate recorded 118.81 and 120.67 mm, respectively, during Somewhat flattened with deep soils and 2.5 1.67 0.83 pasture vegetation. 05-Sept-2007 and 05-October-2013. Figure 4(d) Relatively high hills with shallow soils and 1.5 1 0.5 (monthly rainfall) showed rainy and non-rainy months vegetation is relatively weak Sand, gravel and steep terrain with plenty of 0.88 0.58 0.23 (2003–2014), it helps to understand shifting of rainfall height pattern. Such types of patterns are absent during 2003 High and steep rocky terrain with no 0.43 0.28 0.14 to 2014, however, some non-rainy months with low vegetation rainfall amount (April-06 (20.28 mm), May-09 4.2. Soil map and hydrological soil group (20.54 mm) and May-011 (13.33 mm) was observed. Whereas few high (Feb-07 (26.57 mm) and Feb-2013 The soil of the Jhgarbaria watershed is of Devra clay (34.36 mm)) or high (October-2006 (47.04 mm), soil, Jarkhori sandy loam, Lohgara silty loam, Newaria October-2014 (74.23 mm)) amount of rainfall is also loamy and stony land (Figure 3(a)). The watershed is reported. Figure 4(e), represent monsoon season rain- mainly dominated by Newaria loam (58.26 km or fall pattern and year 2004, 2009, and 2010 (low rainfall 38.82%), Devra clay soils (31.92 km or 21.27%) and with respect to average rainfall). Maximum rainfall in Jarkhori sandy loam soils (27.99 km or 18.65%). last 13 years was recorded for August-2013 as Presence of sand fraction in large quantities under 532.52 mm while minimum rainfall for same year in entire watershed makes it vulnerable to soil erosion. month of September (except month of September- The stoniness of the land (14.63 km or 9.75%) will act 2008). Based on 12 years rainfall data sets, average rain- as barrier to store water however leading to generate fall in study area is 199.57 mm (296.1 mm, from IMD higher amount of runoff (Castiglioni et al., 2010; web http://www.imd.gov.in/section/climate/clima Fathzadeh, 2008).The initial infiltration and transmis- teimp.pdf), which is less (97.53 mm) than from sion of surface water into an aquifer system is a func- 100 year average monsoon (June, July, August and tion of soil type and its texture. From soil classes, September) rainfall. Based on 12 years monsoon further Hydrologic Soil Group (HSG) (Figure 3(b)) month rainfall, average rainfall of month of June, July, map of study area was developed with guidelines August, and September are respectively, 139.49 (88.8 given by Chow et al. (1988). mm, from IMD), 251.86 (280 mm, from IMD), 245.73 (296.1 mm from IMD) and 161.21 mm (185.0 mm from IMD). Figure 4(f), illustrates annual rainfall from 2003 4.3. Rainfall (2003-2014) to 2014, it represents drought years and particular year The daily rainfall data of 12 years are illustrated in rainfall can correlated with particular year crop produc- Figure 4(a-c). These figures show accumulated rainfall tion. Figure 4(f) also explained the drought years (2004, over time (per day from 2003 to 2014) and low rainfall 2009, and 2010) of study area. (<600 mm) in the year 2004, 2009, and 2010 was Figure 2. (a). Land use/land cover map of 2006 of study area. 100 K. S. RAWAT ET AL. Figure 3. (a) Soil types in the study area and their spatial distribution. (b). The Hydrologic Soil Group (HSG) map developed from soil map. 4.4. Land surface temperature (LST) 2013. Hence, the maximum variation was only 1.8° C in 11 years. The 8-days LST was plotted (Figure 5(a-c)) and maximum temperature variation (32.8 to 41.3°C) was noted during 14-April-2010 to 25-Jun-2010 4.5. Runoff from CN method th th (335 to 344 8-days). Figure 5(d) shows monthly LST and reveal that average monthly LST eight Figure 6(a-c) shows destitution of CNn at special times cross 35°C limit line during different month extent and corresponding histogram showing desti- of different years, and maximum monthly average tution of CN at pixel wise (n = I, II, and III) in LST was noted for during June-2010 as 37.3°C. Due images. Runoff calculation from SCS model mainly to average function all peak values (in 8-days LST relied on CN value, which is a function of AMC, data sets, Figure 5(c)) all value range from 14.5 to slope, soil type, and land use. The CN value reflects 37.28°C, while monthly average mean value is the possible runoff generation (Rawat & Singh, noted as 25.3°C. Figure 5(e), represents annual 2017). Under the same rainfall condition, low LST, maximum LST 26.1°C was estimated in the value of CN reflect that the land has a high possi- year 2010 and minimum was 24.3°C in the year bility of water-holding capacity. While high value GEOLOGY, ECOLOGY, AND LANDSCAPES 101 Figure 4. (a). Shows an example of the merged analysis of daily precipitation for 20 July 2001. The merged analysis presents similar spatial distribution patterns with those of satellite estimates while its magnitude is close to the gauge-based analysis over areas with gauge data, (b) shows a final product after merging the inputs. (c). daily rainfall from 2003 to 2014. (d). monthly rainfall from 2003 to 2014. (e). monthly rainfall from 2003 to 2014 during monsoon season. (f). annual rainfall from 2003 to 2014. of CN, precipitation can be held by the land at a of total rainfall was converted into runoff, it comes small extent. Therefore, any class LULC with high as large amount of runoff (60.5% runoff) in next value of CN can generate a high amount of runoff month (September) by 159 mm rainfall. Similarly, which will cause of flood peak. In SCS model, AMC for high runoff (55.9% of total rainfall) during condition has influence on CN values that’s why June-2005 (because high amount of rainfall receives CN and AMC conditions are two major factors that in last days of previous month (22, 23, 25, 26, 27 can affect the runoff analysis in SCS modeling. 29, and 30 May-2005). Figure 7(b) reveals that Figure 7(a) represents a seasonal trend, the varia- during year 2013 October rainfall also produces bility in runoff except high runoff during 05- high runoff 51.9% of 293 mm rainfall, this September-07 (90.11 mm, because of high rainfall October’s runoff is given key information that a 118 mm), 05-October-2013 was noted as highest large amount of rainfall after September becomes runoff (91 mm) due to highest rainfall (120 mm) as runoff because of surface saturation condition. during end of monsoon year of 2013. This was Figure 7(c) graphical representation of annual run- special month (October) of last ten years (2003 to off with rainfall and explain rainfall and runoff 2012) when more rainfall in short time period yield of 12 months. (near about 135 mm within two days) and in year 2010 less runoff. Figure 7(b) represents monthly 4.7. Runoff from surface runoff model monsoon runoff during 2003 to 2014 and showed highest runoff (96 mm and 60.5% of total rainfall Annual runoff was estimated by eight differ - (159 mm)) during September-2007, because in ent surface runoff models (KHM, IDS, TR, August-2007 high amount of rainfall (309 mm) CR, KH, ICAR, LR and JE). Table 5 was received (total 16 days rainfall) but only 22% 102 K. S. RAWAT ET AL. Figure 4. (Continued) describes the comparative runoff results of runoff. However, CR and IDS, the remaining these models. These eight models were inde - models’ runoff predictions are under the pendent of LULC classes, soil categories, acceptable limit (based on % of annual and AMC type. These models are only rainfall). based on annual rainfall and annual tem - perature. From Table 5 , we can easily dis - 4.8. Statistical performance evaluation tinguished two categories, (i) predicted annual runoff was overestimated (CR and Comparative results of runoff estimation are obtained KH) and (ii) predicted annual runoff was through statistical tests (Table 6). Statistical results for underestimated (IDS, TR, and ICAR). Khosla’s method (R = 0.92, RMSE = 49.83, R- Predicted runoff of CR model was always RMSE = 2.3, MAE = 47.64, NRMSE = 1.99, overestimated (because in each year the pre - MBE = −47.64, PBIAS = −65.58, and BIAS = −2.17) dicted runoff was more than the actual pre - with respect to SCS-CN (RS and GIS-based model) has cipitation, like runoff of year 2013 is 34.72% been rejected based on rating of statistical index of annual rainfall), therefore, in first screen - method, because it scored high value of 4.911 ing this model can be discarded. In the same (Rank = 7) which also indicated that based on only way, KH model also predicted high annual R test any model cannot known fully or used as good GEOLOGY, ECOLOGY, AND LANDSCAPES 103 Figure 5. (a). Shows an example of the eight days Land SurfaceTemperature (LST) of India from MOD11A2, for 20 July 2014, (b) extracted LST for Allahabad district of Uttar Pradesh State, India. (c). 8-days LST data set of study area for year 2003 to 2014. (d). Monthly LST data set of study area for year 2003 to 2014. (e). Annual LST data set of study area for year 2003 to 2014. predictor or estimator. Because despite of high value (R = 0.89, RMSE = 7.75, R-RMSE = 0.24, of R = 0.92, the model comes under rank 7. Statistical NRMSE = 0.31, PBIAS = 0.01, BIAS = 0.07, results of IDS model (for plain area) with respect to MAE = 0.002, and MBE = 0.002, (≈ 0, almost zero)) SCS-CN is good because of all statistical tests result have positive responses with respect to SCS-CN 104 K. S. RAWAT ET AL. Figure 6. (a). Curve number (CN) II map and corresponding histogram. (b). Curve Number (CN) I map and corresponding histogram. (c). Curve Number (CN) III map and corresponding histogram. model and the rating process was having the lowest runoff. This may be the reason for adoption of this value of 0.008 (K = 0.001 with first Rank), further- model by JE. The performance of CR model was also more, it revealed that use of this model has satisfactory found satisfactory based on R test. Other models such results compared to SCS-CN model. Similarly, statis- as LR (model rank = 3), TR (model rank = 4), ICAR tical results for JE model showed second ranking. It (model rank= 5) and DII (model rank= 6) have lowest can be used successfully for the prediction of surface value with respect to SCS-CN model. Hence, these GEOLOGY, ECOLOGY, AND LANDSCAPES 105 Figure 7. (a). Daily rainfall- runoff time series at study area from 2002–2014 (monsoon season). (b). Monthly rainfall-runoff time series at study area from 2002–2014 (monsoon season). (c). Yearly rainfall-runoff time series at study area from 2002–2014. models have limited potential to estimate surface run- LULC, soil hydrological group, time interval of pre- off. Significant difference was found among model cipitation and physical characteristics of agricultural with respect to SCS-CN (except IDS). Ghazavi and watershed. Abasali (2003) did not consider Coutagine method and corrected Langbin method, as suitable method in 5. Conclusion arid regions. Khosroshahi (1991) has mentioned that the estimation by ICAR is more than observed value; it Runoff estimation of ungauged watershed is a chal- is more obvious in the agricultural watersheds of more lenge for hydrologists. Discharge value of ungauged than 200 km . Also, Fathzadeh (2008) considered clas- catchments is important for hydrological planning sic Coutagine and Turc approaches as non-suitable and designing of various hydraulic structures. Precise methods with significant errors. In this research, knowledge about the runoff will help in better man- according to the results of statistical tests JE method agement of water resources of the local region. It is was introduced as the best method after Inglis & De difficult to estimate the runoff more accurately from Souza (IDS) method, for runoff estimation in the ungauged watershed with coarse resolution satellite study area. The advantage of IDS model is as simple data due to high uncertainty. SCS-CN model requires and not affected by any factors related to slope, type of input of LULC, soil data, and rainfallthat can be 106 K. S. RAWAT ET AL. Table 5. Runoff estimated from different empirical model. Runoff (in cm) from models A.T A. R S. No Years (°C) (cm) SCS- CN KH IDS DII TR CR ICAR JE LR 1 2003 24.79 93.89 22.58 81.97 28.13 35.73 11.26 2064 10.37 23.11 14.62 2 2004 24.71 56.29 9.04 44.41 8.53 18.83 4.34 744 4.99 8.32 5.61 3 2005 25.80 84.77 31.86 72.36 22.35 31.50 8.87 1629 8.49 18.40 12.10 4 2006 25.93 65.27 16.99 52.8 12.2 22.73 5.35 962 5.79 10.88 7.42 5 2007 24.91 72.32 20.6 60.34 15.52 25.85 6.96 1220 7.08 13.67 9.00 6 2008 24.90 107.64 36.43 95.66 38.07 42.22 14.21 2703 12.56 30.29 18.79 7 2009 25.94 57.20 14.51 44.72 8.87 19.22 4.1 738 4.78 8.35 5.78 8 2010 26.14 59.77 10.87 47.2 9.88 20.33 4.42 801 5.04 9.08 6.28 9 2011 25.44 112.10 40.61 99.86 41.62 44.36 14.82 2881 12.93 32.44 20.23 10 2012 25.13 89.71 24.5 77.62 25.4 33.78 10.2 1863 9.54 20.93 13.44 11 2013 24.34 146.08 51.63 134.38 73.78 61.02 24.29 5073 20.10 56.54 32.58 12 2014 24.79 72.60 20.42 60.4 15.66 25.98 6.82 1211 6.95 13.63 9.06 Annual Temperature (in °C) from MODIS, A.T; Annual Rainfall (in cm) from NOAA, A.R; KH, Khosla; IDS, Inglis & De Souza; DII, Department of irrigation, India; TR, Turc relationship; CR, Coutagine relationship; ICAR, Indian Council of Agricultural Research; JE, Justin Equation; LR, Lacey Relationship; M, Model. Table 6. Statistical test values for different empirical model with respect to SCS-CN. S. No Test IDS JE LR TR ICAR DII KH CR 1 R 0.89 0.89 0.90 0.90 0.90 0.92 0.92 0.88 2 RMSE 7.75 6.43 13.32 16.95 17.99 7.60 49.83 2158.82 3 R-RMSE 0.24 0.27 0.50 0.62 0.63 0.49 2.30 70.88 4 MAE 0.00 4.53 12.09 15.37 15.95 6.79 47.64 1799.10 5 NRMSE 0.31 0.26 0.53 0.68 0.72 0.30 1.99 86.34 6 MBE 0.00 4.53 12.09 15.37 15.95 −6.79 −47.64 −1799.10 7 PBIAS 0.01 22.15 93.69 159.45 176.26 −21.36 −65.58 −98.63 8 RMSE% 2.58 2.14 4.44 5.65 6.00 2.53 16.61 719.52 9 BIAS 0.07 0.21 0.49 0.62 0.62 −0.38 −2.17 −69.33 10 K factor 0.00 0.07 0.13 0.16 0.16 0.23 0.61 0.88 11 Rating 0.01 0.55 1.06 1.29 1.32 1.88 4.91 7.04 12 MR 1 2 3 4 5 6 7 8 Note:- Coefficient of Determination, R ; Root Mean Square Error, RMSE; Relative Root Mean Square Error, R-RMSE; Mean Absolute Error, MAE; Normalized root mean square error, NRMSE; Mean bias error, MBE; RMSE%, Percentage RMSE; bias, BIAS; Mean difference bias, Model Rank, MR. obtained from satellite hence it easily provides the ORCID runoff estimation at macro level. Still there is need of Sudhir Kumar Singh http://orcid.org/0000-0001-8465- some other simple alternate model for estimating annual runoff from ungauged agricultural watershed. That can provide runoff estimate close to SCS-CN model. In this context present study reveals that References Inglis & De Souza (IDS) model is a simple and good Adham, M. I., Shirazi, S. M., Othman, F., Rahman, S., alternative of SCS-CN model. It can serve the purpose Yusop, Z., & Ismail, Z. (2014). Runoff potentiality of a of runoff estimation from ungauged watershed. IDS watershed through SCS and functional data analysis tech- model required input of annual rainfall data. This can nique. The Scientific World Journal, 2014, 1–15. https:// be generated from automatic weather station or satel- doi.org/10.1155/2014/379763 lite-based freely available data. The major drawback of Alizade, A. (1989). Principle of Applied Hydrology. Mashhad, Ghods Boniad Press.860pp all the empirical models except SCS-CN is estimation Balázs, B., Bíró, T., Dyke, G., Singh, S. K., & Szabó, S. (2018). of runoff on annual basis. Still these models provide Extracting water-related features using reflectance data reliable information about the runoff. This informa- and principal component analysis of Landsat images. tion can be utilized by the planners and policy makers Hydrological Sciences Journal, 63(2), 269–284. https:// for management and designing purposes. doi.org/10.1080/02626667.2018.1425802 Bavishi, H., & Bhagat, N. K. (2017). Rainfall Runoff Co- Relationship using Empirical Methods for Lower Mahi Basin, India. International Journal of Civil Engineering Acknowledgments Technology (IJCIET), 8(3), 575–581 Bérod, D. D., Singh, V. P., & Musy, A. (1999). A geomor- The authors are grateful to Dr. Anil Kumar Mishra phologic kinematic-wave (GKW) model for estimation of (Principal Scientist, Water Technology Centre, IARI, New floods from small alpine watersheds. Hydrological Delhi) for his critical input and suggestions on the Processes, 13(9), 1391–1416. https://doi.org/10.1002/ manuscript. (SICI)1099-1085(19990630)13:9<1391::AID-HYP809>3. 0.CO;2-B Castiglioni, S., Lombardi, L., Toth, E., Castellarin, A., & Disclosure statement Montanari, A. (2010). Calibration of rainfall-runoff mod- els in ungauged basins: A regional maximum likelihood No potential conflict of interest was reported by the authors. GEOLOGY, ECOLOGY, AND LANDSCAPES 107 approach. Advances in Water Resources, 33(10), 1235– India.Modeling Earth Systems and Environment, 4(1), 1242. https://doi.org/10.1016/j.advwatres.2010.04.009 295–310. Centeri, C., Szalai, Z., Jakab, G., Barta, K., Farsang, A., Lamb, R. (1999).Calibration of a conceptual rainfall-runoff Szabó, S., & Biró, Z. (2015). Soil erodibility calculations model for flood frequency estimation by continuous based on different particle size distribution measure- simulation. Water Resources Research, 35(10), 3103– ments. Hungarian Geographical Bulletin, 64(1), 17–23. 3114. https://doi.org/10.1029/1999WR900119 https://doi.org/10.15201/hungeobull.64.1.2 Lu, D., Mausel, P., Batistella, M., & Moran, E. (2004). Chow, V. T., Maidment, D. R., & Mays, L. W. (1988). Comparison of land-cover classification methods in the Applied hydrology. McGraw-Hill, Inc. Brazilian Amazon Basin. Photogrammetric Engineering & Devia, G. K., Ganasri, B. P., & Dwarakish, G. S. (2015). A Remote Sensing, 70(6), 723–731. https://doi.org/10.14358/ review on hydrological models. Aquatic Procedia, 4, PERS.70.6.723 1001–1007 doi:10.1016/j.aqpro.2015.02.126 Maliqi, E., & Singh, S. K. (2019). Quantitative Estimation of Dorum, A., Yarar, A., Sevimli, M. F., & Onüçyildiz, M. Soil Erosion Using Open-Access Earth Observation Data (2010). Modelling the rainfall–runoff data of susurluk Sets and Erosion Potential Model. Water Conservation basin. Expert Systems with Applications, 37(9), 6587– Science and Engineering, 4(4), 187-200. Fathzadeh, A. 6593. https://doi.org/10.1016/j.eswa.2010.02.127 (2008). Comparison of Kotain and turc method with its Ghazavi, G. H., & Abasali, V. (2003). Determination of best conversion methods in dry land (M.Sc Thesis). Tarbiat estimation annual runoff empirical method no station Modarres University, College of Natural Resources and watershed in semi-arid land in Darab Province. Journal Marine Sciences. of Natural Resources and Agriculture Science, 10(3), 10– Mohamadi, M. A., & Kavian, A. (2015). Effects of rainfall 25. patterns on runoff and soil erosion in field plots. Golshan, M., & Ebrahimi, P. (2014). Estimation of the International Soil and Water Conservation Research, 3 Runoff by Empirical Equations in Dry and Mid-Dry (4), 273–281. https://doi.org/10.1016/j.iswcr.2015.10.001 Mountainous Area without Stations (Case Study: Moore, R. J., & Clarke, R. T. (1981). A distribution function Madan Watershed, Qazvin Province-Iran). Bulletin of approach to rainfall runoff modeling. Water Resources Environment, Pharmacology and Life Sciences, 3, 77– Research, 17(5), 1367–1382. https://doi.org/10.1029/ 85 WR017i005p01367 Hosking, J. R. M., & Clarke, R. T. (1990). Rainfall-runoff Mosaffaei, Z., Jahani, A., Chahouki, M. A. Z., Goshtasb, H., relations derived from the probability theory of storage. Etemad, V., & Saffariha, M. (2020). Soil texture and plant Water Resources Research, 26(7), 1455–1463. https://doi. degradation predictive model (STPDPM) in national org/10.1029/WR026i007p01455 parks using artificial neural network (ANN).Modeling Iowa Storm water Management 2C-5 Manual (2C-5 NRCS Earth Systems and Environment, 6,–715-729. TR-55 Methodology). (2008). www.ctre.iastate.edu/pubs/ Mutreja, K. N. (1986). Applied hydrology. Tata McGraw- ... /2C-5NRCSTR-55Methodology.pdf Hill. Tata McGraw Hill, New Delhi Jahani, A., Feghhi, J., Makhdoum, M. F., & Omid, M. (2016). Pandey, A., & Dabral, P. P. (2004). Estimation of runoff for Optimized forest degradation model (OFDM): An envir- hilly catchment using satellite data. Journal of the Indian onmental decision support system for environmental Society of Remote Sensing, 32(2), 236–240. https://doi.org/ impact assessment using an artificial neural network. 10.1007/BF03030880 Journal of Environmental Planning and Management, 59 Pásztor, L., Waltner, I., Centeri, C., Belényesi, M., & Takács, (2), 222–244. https://doi.org/10.1080/09640568.2015. K. (2016). Soil erosion of Hungary assessed by spatially 1005732 explicit modelling. Journal of Maps, 12(sup1), 407–414. Jakab, G., Nemeth, T., Csepinszky, B., Madarász, B., Szalai, https://doi.org/10.1080/17445647.2016.1233913 Z., & Kertész, Á. (2013). The influence of short term soil Praveen Kumar, B.J., Pradeep, H., Lokesh, A., Akarshraj, K. sealing and crusting on hydrology and erosion at balaton H., Surendra, H. J., & Avinash, S.D. (2016). Estimation of uplands, Hungary. Carpathian Journal of Earth and Runoff using Empirical Equations and Fuzzy Logic Environmental Sciences, 8(1), 147–155. http://real.mtak. method: A case study. International Journal of Scientific hu/id/eprint/4000. & Engineering Research. 7, (5): 28–37 Kaleris, V., Kourakos, V., & Langousis, A. (2015). Rahmati, O., Panahi, M., Ghiasi, S. S., Deo, R. C., Calibration of rainfall-runoff models: The effect of the Tiefenbacher, J. P., Pradhan, B., Goshtasb, H., temporal distribution of rainfall on uncertainties in Kornejady, A., Shahabi, H., Shirzadi, A., Khosravi, H., model parameter estimation. Geophysical Research Moghaddam, D. D., Mohtashamian, M., Tien Bui, D., & abstracts, 17, EGU2015–13543. https://meetingorgani- Jahani, A. (2020). Hybridized neural fuzzy ensembles for zer.copernicus.org/EGU2015/EGU2015-13543.pdf dust source modeling and prediction. Atmospheric Khosravi, K., Mirzai, H., & Saleh, I. (2013). Assessment of Environment, 224, 117320. https://doi.org/10.1016/j.atmo empirical methods of runoff estimation by statistical test senv.2020.117320 (case study: BanadakSadat Watershed, Yazd Province). Rawat, K. S., & Singh, S. K. (2017). Surface runoff estimation International Journal of Advanced Biological and from semi-arid ungauged agricultural watershed using Biomedical Research, 1(3), 285–301. http://www.ijabbr. SCS-CN and Earth Observation Data Sets. Water com/article_6897.html. Science and Engineering.1: 233–247 https://doi.org/10. Khosroshahi, M. (1991). Water balance calculation in small 1007/s41101-017-0016-4 watershed areas and no hydrometric stations using Rawat, K. S., & Singh, S. K. (2018). Water Quality Indices empirical formulas in the Khorasan region (M.Sc and GIS-based evaluation of a decadal groundwater qual- Thesis). Department of Natural Resources, Tehran ity. Geology, Ecology, and Landscapes, 2(4), 240–255 University. doi:10.1080/24749508.2018.1452462 Kumar, N., Singh, S. K. Singh, V.G., & Dzwairo, B.,(2018). Rawat, K. S., Singh, S. K., Ray, R. L., Szabó, S., & Kumar, S. Investigation of impacts of land use/land cover change on (2020). Parameterizing the modified water cloud model water availability of Tons River Basin, Madhya Pradesh, to improve soil moisture data retrieval using vegetation 108 K. S. RAWAT ET AL. models. Hungarian Geographical Bulletin, 69(1), 17–26. Tedela, N. H., McCutcheon, S. C., Rasmussen, T. C., Hawkins, https://doi.org/10.15201/hungeobull.69.1.2 R. H., Swank, W. T., Campbell, J. L., . . . Tollner, E. W. (2012). Rawat, K. S., Singh, S. K., Singh, M. I., & Garg, B. L. (2019). Runoff Curve Numbers for 10 small forested watersheds in Comparative evaluation of vertical accuracy of elevated the mountains of the Eastern United States. Journal of points with ground control points from ASTERDEM and Hydrologic Engineering, 17(11), 1188–1198. https://doi.org/ SRTMDEM with respect to CARTOSAT-1DEM. Remote 10.1061/(ASCE)HE.1943-5584.0000436 Sensing Applications, Society and Environment, 13, 289– Thakur,, J. K., Singh,, S. K., & Ekanthalu, V. S., (2017). 297. https://doi.org/10.1016/j.rsase.2018.11.005 . Integrating remote sensing, geographic information systems Sadati, S. K. F., Heydari, M., Javaheri, N., & Othman, F. and global positioning system techniques with hydrological (2014). Water yield estimation in polrudwatershed based modeling. Applied Water Science, 7(4), 1595–1608. on empirical methods and modelling in geographic infor- USDA, S. (1986). Urban hydrology for small watersheds TR- mation system (GIS). Journal of River Engineering, 2(7), 55. Technical Release. United States Department for 1–9.https://europub.co.uk/articles/32988/view Agriculture Soil Conservation Service, Washington DC Singh, S. K., Srivastava, P. K., Szabó, S., Petropoulos, G. P., USDA-SCS (1967). United states department of agriculture Gupta, M., & Islam, T. (2017). Landscape transform and soil conservation service, engineering division Irrigation spatial metrics for mapping spatiotemporal land cover Water Requirements, Technical Release, 21, pp. 1–98 dynamics using Earth Observation data-sets. Geocarto Vaze, J., Jordan, P., Beecham, R., Frost, A., & Summerell, G. 2011. International, 32(2), 113–127.https://doi.org/10.1080/ Guidelines for rainfall-runoff modelling: Towards best practice 10106049.2015.1130084. model application. eWater Cooprative Research Centre. Singh, V. K., Kumar, D., Kashyap, P. S., Singh, P. K., Kumar, Waltner, I., Pásztor, L., Centeri, C., Takács, K., Pirkó, B., A., & Singh, S. K. (2020). Modelling of soil permeability Koós, S., & László, P. (2018). Evaluating the new soil using different data driven algorithms based on physical erosion map of Hungary—A semiquantitative properties of soil. Journal of Hydrology, 580, 124223. approach. Land Degradation & Development, 29(4), https://doi.org/10.1016/j.jhydrol.2019.124223 1295–1302.https://doi.org/10.1002/ldr.2916. Szabó, J. A., Jakab, G. I., & Szabó, B. (2015). Spatial and Wheater, H. S., Sorooshian, S., & Sharma, K. (2008). temporal heterogeneity of runoff and soil loss dynamics Modelling hydrological processes in arid and semi-arid under simulated rainfall. Hungarian Geographical areas: an introduction to the workshop. Hydrological Bulletin, 64(1), 25–34. https://doi.org/10.15201/hungeo Modelling in Arid and Semi-Arid Areas.(Cambridge bull.64.1.3 University Press) Cambridge Tadesse, L., Suryabhagavan, K. V., Sridhar, G., & Legesse, G. Zhang, Z., Sheng, L., Yang, J., Chen, X. A., Kong, L., & (2017). Land use and land cover changes and soil erosion Wagan, B. (2015). Effects of land use and slope gradient in Yezat Watershed, North Western Ethiopia. on soil erosion in a red soil hilly watershed of southern International Soil and Water Conservation Research, 5 China. Sustainability, 7(10), 14309–14325. https://doi. (2), 85–94. https://doi.org/10.1016/j.iswcr.2017.05.004 org/10.3390/su71014309 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geology Ecology and Landscapes Taylor & Francis

Comparative evaluation of models to estimate direct runoff volume from an agricultural watershed

Loading next page...
 
/lp/taylor-francis/comparative-evaluation-of-models-to-estimate-direct-runoff-volume-from-1lRyMeMRCW

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
© 2020 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
10.1080/24749508.2020.1833629
Publisher site
See Article on Publisher Site

Abstract

GEOLOGY, ECOLOGY, AND LANDSCAPES 2021, VOL. 5, NO. 2, 94–108 INWASCON https://doi.org/10.1080/24749508.2020.1833629 RESEARCH ARTICLE Comparative evaluation of models to estimate direct runoff volume from an agricultural watershed a b c Kishan Singh Rawat , Sudhir Kumar Singh and Szabo Szilard a b Geo-Informatics, Civil Engineering Department, Graphic Era (Deemed to Be University), Dehradun, India; K. Banerjee Centre of Atmospheric Ocean Studies, IIDS, Nehru Science Centre, University of Allahabad, Allahabad, India; Professor, Head of Department, Department of Physical Geography and Geoinformation Systems, University of Debrecen, Debrecen, Hungary ABSTRACT ARTICLE HISTORY Received 20 April 2020 Generally, runoff records are the most important input data in water resource management; Accepted 4 October 2020 however, their availability is very limited especially in developing country as compared to rainfall records, especially under medium and small-scale catchments. In our study, we esti- KEYWORDS mated runoff from ungauged agricultural watershed with the curve number method and Curve number; runoff; RMSE; empirical mathematical models were compared with SCS-CN. Empirical mathematical models soil erosion; ungauged (Inglis and De Souza Formula (IDS), Turc relationship (TR), Indian Irrigation Department (DII) watershed model, Coutagine relationship (CR), Khosla method (KH), Justin Equation (JE), Lacey relation- ship (LR), and Indian Council of Agricultural Research (ICAR)) model were used to estimate annual runoff (in cm). It was found that IDS model has capability to simulate annual runoff as very close to Soil Conservation Service Curve Number (SCS-CN) model and has lowest Root Mean Square Error (RMSE) value as 7.75, and ranking of this model (based on K factor (value of st 0.001) analysis) was topmost (or 1 ) in comparison to other eight models. This study suggests that empirical mathematical model has potential for annual runoff estimation from ungauged watershed. 1. Introduction (2015), a hydrological model is a simplification of a real-world system, used mainly for the prediction of Soil erosion is one the most important process that hydrological processes based on rainfall, drainage area endangers the soil quality and, therefore, the agricul- (topography), soil properties, vegetation cover, and tural production (Pásztor et al., 2016; Waltner et al., runoff model. It is defined as a set of equations that 2018). Runoff has strong relationship with the rate of enable the estimation of runoff as a function of various erosion, as it is the consequence of the precipitation parameters used for describing watershed (duration, & intensity; Mohamadi & Kavian, 2015), characteristics. the slope characteristics (steepness, length, & shape), The Universal Soil Loss Equation (USLE) is an soil characteristics (infiltration capacity, depth of empirical equation. The Revised Universal Soil Loss humus layer, particle size& initial water content; Equation (RUSLE) is a modification of USLE, espe- Centeri et al., 2015; Szabó et al., 2015), and the vegeta- cially for more complex situations of rill and inter-rill tion/land use (management, density, leaf-area index, erosion in conservation planning and land uses. Both arboreal, or herbaceous; Jakab et al., 2013; Tadesse et erosion-prone models calculate detachment capacity al., 2017). and soil loss. RUSLE model predicts soil degradation Simple empirical equation relates catchment char- and sediment concentrations better using another soil acteristics and complicated physical models are avail- erodibility factor (F-soil factor, based on soil texture). able to estimate the catchment runoff . The application The soil conservation service-curve number (SCS-CN) of conceptual hydrological model to generate runoff method has been used widely (Bérod et al., 1999; from ungauged watershed with limited data have been Pandey & Dabral, 2004; Vaze et al., 2011). The SCS- studied by researchers in past (Kaleris et al., 2015). CN method is simple, predictable, stable, and relies on Regional scale model can explain the variation of the only one parameter, namely the CN. The land use/ model parameters with physiographic factors. These land cover (LULC) class can be integrated with the models did not fully capture the local scale process and hydrologic soil groups (HSG) of the sub basin in GIS, variations. However, the certainty of the calibrated and the weighted CN can be estimated. These esti- model parameters is high enough to simulate the mated weighted CN for the entire area can be used to hydrologic response of ungauged watershed . compute runoff. Moore and Clarke (1981) showed According to Wheater et al. (2008) and Devia et al. CONTACT Sudhir Kumar Singh sudhirinjnu@gmail.com K. Banerjee Centre of Atmospheric Ocean Studies, IIDS, Nehru Science Centre, University of Allahabad, Allahabad, U.P, India © 2020 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. GEOLOGY, ECOLOGY, AND LANDSCAPES 95 that a variety of distributions that can be easily incor- to SCS-CN for generating annual runoff of ungauged porated into this type of model structure and they Jhagrabaria watershed using satellite data. derive analytical equations for the response of differ - ent distributions. Hosking and Clarke (1990) extended 2. Study area and data the work of Moore and Clarke (1981), and reported that the model can be used to derive a relationship The Jhagrabaria watershed is located in the Allahabad between the frequencies of storm rainfall and flow district of state Uttar Pradesh, India (Figure 1). peak magnitude in an analytical form. The UK Geologically the area consists of Upper Vindhayan Institute of Hydrology has shown the model for long formation consist of mainly sandstone and shale. The runs to derive flood frequencies (Dorum et al., 2010; elevation is ranging from 85 to 192 m above mean sea Lamb, 1999). Recently many studies have applied level with nearly flat to gently undulated topography machine learning and soft computing approaches to and small occasional hillocks. The upland area is cov- study the soil properties and erosion (Jahani et al., ered with loam, except in the south-western part of 2016; Singh et al., 2020; Mosaffaei et al., 2020; tehsil Karchhana, where the soil is a mixture of clay Rahmati et al., 2020). and marsh. The area has semi-tropical climate as The earth observation datasets integration within summer and the winter. The area receives about 91% Geographical Information System (GIS) make of the total annual rainfall due to southwest monsoon watershed modeling easy and accurate (Balázs et al. from June to September. The relative humidity is high 2018). The capabilities of these technologies have been during the monsoon month (Rawat & Singh, 2017). successfully utilized by many researchers in rainfall- runoff modeling . Earth-observing satellite provides more reliable input parameters for hydrological mod- 3. Materials and methods eling (Rawat & Singh, 2017; Maliqi & Singh, 2019). 3.1. Datasets used However, GIS processing has become a critical step in hydrologic modeling (Thakur et al., 2017), since it LANDSAT7 ETM+ (path/row: 231/67) was acquired contributes to generate model parameter distribution on June 27, 2006 (Table 1). The image was converted in spatial manner. to apparent reflectance through an image-based calibra- Although researchers delineated several models, the tion method. Atmospheric correction was performed satellite-based inputs in these models were not com- using Fast Line-of-sight Atmospheric Analysis of paratively used and limited model performance was Hypercubes (FLAASH) algorithm. Image was geome- evaluated. Hence, objectives of work are as follows: (i) trically rectified using ground control points collected to estimate daily runoff using SCS-CN and (ii) to find from Survey of India (SOI) topographic sheets using an optimal empirical mathematical model with respect nearest-neighborhood resampling technique and a Figure 1. Location of the study area (Jhagrabaria agriculture watershed), U.P., India. 96 K. S. RAWAT ET AL. Table 1. Specifications of LANDSAT (TM and ETM ) sensors used in the present study. Spatial resolution (meters) Spectral resolution (µm) Spectral bands TM ETM+ TM ETM+ 1 (Blue) 30 30 0.45–0.52 0.45–0.52 2 (Green) 30 30 0.52–0.60 0.53–0.61 3 (Red) 30 30 0.63–0.69 0.63–0.69 4 (Near IR) 30 30 0.76–0.90 0.78–0.90 5 Shortwave Infrared (SWIR) 1 30 30 1.55–1.75 1.55–1.75 6 (Thermal IR)* 120* (30) 60 * (30) 10.4–12.5 10.4–12.5 7 Shortwave Infrared (SWIR) 2 30 30 2.08–2.35 2.09–2.35 8 (Panchromatic)** 15 0.52–0.90 * TM Band 6 was acquired at 120-meter resolution, but products are resampled to 30-meter pixels. * ETM+ Band 6 is acquired at 60-meter resolution, but products are resampled to 30-meter pixels. root-mean-square error with less than one pixel was clipped with the administrative boundary of obtained during the geometric rectification. Land use/ Allahabad district’s study area, LST data were land cover (LULC), viz. barren land, fallow land, vege- extracted. tation, and water bodies/wetlands were identified in the The soil map of the Shankargarh block was col- field and their coordinates were recorded with a hand- lected from Soil Survey Department, Allahabad, U.P., held global positioning system (GPS) device (Garmin India. The map was scanned and then registered with eTrexH). The maximum likelihood classifier is a simple the help of geo-referenced Survey of India (SOI) topo- and easy to use classification algorithm, in which a pixel graphical sheet no. 63 G/11 and 63 G/12, respectively. with the maximum likelihood is classified into the The registered soil maps were digitized and different corresponding class (Singh et al., 2017; Lu et al., soil attributes were assigned to the different soil 2004). Afterward window 3 × 3 size majority filter was groups in digital format. In present study, CN map is applied to remove the “salt and pepper” noise from generated with help of LULC and HSG map, CNII is classified image. the CN for normal condition, CNI is the CN for dry The need of satellite-estimated precipita - condition, CNIII is the CN for wet condition and CN tion arises because of the non-availability or is assigned based on Section 2C-5 – Iowa Storm water poorly distributed ground rainfall data. For Management 2C-5 Manual (2C-5 NRCS TR-55 the work, the daily precipitation data were Methodology) (2008). downloaded from ftp://ftpprd.ncep.noaa. gov/pub/cpc/fews/S.Asia/. Resolution of 3.2. LST role in models rainfall estimates are of 0.1 × 0.1 degree and inputs include Global LST data sets are important because five models (TR, Telecommunication System (GTS) station CR, KH, JE, and ICAR) out of eight (KH, IDS, DII, TR, data, as well as GOES Precipitation Index CR, ICAR, JE, and LR) models required LST as input (GPI) infrared cloud top temperature fields data, to predict runoff. Average function was applied derived from Meteosat and polar-orbiting to calculate monthly and annual LST. In TR tempera- satellite precipitation estimate data from ture is part of denominator, and it is also under square Special Sensor Microwave/Imager (SSM/I) root function therefore its effective yield will be small, on board Defense Meteorological Satellite over all denominator will be a small quantity which Program and Advanced Microwave gives a little fraction of annual rainfall, net result will Sounding Unit (AMSU-B) on board come as high runoff from TR. In CR temperature is NOAA15, 16 and 17. also part of denominator, and it does not has any Land surface temperature (LST) is an important constrain (like square root function); thus, a good parameter in study of water resources. Data available yield will apply in denominator which gives small over tile (1100 km x 1100 km) of 2003–2014; the total fraction of annual rainfall, resulting in the overestima- 495 (per year 45) files, emissivity and quality control tion of runoff. KH model reveals a low annual runoff (QC) files were downloaded from http://glovis.usgs. because a major part is subtracted from annual rainfall gov. LST values were retrieved based on the Split (T/3.74 (in °C)), and will be a big quantity). From JE Window algorithm. Hierarchical Data Format (HDF) model, in denominator temperature has multiple fac- files of 8-day LST were stacked of each year and study tor of 1.8 additional 32 which will give large number at area was subset from tiles and from converted sinu- dominator; therefore, a small yield in JE; Thus, this soidal (SIN) to Universal Transverse Mercator (UTM, model have better result from other models being WGS84) projection in EVNI software. Average 8 days temperature-dependent (TR, CR, KH, and ICAR). images (spatial and temporal) were used to generate ICAR model reveals that temperature is part of monthly LST. The 8-days India’s LST data for 12 years denominator and it is also multiplied by another factor period (2003–2014) were downloaded and later it were which gives a big yield in denominator; therefore, a GEOLOGY, ECOLOGY, AND LANDSCAPES 97 less net annual runoff from ICAR. ICAR may be good overestimation or underestimation of the statistics. for a regional area because it is directly dependent on Mean Absolute Error (MAE, Rawat et al., 2020) is a area, slope and other factors that dominate at regional measure of how models are varied from SCS-CN. scale. Hence, annual runoff fluctuates if annual mean MAE is a more natural measure of average error and surface temperature slightly varies because all equa- is unambiguous. Percentage Bias (PBIAS; error index tions are directly linked to surface temperature. for model, Rawat et al., 2019) measures the average tendency of the simulated data to be larger or smaller than their observed from SCS-CN model. Mean 3.3. Runoff estimation Difference (BIAS, Rawat et al., 2020) is difference between model’s value and value from SCS-CN, if The SCS-CN method was developed to estimate sur- difference is zero, it is called unbiased otherwise face runoff from small agricultural watersheds biased. A low Mean Bias Error (MBE, Rawat et al., (USDA-SCS, 1967). The soils have been classified 2020) is desired; ideally a zero value of MBE should be into four hydrologic groups namely A, B, C, and D obtained. A positive value of MBE shows an over (USDA, 1986), based on infiltration, soil classification, estimate with respect to SCS-CN surface runoff and other criteria (soil’s surface condition (infiltration whereas a negative value show an under-estimate rate) and its horizon (transmission rate). Land use and with respect to SCS-CN surface runoff. management types have been used in the preparation of hydrological soil-cover complex, which has been utilized in estimating direct runoff. Antecedent 3.5 Ranking of empirical mathematical models Moisture Condition (AMC) is an indicator of Factor K was estimated to provide proper weight watershed wetness and availability of soil moisture (Rawat and Singh, 2018) to selected statistical index storage prior to a storm (Rawat & Singh, 2017). SCS (all used statistical test) as: has developed a guide for adjusting CN according to " , # AMC based on total rainfall in the 5 day period pre- � � ceding a storm. Three levels of AMC as: AMC-I (dry), K ¼ 1 (10) AMC-II (normal), and AMC-III (wet) conditions. The n¼1 seasonal rainfall limits for these three antecedent W ¼ and moisture conditions (Table 2). Many hydrologists have discussed relationships of precipitation and 1 ¼ W annual surface runoff with the assumption that physi- cal characteristics of the watershed are constant (Castiglioni et al., 2010). The brief information about n Rating ¼ ðW i Þ (11) the empirical models applied in this study is presented n n n¼1 in Tables 3–4. th where, K is factor, i is n statistical index and W is weight for particular statistical index. Lowest rating 3.4 Performance evaluation model will be on first rank and vice-versa. Model's performance was evaluated using statistical parameters. The reference values were taken of SCS- 4. Results and discussions CN model. Coefficient of Determination (R , Rawat et al., 2020) describes the dispersion of models vs. SCS- 4.1 Land use/land cover (LULC) CN model. The Root Mean Square Error (RMSE, LULC affects the infiltration, erosion, and evapo- Rawat et al., 2020) values define how models over- transpiration hence, it is an important character- estimate or underestimate the measurements with istic of runoff process. Overall 90% accuracy of respect to SCS-CN model. Relative Root Mean classified LULC map was achieved. The area of Square Error (R-RMSE; Rawat et al., 2020) is a stan- 2 2 barren land (36.91 km ), fallow land (36.62 km ), dardization of RMSE. R-RMSE value is expressed in and vegetation (74.71 km ) (Figure 2). The fallow percent and represents the standard variation of the and barren land together have the highest area as model. The R-RMSE assigns equal weight to any 48.99% whereas vegetation area is 47.81%. The area exposed for erosion offer high rate of water Table 2. Classification of antecedent moisture conditions erosion. Several studies have demonstrated the (AMC). role of LULC in hydrologic modeling and runoff Total 5 days Antecedent Rainfall (mm) estimation (Adham et al., 2014; Tedela et al., 2012; AMC Dormant Season Growing Season Kumar et al., 2018). I < 12.7 < 35.6 II 12.7 − 27.9 35.6–53.3 III > 27.9 > 53.3 98 K. S. RAWAT ET AL. Table 3. Specification of empirical mathematical models (EMM), the purpose, mathematical expression and references. Sr. no. EMM Purpose/reason Mathematical Expression Reference 1 SCS-CN method Developed to estimate surface runoff from small agricultural watersheds ðP 0:2SÞ SCS-CN model Q ¼ ðP< 0:2SÞ (1) Pþ0:8S in USDA, Q = is direct runoff (mm), P total precipitation (mm), S is watershed storage (1967) ðP 17:8Þ� P 2 Inglis and De Souza Plains of Maharashtra region of India Mutereja, R ¼ (2) (IDS) (1986) Where, P is annual precipitation (cm), and R is annual runoff (cm) Bavishi and Bhagat (2017) 0:86 3 Indian Irrigation Indian Irrigation Department uses the relationship equation between Rainfall and Praveen Kumar R ¼ P 1:17� P (3) Department (DII) Runoff et al. (2016) Where, P is annual precipitation (cm), and R is annual runoff (cm) rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4 Turc relationship (TR) Relationship for watersheds with the area less than 300 km based on achieved results Khosravi et al. � � R ¼ P D (4) D ¼ P LT ¼ 300þ 25� T þ 0:05� T Where, P is annual from doing a study on 254 watersheds in various climatic and weather conditions (2013) 0:9þ = LT precipitation (cm), R is annual runoff (cm), T is mean annual temperature (°C) and D is annual flow shortage. 5 Coutagine relationship Presented a general relationship Alizade, (1989) R ¼ P λ� P (5) (CR) Sadati et al. Where, P is annual precipitation (cm), R is annual runoff (cm), is coefficient and recommended 1 (2014) as:-λ ¼ = Where, T is mean annual temperature (°C). ð0:8þ 0:14� TÞ 6 Khosla method (KH) Amount of mean annual runoff Golshan and R ¼ P = (6) 3:74 Ebrahimi Where, P is annual precipitation (cm), R is annual runoff (cm) and T is mean annual temperature (° (2014) C). Khosravi et al. (2013) 0:155 2 H H 0:284� S � P max min pffiffi 7 Justin Equation (JE) Estimate runoff with considering parameters such as mean temperature, slope and R ¼ , s ¼ (7) Golshan and 1:8� Tþ32 precipitation in watershed Where, R is annual runoff (cm), P is mean annual precipitation (cm), T is Mean annual temperature Ebrahimi (°C), S is mean slope of watershed, H is elevations of watershed, A is the watershed area (2014) 1:44 1:115� P 8 Indian Council of Based on 17 watershed annual runoff in Nilgiri region that was conducted by Indian R ¼ (8) 1:34 0:0613 T � A Agricultural Council of Agricultural Research Sadati et al. Where, P is annual precipitation (cm), R is annual runoff (cm), A is the watershed area, T is mean Research (ICAR) (2014) annual temperature (°C). 9 Lacey relationship Lacey an Indian scientist investigated several indian watersheds to prepare a Lacey Bavishi and R ¼ (9) 304:8ðF Þ 1þ (LR) equation to estimate annual Bhagat R is mean annual runoff (cm), P is Mean annual precipitation (cm), F is Parameter of rainfall Runoff (2017) duration and physiographic properties, Values of Fz coefficient are given in Table 4. Golshan and Ebrahimi (2014) GEOLOGY, ECOLOGY, AND LANDSCAPES 99 Table 4. Values of F coefficient. observed. This low rainfall in those years leads severe Duration of rainfall to moderate drought. After year 2004 each year one or Catchment area Long Average Short more than one rainfall event (~80 mm) exists except Includes shelf, flat plains with deep soils and 6 4 2 year 2008 to 2011. It reveals that maximum rainfall was vegetation appropriate recorded 118.81 and 120.67 mm, respectively, during Somewhat flattened with deep soils and 2.5 1.67 0.83 pasture vegetation. 05-Sept-2007 and 05-October-2013. Figure 4(d) Relatively high hills with shallow soils and 1.5 1 0.5 (monthly rainfall) showed rainy and non-rainy months vegetation is relatively weak Sand, gravel and steep terrain with plenty of 0.88 0.58 0.23 (2003–2014), it helps to understand shifting of rainfall height pattern. Such types of patterns are absent during 2003 High and steep rocky terrain with no 0.43 0.28 0.14 to 2014, however, some non-rainy months with low vegetation rainfall amount (April-06 (20.28 mm), May-09 4.2. Soil map and hydrological soil group (20.54 mm) and May-011 (13.33 mm) was observed. Whereas few high (Feb-07 (26.57 mm) and Feb-2013 The soil of the Jhgarbaria watershed is of Devra clay (34.36 mm)) or high (October-2006 (47.04 mm), soil, Jarkhori sandy loam, Lohgara silty loam, Newaria October-2014 (74.23 mm)) amount of rainfall is also loamy and stony land (Figure 3(a)). The watershed is reported. Figure 4(e), represent monsoon season rain- mainly dominated by Newaria loam (58.26 km or fall pattern and year 2004, 2009, and 2010 (low rainfall 38.82%), Devra clay soils (31.92 km or 21.27%) and with respect to average rainfall). Maximum rainfall in Jarkhori sandy loam soils (27.99 km or 18.65%). last 13 years was recorded for August-2013 as Presence of sand fraction in large quantities under 532.52 mm while minimum rainfall for same year in entire watershed makes it vulnerable to soil erosion. month of September (except month of September- The stoniness of the land (14.63 km or 9.75%) will act 2008). Based on 12 years rainfall data sets, average rain- as barrier to store water however leading to generate fall in study area is 199.57 mm (296.1 mm, from IMD higher amount of runoff (Castiglioni et al., 2010; web http://www.imd.gov.in/section/climate/clima Fathzadeh, 2008).The initial infiltration and transmis- teimp.pdf), which is less (97.53 mm) than from sion of surface water into an aquifer system is a func- 100 year average monsoon (June, July, August and tion of soil type and its texture. From soil classes, September) rainfall. Based on 12 years monsoon further Hydrologic Soil Group (HSG) (Figure 3(b)) month rainfall, average rainfall of month of June, July, map of study area was developed with guidelines August, and September are respectively, 139.49 (88.8 given by Chow et al. (1988). mm, from IMD), 251.86 (280 mm, from IMD), 245.73 (296.1 mm from IMD) and 161.21 mm (185.0 mm from IMD). Figure 4(f), illustrates annual rainfall from 2003 4.3. Rainfall (2003-2014) to 2014, it represents drought years and particular year The daily rainfall data of 12 years are illustrated in rainfall can correlated with particular year crop produc- Figure 4(a-c). These figures show accumulated rainfall tion. Figure 4(f) also explained the drought years (2004, over time (per day from 2003 to 2014) and low rainfall 2009, and 2010) of study area. (<600 mm) in the year 2004, 2009, and 2010 was Figure 2. (a). Land use/land cover map of 2006 of study area. 100 K. S. RAWAT ET AL. Figure 3. (a) Soil types in the study area and their spatial distribution. (b). The Hydrologic Soil Group (HSG) map developed from soil map. 4.4. Land surface temperature (LST) 2013. Hence, the maximum variation was only 1.8° C in 11 years. The 8-days LST was plotted (Figure 5(a-c)) and maximum temperature variation (32.8 to 41.3°C) was noted during 14-April-2010 to 25-Jun-2010 4.5. Runoff from CN method th th (335 to 344 8-days). Figure 5(d) shows monthly LST and reveal that average monthly LST eight Figure 6(a-c) shows destitution of CNn at special times cross 35°C limit line during different month extent and corresponding histogram showing desti- of different years, and maximum monthly average tution of CN at pixel wise (n = I, II, and III) in LST was noted for during June-2010 as 37.3°C. Due images. Runoff calculation from SCS model mainly to average function all peak values (in 8-days LST relied on CN value, which is a function of AMC, data sets, Figure 5(c)) all value range from 14.5 to slope, soil type, and land use. The CN value reflects 37.28°C, while monthly average mean value is the possible runoff generation (Rawat & Singh, noted as 25.3°C. Figure 5(e), represents annual 2017). Under the same rainfall condition, low LST, maximum LST 26.1°C was estimated in the value of CN reflect that the land has a high possi- year 2010 and minimum was 24.3°C in the year bility of water-holding capacity. While high value GEOLOGY, ECOLOGY, AND LANDSCAPES 101 Figure 4. (a). Shows an example of the merged analysis of daily precipitation for 20 July 2001. The merged analysis presents similar spatial distribution patterns with those of satellite estimates while its magnitude is close to the gauge-based analysis over areas with gauge data, (b) shows a final product after merging the inputs. (c). daily rainfall from 2003 to 2014. (d). monthly rainfall from 2003 to 2014. (e). monthly rainfall from 2003 to 2014 during monsoon season. (f). annual rainfall from 2003 to 2014. of CN, precipitation can be held by the land at a of total rainfall was converted into runoff, it comes small extent. Therefore, any class LULC with high as large amount of runoff (60.5% runoff) in next value of CN can generate a high amount of runoff month (September) by 159 mm rainfall. Similarly, which will cause of flood peak. In SCS model, AMC for high runoff (55.9% of total rainfall) during condition has influence on CN values that’s why June-2005 (because high amount of rainfall receives CN and AMC conditions are two major factors that in last days of previous month (22, 23, 25, 26, 27 can affect the runoff analysis in SCS modeling. 29, and 30 May-2005). Figure 7(b) reveals that Figure 7(a) represents a seasonal trend, the varia- during year 2013 October rainfall also produces bility in runoff except high runoff during 05- high runoff 51.9% of 293 mm rainfall, this September-07 (90.11 mm, because of high rainfall October’s runoff is given key information that a 118 mm), 05-October-2013 was noted as highest large amount of rainfall after September becomes runoff (91 mm) due to highest rainfall (120 mm) as runoff because of surface saturation condition. during end of monsoon year of 2013. This was Figure 7(c) graphical representation of annual run- special month (October) of last ten years (2003 to off with rainfall and explain rainfall and runoff 2012) when more rainfall in short time period yield of 12 months. (near about 135 mm within two days) and in year 2010 less runoff. Figure 7(b) represents monthly 4.7. Runoff from surface runoff model monsoon runoff during 2003 to 2014 and showed highest runoff (96 mm and 60.5% of total rainfall Annual runoff was estimated by eight differ - (159 mm)) during September-2007, because in ent surface runoff models (KHM, IDS, TR, August-2007 high amount of rainfall (309 mm) CR, KH, ICAR, LR and JE). Table 5 was received (total 16 days rainfall) but only 22% 102 K. S. RAWAT ET AL. Figure 4. (Continued) describes the comparative runoff results of runoff. However, CR and IDS, the remaining these models. These eight models were inde - models’ runoff predictions are under the pendent of LULC classes, soil categories, acceptable limit (based on % of annual and AMC type. These models are only rainfall). based on annual rainfall and annual tem - perature. From Table 5 , we can easily dis - 4.8. Statistical performance evaluation tinguished two categories, (i) predicted annual runoff was overestimated (CR and Comparative results of runoff estimation are obtained KH) and (ii) predicted annual runoff was through statistical tests (Table 6). Statistical results for underestimated (IDS, TR, and ICAR). Khosla’s method (R = 0.92, RMSE = 49.83, R- Predicted runoff of CR model was always RMSE = 2.3, MAE = 47.64, NRMSE = 1.99, overestimated (because in each year the pre - MBE = −47.64, PBIAS = −65.58, and BIAS = −2.17) dicted runoff was more than the actual pre - with respect to SCS-CN (RS and GIS-based model) has cipitation, like runoff of year 2013 is 34.72% been rejected based on rating of statistical index of annual rainfall), therefore, in first screen - method, because it scored high value of 4.911 ing this model can be discarded. In the same (Rank = 7) which also indicated that based on only way, KH model also predicted high annual R test any model cannot known fully or used as good GEOLOGY, ECOLOGY, AND LANDSCAPES 103 Figure 5. (a). Shows an example of the eight days Land SurfaceTemperature (LST) of India from MOD11A2, for 20 July 2014, (b) extracted LST for Allahabad district of Uttar Pradesh State, India. (c). 8-days LST data set of study area for year 2003 to 2014. (d). Monthly LST data set of study area for year 2003 to 2014. (e). Annual LST data set of study area for year 2003 to 2014. predictor or estimator. Because despite of high value (R = 0.89, RMSE = 7.75, R-RMSE = 0.24, of R = 0.92, the model comes under rank 7. Statistical NRMSE = 0.31, PBIAS = 0.01, BIAS = 0.07, results of IDS model (for plain area) with respect to MAE = 0.002, and MBE = 0.002, (≈ 0, almost zero)) SCS-CN is good because of all statistical tests result have positive responses with respect to SCS-CN 104 K. S. RAWAT ET AL. Figure 6. (a). Curve number (CN) II map and corresponding histogram. (b). Curve Number (CN) I map and corresponding histogram. (c). Curve Number (CN) III map and corresponding histogram. model and the rating process was having the lowest runoff. This may be the reason for adoption of this value of 0.008 (K = 0.001 with first Rank), further- model by JE. The performance of CR model was also more, it revealed that use of this model has satisfactory found satisfactory based on R test. Other models such results compared to SCS-CN model. Similarly, statis- as LR (model rank = 3), TR (model rank = 4), ICAR tical results for JE model showed second ranking. It (model rank= 5) and DII (model rank= 6) have lowest can be used successfully for the prediction of surface value with respect to SCS-CN model. Hence, these GEOLOGY, ECOLOGY, AND LANDSCAPES 105 Figure 7. (a). Daily rainfall- runoff time series at study area from 2002–2014 (monsoon season). (b). Monthly rainfall-runoff time series at study area from 2002–2014 (monsoon season). (c). Yearly rainfall-runoff time series at study area from 2002–2014. models have limited potential to estimate surface run- LULC, soil hydrological group, time interval of pre- off. Significant difference was found among model cipitation and physical characteristics of agricultural with respect to SCS-CN (except IDS). Ghazavi and watershed. Abasali (2003) did not consider Coutagine method and corrected Langbin method, as suitable method in 5. Conclusion arid regions. Khosroshahi (1991) has mentioned that the estimation by ICAR is more than observed value; it Runoff estimation of ungauged watershed is a chal- is more obvious in the agricultural watersheds of more lenge for hydrologists. Discharge value of ungauged than 200 km . Also, Fathzadeh (2008) considered clas- catchments is important for hydrological planning sic Coutagine and Turc approaches as non-suitable and designing of various hydraulic structures. Precise methods with significant errors. In this research, knowledge about the runoff will help in better man- according to the results of statistical tests JE method agement of water resources of the local region. It is was introduced as the best method after Inglis & De difficult to estimate the runoff more accurately from Souza (IDS) method, for runoff estimation in the ungauged watershed with coarse resolution satellite study area. The advantage of IDS model is as simple data due to high uncertainty. SCS-CN model requires and not affected by any factors related to slope, type of input of LULC, soil data, and rainfallthat can be 106 K. S. RAWAT ET AL. Table 5. Runoff estimated from different empirical model. Runoff (in cm) from models A.T A. R S. No Years (°C) (cm) SCS- CN KH IDS DII TR CR ICAR JE LR 1 2003 24.79 93.89 22.58 81.97 28.13 35.73 11.26 2064 10.37 23.11 14.62 2 2004 24.71 56.29 9.04 44.41 8.53 18.83 4.34 744 4.99 8.32 5.61 3 2005 25.80 84.77 31.86 72.36 22.35 31.50 8.87 1629 8.49 18.40 12.10 4 2006 25.93 65.27 16.99 52.8 12.2 22.73 5.35 962 5.79 10.88 7.42 5 2007 24.91 72.32 20.6 60.34 15.52 25.85 6.96 1220 7.08 13.67 9.00 6 2008 24.90 107.64 36.43 95.66 38.07 42.22 14.21 2703 12.56 30.29 18.79 7 2009 25.94 57.20 14.51 44.72 8.87 19.22 4.1 738 4.78 8.35 5.78 8 2010 26.14 59.77 10.87 47.2 9.88 20.33 4.42 801 5.04 9.08 6.28 9 2011 25.44 112.10 40.61 99.86 41.62 44.36 14.82 2881 12.93 32.44 20.23 10 2012 25.13 89.71 24.5 77.62 25.4 33.78 10.2 1863 9.54 20.93 13.44 11 2013 24.34 146.08 51.63 134.38 73.78 61.02 24.29 5073 20.10 56.54 32.58 12 2014 24.79 72.60 20.42 60.4 15.66 25.98 6.82 1211 6.95 13.63 9.06 Annual Temperature (in °C) from MODIS, A.T; Annual Rainfall (in cm) from NOAA, A.R; KH, Khosla; IDS, Inglis & De Souza; DII, Department of irrigation, India; TR, Turc relationship; CR, Coutagine relationship; ICAR, Indian Council of Agricultural Research; JE, Justin Equation; LR, Lacey Relationship; M, Model. Table 6. Statistical test values for different empirical model with respect to SCS-CN. S. No Test IDS JE LR TR ICAR DII KH CR 1 R 0.89 0.89 0.90 0.90 0.90 0.92 0.92 0.88 2 RMSE 7.75 6.43 13.32 16.95 17.99 7.60 49.83 2158.82 3 R-RMSE 0.24 0.27 0.50 0.62 0.63 0.49 2.30 70.88 4 MAE 0.00 4.53 12.09 15.37 15.95 6.79 47.64 1799.10 5 NRMSE 0.31 0.26 0.53 0.68 0.72 0.30 1.99 86.34 6 MBE 0.00 4.53 12.09 15.37 15.95 −6.79 −47.64 −1799.10 7 PBIAS 0.01 22.15 93.69 159.45 176.26 −21.36 −65.58 −98.63 8 RMSE% 2.58 2.14 4.44 5.65 6.00 2.53 16.61 719.52 9 BIAS 0.07 0.21 0.49 0.62 0.62 −0.38 −2.17 −69.33 10 K factor 0.00 0.07 0.13 0.16 0.16 0.23 0.61 0.88 11 Rating 0.01 0.55 1.06 1.29 1.32 1.88 4.91 7.04 12 MR 1 2 3 4 5 6 7 8 Note:- Coefficient of Determination, R ; Root Mean Square Error, RMSE; Relative Root Mean Square Error, R-RMSE; Mean Absolute Error, MAE; Normalized root mean square error, NRMSE; Mean bias error, MBE; RMSE%, Percentage RMSE; bias, BIAS; Mean difference bias, Model Rank, MR. obtained from satellite hence it easily provides the ORCID runoff estimation at macro level. Still there is need of Sudhir Kumar Singh http://orcid.org/0000-0001-8465- some other simple alternate model for estimating annual runoff from ungauged agricultural watershed. That can provide runoff estimate close to SCS-CN model. In this context present study reveals that References Inglis & De Souza (IDS) model is a simple and good Adham, M. I., Shirazi, S. M., Othman, F., Rahman, S., alternative of SCS-CN model. It can serve the purpose Yusop, Z., & Ismail, Z. (2014). Runoff potentiality of a of runoff estimation from ungauged watershed. IDS watershed through SCS and functional data analysis tech- model required input of annual rainfall data. This can nique. The Scientific World Journal, 2014, 1–15. https:// be generated from automatic weather station or satel- doi.org/10.1155/2014/379763 lite-based freely available data. The major drawback of Alizade, A. (1989). Principle of Applied Hydrology. Mashhad, Ghods Boniad Press.860pp all the empirical models except SCS-CN is estimation Balázs, B., Bíró, T., Dyke, G., Singh, S. K., & Szabó, S. (2018). of runoff on annual basis. Still these models provide Extracting water-related features using reflectance data reliable information about the runoff. This informa- and principal component analysis of Landsat images. tion can be utilized by the planners and policy makers Hydrological Sciences Journal, 63(2), 269–284. https:// for management and designing purposes. doi.org/10.1080/02626667.2018.1425802 Bavishi, H., & Bhagat, N. K. (2017). Rainfall Runoff Co- Relationship using Empirical Methods for Lower Mahi Basin, India. International Journal of Civil Engineering Acknowledgments Technology (IJCIET), 8(3), 575–581 Bérod, D. D., Singh, V. P., & Musy, A. (1999). A geomor- The authors are grateful to Dr. Anil Kumar Mishra phologic kinematic-wave (GKW) model for estimation of (Principal Scientist, Water Technology Centre, IARI, New floods from small alpine watersheds. Hydrological Delhi) for his critical input and suggestions on the Processes, 13(9), 1391–1416. https://doi.org/10.1002/ manuscript. (SICI)1099-1085(19990630)13:9<1391::AID-HYP809>3. 0.CO;2-B Castiglioni, S., Lombardi, L., Toth, E., Castellarin, A., & Disclosure statement Montanari, A. (2010). Calibration of rainfall-runoff mod- els in ungauged basins: A regional maximum likelihood No potential conflict of interest was reported by the authors. GEOLOGY, ECOLOGY, AND LANDSCAPES 107 approach. Advances in Water Resources, 33(10), 1235– India.Modeling Earth Systems and Environment, 4(1), 1242. https://doi.org/10.1016/j.advwatres.2010.04.009 295–310. Centeri, C., Szalai, Z., Jakab, G., Barta, K., Farsang, A., Lamb, R. (1999).Calibration of a conceptual rainfall-runoff Szabó, S., & Biró, Z. (2015). Soil erodibility calculations model for flood frequency estimation by continuous based on different particle size distribution measure- simulation. Water Resources Research, 35(10), 3103– ments. Hungarian Geographical Bulletin, 64(1), 17–23. 3114. https://doi.org/10.1029/1999WR900119 https://doi.org/10.15201/hungeobull.64.1.2 Lu, D., Mausel, P., Batistella, M., & Moran, E. (2004). Chow, V. T., Maidment, D. R., & Mays, L. W. (1988). Comparison of land-cover classification methods in the Applied hydrology. McGraw-Hill, Inc. Brazilian Amazon Basin. Photogrammetric Engineering & Devia, G. K., Ganasri, B. P., & Dwarakish, G. S. (2015). A Remote Sensing, 70(6), 723–731. https://doi.org/10.14358/ review on hydrological models. Aquatic Procedia, 4, PERS.70.6.723 1001–1007 doi:10.1016/j.aqpro.2015.02.126 Maliqi, E., & Singh, S. K. (2019). Quantitative Estimation of Dorum, A., Yarar, A., Sevimli, M. F., & Onüçyildiz, M. Soil Erosion Using Open-Access Earth Observation Data (2010). Modelling the rainfall–runoff data of susurluk Sets and Erosion Potential Model. Water Conservation basin. Expert Systems with Applications, 37(9), 6587– Science and Engineering, 4(4), 187-200. Fathzadeh, A. 6593. https://doi.org/10.1016/j.eswa.2010.02.127 (2008). Comparison of Kotain and turc method with its Ghazavi, G. H., & Abasali, V. (2003). Determination of best conversion methods in dry land (M.Sc Thesis). Tarbiat estimation annual runoff empirical method no station Modarres University, College of Natural Resources and watershed in semi-arid land in Darab Province. Journal Marine Sciences. of Natural Resources and Agriculture Science, 10(3), 10– Mohamadi, M. A., & Kavian, A. (2015). Effects of rainfall 25. patterns on runoff and soil erosion in field plots. Golshan, M., & Ebrahimi, P. (2014). Estimation of the International Soil and Water Conservation Research, 3 Runoff by Empirical Equations in Dry and Mid-Dry (4), 273–281. https://doi.org/10.1016/j.iswcr.2015.10.001 Mountainous Area without Stations (Case Study: Moore, R. J., & Clarke, R. T. (1981). A distribution function Madan Watershed, Qazvin Province-Iran). Bulletin of approach to rainfall runoff modeling. Water Resources Environment, Pharmacology and Life Sciences, 3, 77– Research, 17(5), 1367–1382. https://doi.org/10.1029/ 85 WR017i005p01367 Hosking, J. R. M., & Clarke, R. T. (1990). Rainfall-runoff Mosaffaei, Z., Jahani, A., Chahouki, M. A. Z., Goshtasb, H., relations derived from the probability theory of storage. Etemad, V., & Saffariha, M. (2020). Soil texture and plant Water Resources Research, 26(7), 1455–1463. https://doi. degradation predictive model (STPDPM) in national org/10.1029/WR026i007p01455 parks using artificial neural network (ANN).Modeling Iowa Storm water Management 2C-5 Manual (2C-5 NRCS Earth Systems and Environment, 6,–715-729. TR-55 Methodology). (2008). www.ctre.iastate.edu/pubs/ Mutreja, K. N. (1986). Applied hydrology. Tata McGraw- ... /2C-5NRCSTR-55Methodology.pdf Hill. Tata McGraw Hill, New Delhi Jahani, A., Feghhi, J., Makhdoum, M. F., & Omid, M. (2016). Pandey, A., & Dabral, P. P. (2004). Estimation of runoff for Optimized forest degradation model (OFDM): An envir- hilly catchment using satellite data. Journal of the Indian onmental decision support system for environmental Society of Remote Sensing, 32(2), 236–240. https://doi.org/ impact assessment using an artificial neural network. 10.1007/BF03030880 Journal of Environmental Planning and Management, 59 Pásztor, L., Waltner, I., Centeri, C., Belényesi, M., & Takács, (2), 222–244. https://doi.org/10.1080/09640568.2015. K. (2016). Soil erosion of Hungary assessed by spatially 1005732 explicit modelling. Journal of Maps, 12(sup1), 407–414. Jakab, G., Nemeth, T., Csepinszky, B., Madarász, B., Szalai, https://doi.org/10.1080/17445647.2016.1233913 Z., & Kertész, Á. (2013). The influence of short term soil Praveen Kumar, B.J., Pradeep, H., Lokesh, A., Akarshraj, K. sealing and crusting on hydrology and erosion at balaton H., Surendra, H. J., & Avinash, S.D. (2016). Estimation of uplands, Hungary. Carpathian Journal of Earth and Runoff using Empirical Equations and Fuzzy Logic Environmental Sciences, 8(1), 147–155. http://real.mtak. method: A case study. International Journal of Scientific hu/id/eprint/4000. & Engineering Research. 7, (5): 28–37 Kaleris, V., Kourakos, V., & Langousis, A. (2015). Rahmati, O., Panahi, M., Ghiasi, S. S., Deo, R. C., Calibration of rainfall-runoff models: The effect of the Tiefenbacher, J. P., Pradhan, B., Goshtasb, H., temporal distribution of rainfall on uncertainties in Kornejady, A., Shahabi, H., Shirzadi, A., Khosravi, H., model parameter estimation. Geophysical Research Moghaddam, D. D., Mohtashamian, M., Tien Bui, D., & abstracts, 17, EGU2015–13543. https://meetingorgani- Jahani, A. (2020). Hybridized neural fuzzy ensembles for zer.copernicus.org/EGU2015/EGU2015-13543.pdf dust source modeling and prediction. Atmospheric Khosravi, K., Mirzai, H., & Saleh, I. (2013). Assessment of Environment, 224, 117320. https://doi.org/10.1016/j.atmo empirical methods of runoff estimation by statistical test senv.2020.117320 (case study: BanadakSadat Watershed, Yazd Province). Rawat, K. S., & Singh, S. K. (2017). Surface runoff estimation International Journal of Advanced Biological and from semi-arid ungauged agricultural watershed using Biomedical Research, 1(3), 285–301. http://www.ijabbr. SCS-CN and Earth Observation Data Sets. Water com/article_6897.html. Science and Engineering.1: 233–247 https://doi.org/10. Khosroshahi, M. (1991). Water balance calculation in small 1007/s41101-017-0016-4 watershed areas and no hydrometric stations using Rawat, K. S., & Singh, S. K. (2018). Water Quality Indices empirical formulas in the Khorasan region (M.Sc and GIS-based evaluation of a decadal groundwater qual- Thesis). Department of Natural Resources, Tehran ity. Geology, Ecology, and Landscapes, 2(4), 240–255 University. doi:10.1080/24749508.2018.1452462 Kumar, N., Singh, S. K. Singh, V.G., & Dzwairo, B.,(2018). Rawat, K. S., Singh, S. K., Ray, R. L., Szabó, S., & Kumar, S. Investigation of impacts of land use/land cover change on (2020). Parameterizing the modified water cloud model water availability of Tons River Basin, Madhya Pradesh, to improve soil moisture data retrieval using vegetation 108 K. S. RAWAT ET AL. models. Hungarian Geographical Bulletin, 69(1), 17–26. Tedela, N. H., McCutcheon, S. C., Rasmussen, T. C., Hawkins, https://doi.org/10.15201/hungeobull.69.1.2 R. H., Swank, W. T., Campbell, J. L., . . . Tollner, E. W. (2012). Rawat, K. S., Singh, S. K., Singh, M. I., & Garg, B. L. (2019). Runoff Curve Numbers for 10 small forested watersheds in Comparative evaluation of vertical accuracy of elevated the mountains of the Eastern United States. Journal of points with ground control points from ASTERDEM and Hydrologic Engineering, 17(11), 1188–1198. https://doi.org/ SRTMDEM with respect to CARTOSAT-1DEM. Remote 10.1061/(ASCE)HE.1943-5584.0000436 Sensing Applications, Society and Environment, 13, 289– Thakur,, J. K., Singh,, S. K., & Ekanthalu, V. S., (2017). 297. https://doi.org/10.1016/j.rsase.2018.11.005 . Integrating remote sensing, geographic information systems Sadati, S. K. F., Heydari, M., Javaheri, N., & Othman, F. and global positioning system techniques with hydrological (2014). Water yield estimation in polrudwatershed based modeling. Applied Water Science, 7(4), 1595–1608. on empirical methods and modelling in geographic infor- USDA, S. (1986). Urban hydrology for small watersheds TR- mation system (GIS). Journal of River Engineering, 2(7), 55. Technical Release. United States Department for 1–9.https://europub.co.uk/articles/32988/view Agriculture Soil Conservation Service, Washington DC Singh, S. K., Srivastava, P. K., Szabó, S., Petropoulos, G. P., USDA-SCS (1967). United states department of agriculture Gupta, M., & Islam, T. (2017). Landscape transform and soil conservation service, engineering division Irrigation spatial metrics for mapping spatiotemporal land cover Water Requirements, Technical Release, 21, pp. 1–98 dynamics using Earth Observation data-sets. Geocarto Vaze, J., Jordan, P., Beecham, R., Frost, A., & Summerell, G. 2011. International, 32(2), 113–127.https://doi.org/10.1080/ Guidelines for rainfall-runoff modelling: Towards best practice 10106049.2015.1130084. model application. eWater Cooprative Research Centre. Singh, V. K., Kumar, D., Kashyap, P. S., Singh, P. K., Kumar, Waltner, I., Pásztor, L., Centeri, C., Takács, K., Pirkó, B., A., & Singh, S. K. (2020). Modelling of soil permeability Koós, S., & László, P. (2018). Evaluating the new soil using different data driven algorithms based on physical erosion map of Hungary—A semiquantitative properties of soil. Journal of Hydrology, 580, 124223. approach. Land Degradation & Development, 29(4), https://doi.org/10.1016/j.jhydrol.2019.124223 1295–1302.https://doi.org/10.1002/ldr.2916. Szabó, J. A., Jakab, G. I., & Szabó, B. (2015). Spatial and Wheater, H. S., Sorooshian, S., & Sharma, K. (2008). temporal heterogeneity of runoff and soil loss dynamics Modelling hydrological processes in arid and semi-arid under simulated rainfall. Hungarian Geographical areas: an introduction to the workshop. Hydrological Bulletin, 64(1), 25–34. https://doi.org/10.15201/hungeo Modelling in Arid and Semi-Arid Areas.(Cambridge bull.64.1.3 University Press) Cambridge Tadesse, L., Suryabhagavan, K. V., Sridhar, G., & Legesse, G. Zhang, Z., Sheng, L., Yang, J., Chen, X. A., Kong, L., & (2017). Land use and land cover changes and soil erosion Wagan, B. (2015). Effects of land use and slope gradient in Yezat Watershed, North Western Ethiopia. on soil erosion in a red soil hilly watershed of southern International Soil and Water Conservation Research, 5 China. Sustainability, 7(10), 14309–14325. https://doi. (2), 85–94. https://doi.org/10.1016/j.iswcr.2017.05.004 org/10.3390/su71014309

Journal

Geology Ecology and LandscapesTaylor & Francis

Published: Apr 3, 2021

Keywords: Curve number; runoff; RMSE; soil erosion; ungauged watershed

References