- Email: [email protected]
Effect of spatial scale on runoff coefficient: Evidence from the Ethiopian highlands
International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
HOSTED BY
Contents lists available at ScienceDirect
International Soil and Water Conservation Research journal homepage: www.elsevier.com/locate/iswcr
Original Research Article
Effect of spatial scale on runoff coefficient: Evidence from the Ethiopian highlands Tesfaye Mebrate Lemma a,n, Gizaw Desta Gessesse b, Asfaw Kebede Kassa c, Desalegn Chemeda Edossa d a
Debre Berhan University, Debre Berhan, Ethiopia Water and Land Resources Center, Addis Ababa, Ethiopia c Haramaya Institute of Technology, Haramaya, Ethiopia d South-Man Engineering, Manitoba, Winnipeg, Canada b
art ic l e i nf o
a b s t r a c t
Article history: Received 12 November 2017 Received in revised form 15 August 2018 Accepted 16 August 2018
The runoff coefficient (RC) is the ratio between the runoff and rainfall amounts and is scale dependent, which is due in part to the heterogeneity of watershed characteristics. This study quantified the spatial scale effects on runoff using long-term rainfall-runoff data on runoff plots and small watersheds. Effect of spatial scale on RC was studied for 12 runoff plots (2 m by 15 m) and three small watersheds (113–477 ha) in the highlands of Ethiopia using a total of 4397 and 13,925 15-day cumulative pairs of rainfall and runoff data at watershed and runoff plot scales, respectively. The observed average RC of runoff plots was extrapolated based on the extent of representation of a particular watershed in terms of slope, land use, cover and soil type. The weighted RC of plots was then compared with the observed RC of the watershed to determine a scale factor for extrapolation. A decrease in RC from plot to the watershed was observed in Anjeni and Andit Tid watersheds, while an increase in RC in Maybar watershed illustrates the role of specific watershed conditions in determining the scale effect. This, in turn, suggests that the variation in scale factor is not well explained by the difference in the area alone. The scale effect of runoff generation was better explained by extrapolating the RC based on the representation of different watershed characteristics. Thus, extrapolation exercises in runoff modeling and scaling efforts of soil and water conservation practices should consider the scale effect cautiously. & 2018 International Research and Training Center on Erosion and Sedimentation and China Water and Power Press. Production and Hosting by Elsevier B.V. This is an open access article under the CC BY-NCND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Keywords: Runoff coefficient Spatial scale Rainfall-runoff relation Ethiopian highlands
1. Introduction Sustainable watershed management is a guarantee for reducing watershed degradation (Temesgen, 2015). Thorough understanding of the hydrological processes, amongst which surface runoff generation is a fundamental process of interest (Stewart, Liu, Rupp, Higgins, & Selker, 2015), is crucial in watershed management. It has greater importance for prioritizing watersheds and selection of hot spot areas (Ramana, 2014). Runoff coefficient is the ratio of runoff to rainfall, and it is the basic input parameter in the hydrologic designs (Merz, Blö schl, & Parajka, 2006). Detailed
n
Corresponding author. E-mail addresses: [email protected] (T.M. Lemma), [email protected] (G.D. Gessesse), [email protected] (A.K. Kassa), [email protected] (D.C. Edossa). Peer review under responsibility of International Research and Training Center on Erosion and Sedimentation and China Water and Power Press.
understanding runoff coefficient of an area helps to facilitate upscaling of best soil and water conservation practices (BSWCP) that generate low runoff. In the process of up-scaling, extrapolating results from smaller to larger scales requires a due consideration of the scale effect (Gebirrye, 2004). However, extrapolation is possible only when scale factors are well understood in specific conditions. Due to lack of measured reliable data across different scales, many studies emphasize on model-based extrapolations. So far, many studies give less focus to scale problems and its determinant factors using long-term measured data (Yair & RazYassif, 2004). Several studies indicated that smaller areas have generated more runoff due to greater runoff coefficient than larger areas (Feng & Li, 2008; Penna, Mantese, Gobbi, & Borga, 2011). The result from a study on scale effect using 0.05 ha, 90 ha, and 1100 ha stated that the runoff coefficient was observed to decrease as area increases (Cerdan et al., 2004). In their study, the soil type and slope were kept similar while the percentage of arable land was
https://doi.org/10.1016/j.iswcr.2018.08.002 2095-6339/& 2018 International Research and Training Center on Erosion and Sedimentation and China Water and Power Press. Production and Hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
2
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
heterogeneous and found to be the major factor for the difference in runoff coefficient. On the contrary, a direct relationship between runoff coefficient and the watershed area was also reported (Prats, Wagenbrenner, Martins, Malvar, & Keizer, 2015). According to Feng and Li (2008), a study undertaken using six sub-basins having different drainage area shows that a smaller basin is found to have smaller runoff coefficient. Moreover, a runoff between 1 m2 and 10 m2 plots showed the non-significant difference (Asadzadeh, Gorji, Vaezi, & Shorafa, 2012). Stone, Nichols, Goodrich, and Buono (2008) generalize that humid area show an inverse relationship between runoff coefficient and drainage area while semi-arid areas exhibit the direct relationship. This implies that scale effect is not only explained by area but also other important watershed characteristics. Thus, quantifying the magnitude of scaling factor (SF) and recognizing the controlling factors is found critical to the rainfall-runoff relations (Leys, Govers, Gillijns, BeRCkmoes, & Takken, 2010; Wu, Jones, Li, & Loucks, 2006). Inference about larger watersheds using results from smaller watersheds is often misleading. Simple linear up-scaling or extrapolation would not be valid since a watershed could not be considered as the sum of individual plots (Cerdan et al., 2004). Penna et al. (2011) underline that a rainfall-runoff process examined at small experimental plots can offer insights about the same process occurring at larger scales. On the contrary, Sadeghi, Gholami, Sharifi, Khaledi, and Homaee (2015) reported that plots could not give accurate data for larger scale outcomes. The low accuracy of extrapolation is often associated with the subjective interpretation of data and inconsistencies in parameter estimation at larger watersheds (Hughes, Kapangaziwiri, & Tanner, 2013). Spatial scale effects in hydrological processes are usually described by watershed heterogeneity in terms of slope, soil, and land uses (Wu et al., 2006). Separate and combined effect of heterogeneous characteristics can cause different runoff generation conditions in an area (Gomi, Sidle, Miyata, Kosugi, & Onda, 2008; Lin & Wang, 2010; Mounirou, Hamma, Harouna, Jean-Emmanuel, & Gil, 2012). Yildiz and Barros (2009) state the importance of incorporation of detailed information that has consequences at the spatial scales of interest will help to effectively forecast spatial variability of hydrologic responses. This situation, thus, demands further investigation of the spatial scale effect with respect to the extent of representation of plot characteristics on heterogeneous watersheds. One can use the runoff coefficient corresponding to the different plots with various characteristics to explain the spatial scale effect on a heterogeneous watershed. Therefore, quantifying the magnitude of scale effects using runoff plot and watersheds are found to be critical undertakings in a situation where measured data is not available. Addressing spatial scale effect with reference to the representation of characteristics of small plots to the particular watersheds help to guide land and water management implementation efforts. In due course, effective watershed management under up-scaling approach could be promoted. The present study was conducted in humid and sub-humid watersheds in the Ethiopian highlands. Therefore, the objective of the study was to analyze and model effects of spatial scale on the runoff coefficient. The scale effect was analyzed at watershed (113–477 ha) and runoff plot (30 m2) scales. Long-term rainfall and runoff data both at plot and watershed level were aggregated into fifteen-day interval and ordered separately based on ranking. To quantify the runoff coefficient at the watershed level, the relative magnitude runoff coefficient of runoff plots were weighted by the areal extent of combined plot characteristics including slope, land use, and soil. Eventually, the scale effect was analyzed by comparing the weighted runoff coefficient of runoff plots with the measured runoff coefficient at the watershed level.
2. Materials and methods 2.1. Description of the study areas The study was conducted in three experimental watersheds located in the Ethiopian highland, namely Anjeni, Andit Tid, and Maybar watersheds. The watersheds were established in the early 1980s by the Soil Conservation Research Project (SCRP). The watersheds are located in humid and sub-humid agro-ecological zones of the country with an altitude range between 2407 and 3548 m a.s.l. (Fig. 1). Cereal-based cropping system mixed with livestock is the dominant farming system. The major biophysical features of the watersheds are presented in Table 1. The main reason to focus on these watersheds is that the watersheds are equipped with the hydro-meteorological monitoring system at watershed and runoff plot level and availability of long-term rainfall-runoff data since the 1980s. Daily rainfall and runoff data from runoff plot and daily discharge data from watershed level were recorded for the last three and half decades. Each watershed has four 2 m by 15 m runoff plots. The different runoff plot setup represents different classes of the slope, soil type, and crop cover types by considering the heterogeneity of the watersheds. As depicted in Table 2, the slope, soil, and land use characteristics of the plots and their area proportion from the total watershed are described by different research works (Abebe, Hurni, & Gete, 2013; Abrham, Tilashwork, Tesfaye, & Abdlesemed, 2016; Binyam, 2009; Elias, 2009; Tegenu, 2009; Woubet et al., 2013; Yakob, 2009). 2.2. Rainfall and runoff data A long-term data on rainfall, runoff at the plot level, and streamflow at the outlet of the watersheds have been collected since the early 1990s and used for the analysis. A cleaned dataset was collected from Water and Land Resource Center (WLRC). Out of the total dataset, Thirteen years (i.e., in the period of 1987–1993, 1996–1998, 2000 and 2008), six years (1987–1992) and twenty years (i.e., in the period of 1988–1989, 1991–1993, 1995, and 2000– 2013) rainfall and runoff data were used respectively for Anjeni, Andit Tid, and Maybar watersheds. Missing data for any single month was excluded from analysis. Since the study watersheds are small, rainfall was considered to be similar throughout each watershed area. Daily rainfall depths were computed from successive rainfall events. In the runoff test plots, runoff collection tanks were emptied every morning at 8:00 a. m. Therefore, daily rainfall and runoff amount were considered to be of the total duration between two consecutive tank emptying periods. Even among plots, emptying a tank and taking a soil sample and water depth measurements were time-consuming task compelling to reach the next test plot lately (Bayabil, Tebebu, Stoof, & Steenhuis, 2015; Junker, 2012). Total daily runoff from the test plot was computed using procedures proposed by SCRP (1982, 1984). The river stage-discharge relationships for each watershed is developed and calibrated by Bosshart (1996) and was adopted to generate discharge data of each watershed. Thus, total runoff in each study watershed for any rainfall event was computed using the adopted rating curve equations relating streamflow (Q) and streamflow stage (H). Base flow was determined using constant-discharge method (Raghunath, 2006). Thereafter, surface runoff was computed by deducting base flow from the total runoff. Daily rainfall-runoff data was then obtained by adding all events on a specific day. In order to consider watershed sizes, the runoff volumes computed were divided by the plot as well as the watershed area and resulted in runoff depth. The rainfall and runoff data were subject to screening to exclude the outliers and avoid unnecessary errors. Because of low rainfall resulting in unrealistic high runoff flow, it is customary to use large storm events the lowest limit being 25.4 mm (Ajmal,
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
3
Fig. 1. Location map of the study watersheds (Source: Authors’ illustration).
Khan, & Kim, 2016; Soulis & Valiantzas, 2012). Besides, a pair of rainfall-runoff data resulting in runoff coefficient greater than one was also excluded. To better separate the runoff events, researchers argue in favor of using longer period rainfall-runoff data (Engda et al., 2012; Giovanoli, 2014; Gizaw, 2003; Guzman, Tilahun, Zegeye, & Steenhuis, 2013; Liu et al., 2008; Meher, 2014; Merz et al., 2006; Tebebu et al., 2015 Jayantilal, 2016). Eventually, paired rainfall-runoff data satisfying the predetermined conditions were selected. A total of 4397 rank-order paired dataset at watershed level and 13,925 at runoff test plot level were used for runoff coefficient analysis. Using the paired rainfall and runoff data, runoff coefficients were computed at plot and watershed level. For the rank-order method to be applied, the collected natural pairs of rainfall and runoff data were ranked separately from largest to smallest forming new pairs of rainfall-runoff datasets. According
to EWRI (2009) and Hawkins, Ward, Woodward, and Mullen (2010), this approach is important in water design works. The method enables matching between the frequency of the rainfall to the frequency of the runoff through reducing the difference in rainfall frequency and runoff frequency. 2.3. Analyses of spatial scale effect Runoff coefficient is a dynamic and broadly used diagnostic variable in studies of runoff generation (Merze et al., 2006; Feng & Li, 2008). Scaling and comparing of runoff on different spatial scales can be made conveniently using the runoff coefficient (Patin et al., 2012). Therefore, in the subsequent spatial scale analyses, the differences revealed in the runoff coefficient of a plot to the particular watershed were considered.
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
4
Table 1 Description of the study watersheds. (Source: SCRP, 2000a, 2000b, 2000c). Watershed Features
Anjeni
AnditTid
Maybar
Location Basin Watershed Area (ha) Elevation (m) Mean annual rainfall (mm) Rainfall pattern Climate Mean annual Temp. (°C) Major soil types Land under cultivation (%)
37°31’E, 10°40′N Abay Basin 113.4 2407–2507 1690 Unimodal Sub-humid 16 Alisols, Nitosols, Cambisols 80
39°43’E, 9°48′N Abay Basin 477.5 3040–3548 1467 Bimodal Humid 12.6 Andisol, Aluvisol, Regosols, Lithosols 30
39°40’E, 11°00′N Awash Basin 112.8 2530–2858 1417 Bimodal Sub-humid 16.4 Phaeozems, Lithosols, Gleysols 60
Table 2 Characteristics of runoff test plots and proportional area coverage of test plot feature from the total watershed area in percent (parenthesis). (Source: SCRP, 2000a, 2000b, 2000c); TP is test plot. Plot feature Slope
Land use
Soil type
TPs TP-1 TP-2 TP-3 TP-4 TP-1 TP-2 TP-3 TP-4 TP-1 TP-2 TP-3 TP-4
Anjeni 28% (17.18) 12% (20.39) 16% (19.67) 22% (19.67) Cultivated (76.1) Cultivated (76.1) Grass (10.65) Fallow (6.14) VerticLuvisol (3.9) EutricNitosol (23.8) Eutric Regosol (9.2) EutricNitosol (23.8)
The scale effect on runoff coefficient was studied in relation to the relative extent of representation of runoff plot characteristics in a watershed under the situation where the watershed heterogeneous The analysis relied on the separate and combined effects of slope, soil, and land use of runoff test plots and their extent of representing the particular watershed. Each of the three characteristics in each test plot was assessed in terms of their relative percentage area in the particular watershed. Specific characteristics dominantly affecting the spatial scale effect were analyzed. The area-weighted representation was analyzed and resulted in area-weighted new Q and RC at watershed scales. The areaweighted runoff coefficient was then compared with the average runoff coefficient estimated at the outlet of the same watershed. This analysis helps to further explain the watershed heterogeneity approach by considering soil, slope, and land use characteristics of a plot and the respective watershed. Results were summarized using the scaling factor (SF). Scaling factor is a ratio of runoff coefficient of the differently sized watersheds (Heras, Nicolau, Martín, & Wilcox, 2010). It helps to analyze the influence of varying surface conditions on modifying the effect of spatial scale on runoff. Moreover, scaling factor is an important conversion factor to develop similarity between characteristics of one system (runoff test plot) to the corresponding characteristics of the other system (watershed) (Woods, 2016). Therefore, the scaling factor is the ratio of weighted RC of a watershed divided by observed RC at the watershed outlet.
AnditTid
Maybar
23% (11) 39% (65) 48% (65) 48% (65) Cultivated (34.61) Cultivated (34.61) Grazing (17.2) Cultivated (34.61) EutricRegosol (32.4) Chromic Cambisol (1) EutricRegosol (34.61) OchricAndosol (66)
16% (20) 64% (36) 43% (38) 37% (38) Cultivated (60) Grass (20) Grass (20) Cultivated (60) Deep Phaeozem (14) PhaeozemLithosols (11) PhaeozemLithosols (40) HaplicPhaeozems (12)
3. Results and discussion
practices and topographic conditions prevalent to specific plots and watersheds. It also indicates the mechanism and trend of runoff generation throughout the rainfall season. The comparison of rainfall and runoff can quantify the relative response of plot conditions. The relationship between rainfall and runoff at plot and watershed levels are illustrated in Fig. 2. The total watersheds’ fifteen-day aggregated rank-order rainfall-runoff dataset fitted a polynomial function showed a good matching with R2 of 0.93, 0.99, and 0.99 in Anjeni, Andit Tid, and Maybar, respectively. While, the pooled plots’ 15-day rainfall-runoff dataset showed better matching with R2 of 0.99, 0.99, and 0.97 in Anjeni, Andit Tid, and Maybar, respectively. The overall watershed and plots observed runoff coefficients (RCobs) were correlated positively and strongly with r-values of 0.98, 0.87, and 0.98 respectively for Anjeni, Andit Tid, and Maybar (Fig. 3). Moreover, the runoff and runoff coefficients of plots and the respective watersheds were found significantly different (P o 0.01). The observed runoff and runoff coefficient showed a consistent increment from plot to watershed scales at Anjeni and Andit Tid watersheds except in Maybar. Despite the great variation in the soil and agro-ecological conditions, the average observed runoff coefficients were more or less similar across the three watersheds (28–38%). At plot scale, the average runoff coefficient was 45.52%, 48.67%, and 14.58%, respectively at Anjeni, Andit Tid, and Maybar. Similar results in RC, 20–50% (Gizaw, 2003) in Andit Tid, and 27% (Hurni, Kebede, & Gete, 2005) and 10–32.4% (Gizaw, 2003) in Maybar, and 45.2% in Anjeni (Nyssen et al., 2010) as cited in Boshart (1998a) were reported. Runoff coefficients estimated in all sites were smaller than those in Liu et al. (2008), Engda et al. (2012), and Lemann, Roth, and Gete (2015).
3.1. Rainfall-runoff relationship
3.2. Scale effect on runoff coefficient in heterogeneous watersheds
Analysis of rainfall-runoff relationship is useful to understand the hydrologic conditions and the response of land management
The measured runoff coefficients of runoff plots are explained by slope, soil and land use characteristics. This measured runoff
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
5
Fig. 2. Rank-order rainfall (P) and runoff (Q) data relationship at watersheds and runoff test plots.
coefficient of runoff plots further extrapolated into watershed level runoff coefficient based on the extent of representation of each characteristic of individual runoff plots in a watershed. The extrapolation of runoff coefficient was then computed by weighting the runoff coefficient of runoff plots using the relative proportional area coverage of each characteristic of plots from the total watershed. The extent of representation of plot characteristics (slope, soil and land use/cover) to the particular watershed were analyzed using the general information from Tables 1 and 2. In the three watersheds, depending on the size of the watershed, initial setup of runoff plots and change in land use/cover, the representation of each watershed by the characteristics of the four test plots was different. The magnitude of representation of each watershed by the characteristics of four runoff plots varies in the three watersheds. Of the total area of the Anjeni watershed, 94% of the land use/cover, 57% of the slope, and 37% of the soil type are represented by the four runoff plots. Whereas in the Andit Tid
watershed, 52% of the land use/cover, 76% of the slope, and100% of the soil, are represented by the runoff plots. Likewise, in the Maybar watershed, 80% of the land use/cover, 95% of the slope, and 76% of the soil were represented by its plots. This shows the whole watershed area was not represented by the four plots. Therefore, an intermediate runoff coefficient values were estimated and assigned for those slope, land use/cover and soil classes that are not represented by the four runoff plots. But, the larger extents of representations of watersheds by the plots seem to reflect enhanced transferability of information from the runoff plot to the watershed. The non-uniform representation of slope, land use/cover and soil type will be discussed in terms of the relative contribution of the factors to the runoff coefficient. Nevertheless, the three watersheds were not equally represented and thus results were interpreted independently (Table 3). Based on the extent of plot characteristics in the watersheds, area-weighted runoff coefficients were estimated and presented in Table 4. The relative influence of the sources of variation in the
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
6
Fig. 3. The relationship between watershed and plot scale runoff coefficients.
Table 3 Observed runoff and runoff coefficient at watershed and plots.
Anjeni watershed Anjeni Plot average Anjeni-TP-1 Anjeni-TP-2 Anjeni-TP-3 Anjeni-TP-4 Andit Tid watershed Andit Tid Plot average AnditTid-TP-1 AnditTid-TP-2 AnditTid-TP-3 AnditTid-TP-4 Maybar watershed Maybar Plot average Maybar-TP-1 Maybar-TP-2 Maybar-TP-3 Maybar-TP-4
P (mm)
Qobs (mm)
RCobs (%)
142.03 136.67 136.94 133.50 137.85 138.37 106.00 116.09 114.05 113.33 114.11 122.87 117.57 130.01 130.09 131.11 129.07 129.76
42.82 62.08 67.27 75.39 43.79 61.86 40.27 56.19 56.53 70.41 53.73 44.09 33.29 18.94 27.31 10.12 16.03 22.28
30.15 45.52 49.12 56.47 31.77 44.70 38.00 48.67 49.57 62.13 47.09 35.88 28.32 14.58 21.00 7.72 12.42 17.17
*P and Q are averages of 15-day aggregated rainfall and runoff.
runoff coefficients was analyzed under heterogeneous surface characteristics. The weighted observed RC of the watersheds was found at 48.6%, 43.1%, and 14.4% for Anjeni, Andit Tid, and Maybar respectively. There were only slightly higher weighted RCs for Anjeni and Andit Tid watersheds. The weighted RCs in relation to slope, land use/cover and soil type representation were 51%, 49.2%, and 45.5%, at Anjeni; 48%, 41%, and 40.4% at Andit Tid and 13.9% 15.5%; and 13.7% at Maybar watersheds. In general, the relative contribution of factors to the magnitude of runoff coefficient is in the order of slope followed by land use/cover and then soil type. The scaling factor of the runoff coefficient of area-weighted observed RC to the observed RC at watershed scale was 1.61, 1.13, and 0.51 respectively for Anjeni, Andit Tid, and Maybar. Overestimation
of weighted RC at Anjeni and Andit Tid and underestimation at Maybar compared to the measured RC might arise due to lack of complete watershed representation by the four runoff plots and the relative location specific importance of plot characteristics. Cultivated areas in Anjeni and Maybar watersheds cover 80% and 60%, respectively. These extents were consistent with the order of extent of representativeness of plots’ characteristics signifying the influence of land use/cover (Li & Wang, 2010). The estimated area-weighted runoff coefficient increased for runoff plots’ slopes up to 50% and reduced for steeper slopes in Andit Tid and Maybar watersheds. Similar trends for the same ranges of slopes were reported (Patin et al., 2012). However, the slope-runoff coefficient trend in Anjeni was found to be random across runoff plots. In Anjeni, about 65% of the area is covered with deep to moderately deep, drained to moderately drained soils (SCRP, 2000a). In Andit Tid most of the areas are covered with shallow, stony, up to 2 m deep and excellent water storing capacity soils (SCRP, 2000b). About 75% areas of Maybar was steep to very steep, covered by shallow (average of 15 cm) soils that are excessively drained and well structured (SCRP, 2000c). The remaining, from gentle to the flat valley area, is covered with deep to very deep and poorly drained soil. In these flatter and valley floor areas, the water table can reach from 20 to 50 cm. According to Bayabil, Tilahun, Collick, Birru, and Steenhui (2010), and Engda et al. (2012), in Maybar watershed rainfall infiltrated into large steep areas and reached the downslope area through interflow resulted in the rise of the water table. As explained earlier, runoff test plots share similar characteristics with their watersheds reflecting lower runoff coefficient in the watershed than the plots. In spite of the average 83.67% representation of runoff plots in Maybar, much of the runoff is generated in the quarter saturated valley floor, and closer to the stream area. This area is not represented by the runoff plots. This situation substantiates the influence of spatial scale on runoff generation (Li & Wang, 2010). Besides, it necessitates further investigation of relationships between the spatial distribution of
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
7
Table 4 Representation of plots to the watershed (SF) in slope, soil, and land use characteristics. Plot features
Slope Land use Soil type Weighted Avg. RCobs Watershed Avg. RCobs
Anjeni
Andit Tid
Maybar
Weighted RCobs (%)
SFobs
Weighted RCobs (%)
SFobs
Weighted RCobs (%)
SFobs
51.03 49.26 45.55 48.6 30.15
1.69 1.63 1.51 1.61
47.98 41.01 40.36 43.10 38.0
1.26 1.08 1.06 1.13
13.86 15.48 13.73 14.4 28.3
0.49 0.55 0.59 0.51
SF is Scaling Factor.
plots and potential runoff generating areas within the watershed. In Anjeni and Andit Tid, plots are situated where much of the runoff is produced. For example, unlike in Maybar, 60% of the area in Anjeni contributes to base flow and the runoff is not generated in some specific part (Elias, 2009). This situation dictates that runoff generation is not affected by topography rather by soil type or land use. In Maybar, although the three surface characteristics represented nearly in equal share, much of the runoff is generated on specific topographic positions. That topographic position is the small valley floor where the runoff coefficient of the watershed is determined. The variability of runoff coefficient in plots within a watershed revealed that the variation is associated with the probable differences in the opportunity of flow to move extra distance and infiltrate into the ground (Gomi et al., 2008). Consequently, research results showed that variations in runoff coefficient are implications of lost/gained opportunities for transfer of flow through differences in soil, land use, and slope. Rainfall-runoff process in a watershed is governed by lots of known and unknown factors requiring special reasoning for special responses (Ramana, 2014; Shi, Chen, Fang, Qin, & Cai, 2009). The connectivity of surface flow is directly linked to the dominant surface characteristics i.e., soil, vegetation (land use), and slope conditions that control the hydrological response (Gomi et al., 2008; Lin & Wang, 2010). Unless interventions are proposed, designed, and implemented in reference to the proper transfer of hydrological information from small scale to larger scale, negative externalities could emerge within and out of the area intervened. This is because that result from small-scale experiments usually considers only part of the dominant hydrologic processes. As such, recommended measures could not have a proportional effect on the watershed level. Although scale effect is related to the scale of interest, undermining its influence that impacts runoff generation capacity of an area could cause problems reaching to downstream impacts (Hunink, Droogers, Kauffman, Mwaniki, & Bouma, 2012). To promote watershed development through up-scaling of BSWCPs to the areas where they can best fit with minimized negative externalities, scaling factor (SF) needs inferences in hydrologic processes. The dramatic change in runoff coefficients resulted because of the high correlation of surface characteristics with runoff coefficient at plot scale than at watershed scale (Merze et al., 2006). In cases where pronounced variability in surface condition between small and large areas occurred, the customary consistent pattern of runoff generation (decreasing runoff with increasing area) could be altered (Gomi et al., 2008). This inverse relation between runoff and drainage area is prevalent in humid and sub-humid areas (Stone et al., 2008) with the exception of watersheds like Maybar. Therefore, the results of this study dictate that assessment should be done at watershed scale or else runoff plot results are used with appropriate scaling factors. Otherwise, under or overestimation of extrapolation of runoff in plots experiments might lead to erroneous management decisions (Leys et al., 2010).
4. Conclusion and recommendation The purpose of the study was to analyze the effects of spatial scale on hydrological processes amongst which runoff coefficient is of critical interest. Long-term rainfall and runoff data from three different watersheds and four equal sized runoff plots in each of the watershed were used for the analysis. In all the three study sites, the runoff coefficient scaling factor from runoff plot to the watershed was determined. Area-weighted runoff coefficients estimated by considering the heterogeneity of watersheds represented by the extent of plot characteristics (slope, land use/cover and soil type). The results showed that runoff coefficient was reduced from plot to the watershed in Anjeni and Andit Tid, while it was the reverse in Maybar. Lower runoff coefficient at the plot than watershed values in Maybar challenges the common results of high runoff coefficient from a small area and low runoff coefficient from the larger area. Thus, taking the variable response of spatial scale on the runoff coefficient into account, the proportional effect of small plots to large watersheds is not always expected. Therefore, extrapolation of results from a small area to the enclosing larger area for up-scaling purposes shall be done with caution. The scale of representativeness of plots to the particular watershed with marked variability and the combined effect of surface characteristics need to be areas of future research by considering local conditions. Furthermore, it is necessary to investigate relationships between the spatial location of plots and the potential runoff generating areas within a watershed.
Acknowledgments The thoughtful supply of raw data and information was from Water and Land Resources Center of Ethiopia. Dr. Daniel Bekele from Melkasa Agricultural Research Center has helped on the manuscript structure and Amsalu Weldei (Ph.D. candidate) provided me support in editing the English.
References Abebe, S., Hurni, H., & Gete, Z. (2013). Long-term changes in soil-based ecological services at three sites in Ethiopia. Journal of Ecology and the Natural Environment, 5(8), 172–180. Abrham, T. G., Tilashwork, C., Tesfaye, F., & Abdlesemed, J. (2016). Impact assessment of land use /change on soil erosion and rural livelihood in Andit Tid watershed, North Shewa, Ethiopia. Asia Pacific Journal of Energy and Environment, 2, 125–132. Ajmal, M., Khan, T. A., & Kim, T. (2016). A CN-based ensembled hydrological model for enhanced watershed runoff prediction Muhammad. Water, 8, 20. Asadzadeh, F. M., Gorji, A., Vaezi, Sokouti R., & Shorafa, M. (2012). Scale effect on runoff from filed plots under natural rainfall. American-Eurasian Journal of Agriculture and Environmental Science, 12(9), 1148–1152. Bayabil, H. K., Tebebu, T. Y., Stoof, C. R., & Steenhuis, T. S. (2015). Spatial and temporal runoff processes in the degraded Ethiopian Highlands: The Anjeni Watershed. Journal of Hydrology and Earth System Sciences, 12, 4387–4411. Bayabil, H. K., Tilahun, S. A., Collick, A. S., Birru, Y., & Steenhui, T. S. (2010). Are runoff processes ecologically or topographically driven in the (sub) humid Ethiopian
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
8
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
highlands? The Case of the Maybar Watershed. Ecohydrology(3), 457–466. Binyam, B. (2009). SWAT to identify watershed management options: (Anjeni Watershed, Blue Nile Basin, Ethiopia) (M.Sc. Thesis). USA: Cornell University. Bosshart, U. (1996). Measurement of river discharge for the SCRP research catchments: Methodology and theoretical background. Switzerland: University of Berne (Research Report 29). Cerdan, Olivier, Le Bissonnais, Yves, Govers, Gerard, Lecomte, V., Oost, Kristof, Couturier, A., … Dubreuil, N. (2004). Scale effect on runoff from experimental plots to catchments in agricultural areas in Normandy. Journal of Hydrology, 299, 4–14. http://dx.doi.org/10.1016/j.jhydrol.2004.02.017. Elias, S., L. (2009). Modeling rainfall-runoff relationships for the Anjeni Watershed in the Blue Nile Basin (M.Sc. Thesis). USA: Cornell University. Engda, T., A., Bayabil, H., Legesse, E., Ayana, E., Tilahun, S., Collick, Amy S., … Tammo, S. Steenhui (2012). Watershed hydrology of the (Semi) humid Ethiopian highlands In: Nile River Basin (pp. 145–162). Springer Science. EWRI (Environmental and Water Resources Institute) (2009). Curve number hydrology: State of the practice. Alexander Bell Drive Reston, Virginia, USA: American Society of Civil Engineers. Feng, P., & Li, J. Z. (2008). Scale effects on runoff coefficient in mesoscale and largescale sub-basins in the Luanhe River Basin. Journal of Hydrology and Earth System Sciences, 5, 1511–1531. Gebirrye, T. D. (2004). Investigating scale, rainfall-runoff sequences and bmp effects on phosphorus, runoff and sediment yield (Doctoral dissertation). USA: Oklahoma State University. Giovanoli, L. (2014). Linking land use mapping and runoff estimation in the Gerda Watershed in the Ethiopian Highlands (M.Sc. Thesis). Switzerland: University of Bern. Gizaw, D. (2003). Estimation of runoff coefficients at different growth stages of crops in the highlands of Amhara Region (M.Sc. Thesis). Ethiopia: Alemaya University. Gomi, T., Sidle, R. C., Miyata, S., Kosugi, K., & Onda, Y. (2008). Dynamic runoff connectivity of overland flow on steep forested hillslopes: Scale effects and runoff transfer. Water Resources Research, 44, W08411. http://dx.doi.org/ 10.1029/2007WR005894. Guzman, C. D., Tilahun, S. A., Zegeye, A. D., & Steenhuis, T. S. (2013). Suspended sediment concentration-discharge relationships in the (sub-) humid Ethiopian highlands. Journal of Hydrology and Earth Science System, 17, 1067–1077. Hawkins, R. H., Ward T. J., Woodward, D. E., and Mullen J. A. (2010). Continuing Evolution of Rainfall-Runoff and the Curve Number. 2nd Joint Federal Interagency Conference, June 27 - July 1, 2010, Las Vegas, NV USA. Heras, M. M., Nicolau, J. M., Martín, L. M., & Wilcox, B. P. (2010). Plot-scale effects on runoff and erosion along a slope degradation gradient. Water Resources Research, 46, W04503. http://dx.doi.org/10.1029/2009WR007875. Hughes, D. A., Kapangaziwiri, E., & Tanner, J. (2013). Spatial scale effects on model parameter estimation and predictive uncertainty in ungauged basins. Journal of Hydrology Research, 44(3), 441–453. http://dx.doi.org/10.2166/nh.2012.049. Hunink, J. E., Droogers, P., Kauffman, S., Mwaniki, B. M., & Bouma, J. (2012). Quantitative simulation tools to analyze up- and downstream interactions of soil and water conservation measures: Supporting policymaking in the Green Water credits program of Kenya. Journal of Environmental Management, 111, 187–194. Hurni, H., Kebede, Tato, & Gete, Zeleke (2005). The implications of changes in population, land use, and land management for surface runoff in the upper Nile Basin Area of Ethiopia. Mountain Research and Development, 25(2), 147–154. Junker, L. (2012). Evaluation of the AnnAGNPS model used for river discharge and sediment yield predictions in a microscale catchment in the Highlands of Ethiopia (Master of Science Thesis). Switzerland: University of Berne. Lemann, T., Roth, V., & Gete, Z. (2015). Impact of precipitation and temperature on hydrological responses in the Ethiopian highlands. Hydrological Sciences, 62(2), 270–282. Leys, A., Govers, G., Gillijns, K., BeRCkmoes, E., & Takken, I. (2010). Scale effects on runoff and erosion losses from arable land under conservation and conventional tillage: The role of residue cover. Journal of Hydrology, 390(2010), 143–154. Lin, Y., & Wang, G. X. (2010). Scale effect on runoff in alpine mountain catchments on China's Gongga Mountain. Hydrology Earth System Science Discussion, 7, 2157–2186. Liu, B. M., Collick, A. S., Gete, Z., Enyew, A., Easton, Z. M., & Steenhuis, T. S. (2008). Rainfall-discharge relationships for a monsoonal climate in the Ethiopian Highlands. Journal of Hydrological Processes, 22, 1059–1067. Meher, J. (2014). Rainfall and runoff estimation using hydrological models and ANN techniques (Doctoral Dissertation). India: National Institute of Technology Rourkela. Merz, R., Blö schl, G., & Parajka, J. (2006). Spatio-temporal variability of event runoff coefficients. Journal of Hydrology, 331, 591–604. Mounirou, L. A., Hamma, Y., Harouna, K., Jean-Emmanuel, P., & Gil, M. (2012). Measuring runoff by plots at different scales: Understanding and analyzing the sources of variation. Hydrology and, Environment(20), 865–890. Nyssen, J., Clymans, W., Descheemaeker, K., Poesen, J., Vandecasteele, I., VanmaeRCke, M., & Walraevens, K. (2010). Impact of soil and water conservation measures on watershed hydrological response – a case in North Ethiopia. Journal of Hydrological Processes, 24, 1880–1895. Patin, J., Mouche, E., Ribolzi, O., Chaplot, V., Sengtahevanghoung, O., Latsachak, K. O., & Valentin, C. (2012). Analyses of runoff production at the plot scale during a
long-term survey of a small agricultural catchment in Lao PDR. Journal of Hydrology, 426–427(2012), 79–92. Penna, N., D. Mantese, N., Gobbi, A., & Borga, M. (2011). Runoff response at different spatial scales: Moving from small experimental areas to mesoscale watersheds. Research Support, Innovation & Technology Transfer, 62, 1–4. Prats, S. A., Wagenbrenner, Joseph W., Martins, Martinho A. S., Malvar, Maruxa Cortizo, & Keizer, Jan Jacob (2015). Hydrologic implications of post-fire mulching across different spatial scales. Land Degradation and Development, 27, 1440–1452. Raghunath, H. M. (2006). Hydrology: Principles. Analyses. Design. Revised ((2nd ed.). New Delhi, India: New Age International (P) Limited, Publishers. Ramana, G. V. (2014). Regression analyses of rainfall and runoff process of a typical watershed. International Journal of Sciences and Applied Information Technology, 3(1), 16–26. Sadeghi, S. R., Gholami, L., Sharifi, E., Khaledi, A. D., & Homaee, M. (2015). Scale effect on runoff and soil loss control using rice straw mulch under laboratory conditions. Solid Earth, 6, 1–8. SCRP (Soil Conservation Research Program) (1982). SCRP – First Progress Report (the year 1981)Volume 2. University of Bern, Switzerland, in association with The United Nations University, Tokyo, Japan. SCRP (Soil Conservation Research Program) (1984). Compilation of phase –I progress Reports (the year 1981, 1982, and 1983) – by Hans Hurni – 1984. The University of Bern, Switzerland in association with The United Nations University, Tokyo, Japan. SCRP (Soil Conservation Research Program) (2000a). Area of Anjeni, Gojam, Ethiopia: Long-term Monitoring of the Agricultural Environment 1984 – 1994. Soil Conservation Research Program (SCRP). Center for Development and Environment, University of Berne, Switzerland, in Association with the Ministry ofAgriculture, Ethiopia. SCRP (Soil Conservation Research Program) (2000b). Area of Andit Tid, Shewa, Ethiopia: Long-term Monitoring of the Agricultural Environment 1982 – 1994. Soil Conservation Research Program (SCRP). Center for Development and Environment, University of Berne, Switzerland, in Association with the Ministry of Agriculture, Ethiopia. SCRP (Soil Conservation Research Program) (2000c). Area of Maybar, Wello, Ethiopia: Long-term Monitoring of the Agricultural Environment 1981 – 1994. Soil Conservation Research Program (SCRP). Center for Development and Environment, University of Berne, Switzerland, in Association with the Ministry ofAgriculture, Ethiopia. Shi, Z., Chen, L., Fang, N., Qin, D., & Cai, C. (2009). Research on the SCS-CN initial abstraction ratio using rainfall-runoff event analyses in the three Gorges area, China. Catena, 77, 1–7 (2009). Soulis, K. X., & Valiantzas, J. D. (2012). Identification of the SCS-CN parameter spatial distribution using rainfall-runoff data in heterogeneous watersheds. Journal of Water Resource Management, 27, 1737–1749. Stewart, R. D., Liu, Z., Rupp, D. E., Higgins, C. W., & Selker, J. S. (2015). A new instrument to measure plot-scale runoff. Geoscientific Instrumentation Methods and Data System, 4, 57–64. Stone, J. J., Nichols, M. H., Goodrich, D. C., & Buono, J. (2008). Long-term runoff database, Walnut Gulch Experimental Watershed, Arizona, United States. Journal Water Resources Research (p. 44), 44. Tebebu, T. Y., Steenhuis, T. S., Dagnew, D. C., Guzman, C. D., Bayabil, H. K., Zegeye, A. D., & Tilahun, S. A. (2015). Improving efficacy of landscape interventions in the (sub) humid Ethiopian highlands by improved understanding of runoff processes. Frontiers of Earth Science, 3, 49. Tegenu, A., E. (2009). Modeling rainfall, runoff and soil loss relationships in the northeastern highlands of Ethiopia, Andit Tid Watershed (M.Sc. Thesis). USA: Cornell University. Temesgen, G. (2015). The implications of watershed management for reversing land degradation in Ethiopia. Research Journal of Agriculture and Environmental Management, 4(1), 5–12. Woods, R. (2016). The Search for Catchment -Scale Physics: Spatial Scaling & Similarity with Hydrological Heterogeneity. Symposium in Honor of Eric Wood: Observations and Modeling across Scales, June2-3, 2016 University of Bristol, Princeton, USA. Woubet, G. Alemu, Tadele, A., Birru, Y., Selassie, Yihenew G., Wolfgramm, Bettina, & Hurni, H.( (2013). Impacts of soil and water conservation on land suitability to crops: The case of Anjeni watershed, Northwest Ethiopia. Journal of Agricultural Science, 5(2). Wu, J., Jones, K. B., Li, H., & Loucks, O. L. (2006). Scaling and Uncertainty Analysis in Ecology: Methods and Applications ((1st ed.). . Springer (Printed in the Netherlands). Yair, A., & Raz-Yassif, N. (2004). Hydrological processes in a small arid catchment: Scale effects of rainfall and slope length. Geomorphology, 61(1), 155–169. Yakob, Mohammed (2009). Climate change impact assessment on soil water availability and crop production in Anjeni Watershed Blue Nile Basin (M.Sc. Thesis). Ethiopia: Arbraminch University. Yildiz, O., & Barros, A. P. (2009). Evaluating spatial variability and scale effects on hydrologic processes in a midsize river basin. Scientific Research and Essay(4), 217–225.
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
HOSTED BY
Contents lists available at ScienceDirect
International Soil and Water Conservation Research journal homepage: www.elsevier.com/locate/iswcr
Original Research Article
Effect of spatial scale on runoff coefficient: Evidence from the Ethiopian highlands Tesfaye Mebrate Lemma a,n, Gizaw Desta Gessesse b, Asfaw Kebede Kassa c, Desalegn Chemeda Edossa d a
Debre Berhan University, Debre Berhan, Ethiopia Water and Land Resources Center, Addis Ababa, Ethiopia c Haramaya Institute of Technology, Haramaya, Ethiopia d South-Man Engineering, Manitoba, Winnipeg, Canada b
art ic l e i nf o
a b s t r a c t
Article history: Received 12 November 2017 Received in revised form 15 August 2018 Accepted 16 August 2018
The runoff coefficient (RC) is the ratio between the runoff and rainfall amounts and is scale dependent, which is due in part to the heterogeneity of watershed characteristics. This study quantified the spatial scale effects on runoff using long-term rainfall-runoff data on runoff plots and small watersheds. Effect of spatial scale on RC was studied for 12 runoff plots (2 m by 15 m) and three small watersheds (113–477 ha) in the highlands of Ethiopia using a total of 4397 and 13,925 15-day cumulative pairs of rainfall and runoff data at watershed and runoff plot scales, respectively. The observed average RC of runoff plots was extrapolated based on the extent of representation of a particular watershed in terms of slope, land use, cover and soil type. The weighted RC of plots was then compared with the observed RC of the watershed to determine a scale factor for extrapolation. A decrease in RC from plot to the watershed was observed in Anjeni and Andit Tid watersheds, while an increase in RC in Maybar watershed illustrates the role of specific watershed conditions in determining the scale effect. This, in turn, suggests that the variation in scale factor is not well explained by the difference in the area alone. The scale effect of runoff generation was better explained by extrapolating the RC based on the representation of different watershed characteristics. Thus, extrapolation exercises in runoff modeling and scaling efforts of soil and water conservation practices should consider the scale effect cautiously. & 2018 International Research and Training Center on Erosion and Sedimentation and China Water and Power Press. Production and Hosting by Elsevier B.V. This is an open access article under the CC BY-NCND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Keywords: Runoff coefficient Spatial scale Rainfall-runoff relation Ethiopian highlands
1. Introduction Sustainable watershed management is a guarantee for reducing watershed degradation (Temesgen, 2015). Thorough understanding of the hydrological processes, amongst which surface runoff generation is a fundamental process of interest (Stewart, Liu, Rupp, Higgins, & Selker, 2015), is crucial in watershed management. It has greater importance for prioritizing watersheds and selection of hot spot areas (Ramana, 2014). Runoff coefficient is the ratio of runoff to rainfall, and it is the basic input parameter in the hydrologic designs (Merz, Blö schl, & Parajka, 2006). Detailed
n
Corresponding author. E-mail addresses: [email protected] (T.M. Lemma), [email protected] (G.D. Gessesse), [email protected] (A.K. Kassa), [email protected] (D.C. Edossa). Peer review under responsibility of International Research and Training Center on Erosion and Sedimentation and China Water and Power Press.
understanding runoff coefficient of an area helps to facilitate upscaling of best soil and water conservation practices (BSWCP) that generate low runoff. In the process of up-scaling, extrapolating results from smaller to larger scales requires a due consideration of the scale effect (Gebirrye, 2004). However, extrapolation is possible only when scale factors are well understood in specific conditions. Due to lack of measured reliable data across different scales, many studies emphasize on model-based extrapolations. So far, many studies give less focus to scale problems and its determinant factors using long-term measured data (Yair & RazYassif, 2004). Several studies indicated that smaller areas have generated more runoff due to greater runoff coefficient than larger areas (Feng & Li, 2008; Penna, Mantese, Gobbi, & Borga, 2011). The result from a study on scale effect using 0.05 ha, 90 ha, and 1100 ha stated that the runoff coefficient was observed to decrease as area increases (Cerdan et al., 2004). In their study, the soil type and slope were kept similar while the percentage of arable land was
https://doi.org/10.1016/j.iswcr.2018.08.002 2095-6339/& 2018 International Research and Training Center on Erosion and Sedimentation and China Water and Power Press. Production and Hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
2
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
heterogeneous and found to be the major factor for the difference in runoff coefficient. On the contrary, a direct relationship between runoff coefficient and the watershed area was also reported (Prats, Wagenbrenner, Martins, Malvar, & Keizer, 2015). According to Feng and Li (2008), a study undertaken using six sub-basins having different drainage area shows that a smaller basin is found to have smaller runoff coefficient. Moreover, a runoff between 1 m2 and 10 m2 plots showed the non-significant difference (Asadzadeh, Gorji, Vaezi, & Shorafa, 2012). Stone, Nichols, Goodrich, and Buono (2008) generalize that humid area show an inverse relationship between runoff coefficient and drainage area while semi-arid areas exhibit the direct relationship. This implies that scale effect is not only explained by area but also other important watershed characteristics. Thus, quantifying the magnitude of scaling factor (SF) and recognizing the controlling factors is found critical to the rainfall-runoff relations (Leys, Govers, Gillijns, BeRCkmoes, & Takken, 2010; Wu, Jones, Li, & Loucks, 2006). Inference about larger watersheds using results from smaller watersheds is often misleading. Simple linear up-scaling or extrapolation would not be valid since a watershed could not be considered as the sum of individual plots (Cerdan et al., 2004). Penna et al. (2011) underline that a rainfall-runoff process examined at small experimental plots can offer insights about the same process occurring at larger scales. On the contrary, Sadeghi, Gholami, Sharifi, Khaledi, and Homaee (2015) reported that plots could not give accurate data for larger scale outcomes. The low accuracy of extrapolation is often associated with the subjective interpretation of data and inconsistencies in parameter estimation at larger watersheds (Hughes, Kapangaziwiri, & Tanner, 2013). Spatial scale effects in hydrological processes are usually described by watershed heterogeneity in terms of slope, soil, and land uses (Wu et al., 2006). Separate and combined effect of heterogeneous characteristics can cause different runoff generation conditions in an area (Gomi, Sidle, Miyata, Kosugi, & Onda, 2008; Lin & Wang, 2010; Mounirou, Hamma, Harouna, Jean-Emmanuel, & Gil, 2012). Yildiz and Barros (2009) state the importance of incorporation of detailed information that has consequences at the spatial scales of interest will help to effectively forecast spatial variability of hydrologic responses. This situation, thus, demands further investigation of the spatial scale effect with respect to the extent of representation of plot characteristics on heterogeneous watersheds. One can use the runoff coefficient corresponding to the different plots with various characteristics to explain the spatial scale effect on a heterogeneous watershed. Therefore, quantifying the magnitude of scale effects using runoff plot and watersheds are found to be critical undertakings in a situation where measured data is not available. Addressing spatial scale effect with reference to the representation of characteristics of small plots to the particular watersheds help to guide land and water management implementation efforts. In due course, effective watershed management under up-scaling approach could be promoted. The present study was conducted in humid and sub-humid watersheds in the Ethiopian highlands. Therefore, the objective of the study was to analyze and model effects of spatial scale on the runoff coefficient. The scale effect was analyzed at watershed (113–477 ha) and runoff plot (30 m2) scales. Long-term rainfall and runoff data both at plot and watershed level were aggregated into fifteen-day interval and ordered separately based on ranking. To quantify the runoff coefficient at the watershed level, the relative magnitude runoff coefficient of runoff plots were weighted by the areal extent of combined plot characteristics including slope, land use, and soil. Eventually, the scale effect was analyzed by comparing the weighted runoff coefficient of runoff plots with the measured runoff coefficient at the watershed level.
2. Materials and methods 2.1. Description of the study areas The study was conducted in three experimental watersheds located in the Ethiopian highland, namely Anjeni, Andit Tid, and Maybar watersheds. The watersheds were established in the early 1980s by the Soil Conservation Research Project (SCRP). The watersheds are located in humid and sub-humid agro-ecological zones of the country with an altitude range between 2407 and 3548 m a.s.l. (Fig. 1). Cereal-based cropping system mixed with livestock is the dominant farming system. The major biophysical features of the watersheds are presented in Table 1. The main reason to focus on these watersheds is that the watersheds are equipped with the hydro-meteorological monitoring system at watershed and runoff plot level and availability of long-term rainfall-runoff data since the 1980s. Daily rainfall and runoff data from runoff plot and daily discharge data from watershed level were recorded for the last three and half decades. Each watershed has four 2 m by 15 m runoff plots. The different runoff plot setup represents different classes of the slope, soil type, and crop cover types by considering the heterogeneity of the watersheds. As depicted in Table 2, the slope, soil, and land use characteristics of the plots and their area proportion from the total watershed are described by different research works (Abebe, Hurni, & Gete, 2013; Abrham, Tilashwork, Tesfaye, & Abdlesemed, 2016; Binyam, 2009; Elias, 2009; Tegenu, 2009; Woubet et al., 2013; Yakob, 2009). 2.2. Rainfall and runoff data A long-term data on rainfall, runoff at the plot level, and streamflow at the outlet of the watersheds have been collected since the early 1990s and used for the analysis. A cleaned dataset was collected from Water and Land Resource Center (WLRC). Out of the total dataset, Thirteen years (i.e., in the period of 1987–1993, 1996–1998, 2000 and 2008), six years (1987–1992) and twenty years (i.e., in the period of 1988–1989, 1991–1993, 1995, and 2000– 2013) rainfall and runoff data were used respectively for Anjeni, Andit Tid, and Maybar watersheds. Missing data for any single month was excluded from analysis. Since the study watersheds are small, rainfall was considered to be similar throughout each watershed area. Daily rainfall depths were computed from successive rainfall events. In the runoff test plots, runoff collection tanks were emptied every morning at 8:00 a. m. Therefore, daily rainfall and runoff amount were considered to be of the total duration between two consecutive tank emptying periods. Even among plots, emptying a tank and taking a soil sample and water depth measurements were time-consuming task compelling to reach the next test plot lately (Bayabil, Tebebu, Stoof, & Steenhuis, 2015; Junker, 2012). Total daily runoff from the test plot was computed using procedures proposed by SCRP (1982, 1984). The river stage-discharge relationships for each watershed is developed and calibrated by Bosshart (1996) and was adopted to generate discharge data of each watershed. Thus, total runoff in each study watershed for any rainfall event was computed using the adopted rating curve equations relating streamflow (Q) and streamflow stage (H). Base flow was determined using constant-discharge method (Raghunath, 2006). Thereafter, surface runoff was computed by deducting base flow from the total runoff. Daily rainfall-runoff data was then obtained by adding all events on a specific day. In order to consider watershed sizes, the runoff volumes computed were divided by the plot as well as the watershed area and resulted in runoff depth. The rainfall and runoff data were subject to screening to exclude the outliers and avoid unnecessary errors. Because of low rainfall resulting in unrealistic high runoff flow, it is customary to use large storm events the lowest limit being 25.4 mm (Ajmal,
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
3
Fig. 1. Location map of the study watersheds (Source: Authors’ illustration).
Khan, & Kim, 2016; Soulis & Valiantzas, 2012). Besides, a pair of rainfall-runoff data resulting in runoff coefficient greater than one was also excluded. To better separate the runoff events, researchers argue in favor of using longer period rainfall-runoff data (Engda et al., 2012; Giovanoli, 2014; Gizaw, 2003; Guzman, Tilahun, Zegeye, & Steenhuis, 2013; Liu et al., 2008; Meher, 2014; Merz et al., 2006; Tebebu et al., 2015 Jayantilal, 2016). Eventually, paired rainfall-runoff data satisfying the predetermined conditions were selected. A total of 4397 rank-order paired dataset at watershed level and 13,925 at runoff test plot level were used for runoff coefficient analysis. Using the paired rainfall and runoff data, runoff coefficients were computed at plot and watershed level. For the rank-order method to be applied, the collected natural pairs of rainfall and runoff data were ranked separately from largest to smallest forming new pairs of rainfall-runoff datasets. According
to EWRI (2009) and Hawkins, Ward, Woodward, and Mullen (2010), this approach is important in water design works. The method enables matching between the frequency of the rainfall to the frequency of the runoff through reducing the difference in rainfall frequency and runoff frequency. 2.3. Analyses of spatial scale effect Runoff coefficient is a dynamic and broadly used diagnostic variable in studies of runoff generation (Merze et al., 2006; Feng & Li, 2008). Scaling and comparing of runoff on different spatial scales can be made conveniently using the runoff coefficient (Patin et al., 2012). Therefore, in the subsequent spatial scale analyses, the differences revealed in the runoff coefficient of a plot to the particular watershed were considered.
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
4
Table 1 Description of the study watersheds. (Source: SCRP, 2000a, 2000b, 2000c). Watershed Features
Anjeni
AnditTid
Maybar
Location Basin Watershed Area (ha) Elevation (m) Mean annual rainfall (mm) Rainfall pattern Climate Mean annual Temp. (°C) Major soil types Land under cultivation (%)
37°31’E, 10°40′N Abay Basin 113.4 2407–2507 1690 Unimodal Sub-humid 16 Alisols, Nitosols, Cambisols 80
39°43’E, 9°48′N Abay Basin 477.5 3040–3548 1467 Bimodal Humid 12.6 Andisol, Aluvisol, Regosols, Lithosols 30
39°40’E, 11°00′N Awash Basin 112.8 2530–2858 1417 Bimodal Sub-humid 16.4 Phaeozems, Lithosols, Gleysols 60
Table 2 Characteristics of runoff test plots and proportional area coverage of test plot feature from the total watershed area in percent (parenthesis). (Source: SCRP, 2000a, 2000b, 2000c); TP is test plot. Plot feature Slope
Land use
Soil type
TPs TP-1 TP-2 TP-3 TP-4 TP-1 TP-2 TP-3 TP-4 TP-1 TP-2 TP-3 TP-4
Anjeni 28% (17.18) 12% (20.39) 16% (19.67) 22% (19.67) Cultivated (76.1) Cultivated (76.1) Grass (10.65) Fallow (6.14) VerticLuvisol (3.9) EutricNitosol (23.8) Eutric Regosol (9.2) EutricNitosol (23.8)
The scale effect on runoff coefficient was studied in relation to the relative extent of representation of runoff plot characteristics in a watershed under the situation where the watershed heterogeneous The analysis relied on the separate and combined effects of slope, soil, and land use of runoff test plots and their extent of representing the particular watershed. Each of the three characteristics in each test plot was assessed in terms of their relative percentage area in the particular watershed. Specific characteristics dominantly affecting the spatial scale effect were analyzed. The area-weighted representation was analyzed and resulted in area-weighted new Q and RC at watershed scales. The areaweighted runoff coefficient was then compared with the average runoff coefficient estimated at the outlet of the same watershed. This analysis helps to further explain the watershed heterogeneity approach by considering soil, slope, and land use characteristics of a plot and the respective watershed. Results were summarized using the scaling factor (SF). Scaling factor is a ratio of runoff coefficient of the differently sized watersheds (Heras, Nicolau, Martín, & Wilcox, 2010). It helps to analyze the influence of varying surface conditions on modifying the effect of spatial scale on runoff. Moreover, scaling factor is an important conversion factor to develop similarity between characteristics of one system (runoff test plot) to the corresponding characteristics of the other system (watershed) (Woods, 2016). Therefore, the scaling factor is the ratio of weighted RC of a watershed divided by observed RC at the watershed outlet.
AnditTid
Maybar
23% (11) 39% (65) 48% (65) 48% (65) Cultivated (34.61) Cultivated (34.61) Grazing (17.2) Cultivated (34.61) EutricRegosol (32.4) Chromic Cambisol (1) EutricRegosol (34.61) OchricAndosol (66)
16% (20) 64% (36) 43% (38) 37% (38) Cultivated (60) Grass (20) Grass (20) Cultivated (60) Deep Phaeozem (14) PhaeozemLithosols (11) PhaeozemLithosols (40) HaplicPhaeozems (12)
3. Results and discussion
practices and topographic conditions prevalent to specific plots and watersheds. It also indicates the mechanism and trend of runoff generation throughout the rainfall season. The comparison of rainfall and runoff can quantify the relative response of plot conditions. The relationship between rainfall and runoff at plot and watershed levels are illustrated in Fig. 2. The total watersheds’ fifteen-day aggregated rank-order rainfall-runoff dataset fitted a polynomial function showed a good matching with R2 of 0.93, 0.99, and 0.99 in Anjeni, Andit Tid, and Maybar, respectively. While, the pooled plots’ 15-day rainfall-runoff dataset showed better matching with R2 of 0.99, 0.99, and 0.97 in Anjeni, Andit Tid, and Maybar, respectively. The overall watershed and plots observed runoff coefficients (RCobs) were correlated positively and strongly with r-values of 0.98, 0.87, and 0.98 respectively for Anjeni, Andit Tid, and Maybar (Fig. 3). Moreover, the runoff and runoff coefficients of plots and the respective watersheds were found significantly different (P o 0.01). The observed runoff and runoff coefficient showed a consistent increment from plot to watershed scales at Anjeni and Andit Tid watersheds except in Maybar. Despite the great variation in the soil and agro-ecological conditions, the average observed runoff coefficients were more or less similar across the three watersheds (28–38%). At plot scale, the average runoff coefficient was 45.52%, 48.67%, and 14.58%, respectively at Anjeni, Andit Tid, and Maybar. Similar results in RC, 20–50% (Gizaw, 2003) in Andit Tid, and 27% (Hurni, Kebede, & Gete, 2005) and 10–32.4% (Gizaw, 2003) in Maybar, and 45.2% in Anjeni (Nyssen et al., 2010) as cited in Boshart (1998a) were reported. Runoff coefficients estimated in all sites were smaller than those in Liu et al. (2008), Engda et al. (2012), and Lemann, Roth, and Gete (2015).
3.1. Rainfall-runoff relationship
3.2. Scale effect on runoff coefficient in heterogeneous watersheds
Analysis of rainfall-runoff relationship is useful to understand the hydrologic conditions and the response of land management
The measured runoff coefficients of runoff plots are explained by slope, soil and land use characteristics. This measured runoff
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
5
Fig. 2. Rank-order rainfall (P) and runoff (Q) data relationship at watersheds and runoff test plots.
coefficient of runoff plots further extrapolated into watershed level runoff coefficient based on the extent of representation of each characteristic of individual runoff plots in a watershed. The extrapolation of runoff coefficient was then computed by weighting the runoff coefficient of runoff plots using the relative proportional area coverage of each characteristic of plots from the total watershed. The extent of representation of plot characteristics (slope, soil and land use/cover) to the particular watershed were analyzed using the general information from Tables 1 and 2. In the three watersheds, depending on the size of the watershed, initial setup of runoff plots and change in land use/cover, the representation of each watershed by the characteristics of the four test plots was different. The magnitude of representation of each watershed by the characteristics of four runoff plots varies in the three watersheds. Of the total area of the Anjeni watershed, 94% of the land use/cover, 57% of the slope, and 37% of the soil type are represented by the four runoff plots. Whereas in the Andit Tid
watershed, 52% of the land use/cover, 76% of the slope, and100% of the soil, are represented by the runoff plots. Likewise, in the Maybar watershed, 80% of the land use/cover, 95% of the slope, and 76% of the soil were represented by its plots. This shows the whole watershed area was not represented by the four plots. Therefore, an intermediate runoff coefficient values were estimated and assigned for those slope, land use/cover and soil classes that are not represented by the four runoff plots. But, the larger extents of representations of watersheds by the plots seem to reflect enhanced transferability of information from the runoff plot to the watershed. The non-uniform representation of slope, land use/cover and soil type will be discussed in terms of the relative contribution of the factors to the runoff coefficient. Nevertheless, the three watersheds were not equally represented and thus results were interpreted independently (Table 3). Based on the extent of plot characteristics in the watersheds, area-weighted runoff coefficients were estimated and presented in Table 4. The relative influence of the sources of variation in the
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
6
Fig. 3. The relationship between watershed and plot scale runoff coefficients.
Table 3 Observed runoff and runoff coefficient at watershed and plots.
Anjeni watershed Anjeni Plot average Anjeni-TP-1 Anjeni-TP-2 Anjeni-TP-3 Anjeni-TP-4 Andit Tid watershed Andit Tid Plot average AnditTid-TP-1 AnditTid-TP-2 AnditTid-TP-3 AnditTid-TP-4 Maybar watershed Maybar Plot average Maybar-TP-1 Maybar-TP-2 Maybar-TP-3 Maybar-TP-4
P (mm)
Qobs (mm)
RCobs (%)
142.03 136.67 136.94 133.50 137.85 138.37 106.00 116.09 114.05 113.33 114.11 122.87 117.57 130.01 130.09 131.11 129.07 129.76
42.82 62.08 67.27 75.39 43.79 61.86 40.27 56.19 56.53 70.41 53.73 44.09 33.29 18.94 27.31 10.12 16.03 22.28
30.15 45.52 49.12 56.47 31.77 44.70 38.00 48.67 49.57 62.13 47.09 35.88 28.32 14.58 21.00 7.72 12.42 17.17
*P and Q are averages of 15-day aggregated rainfall and runoff.
runoff coefficients was analyzed under heterogeneous surface characteristics. The weighted observed RC of the watersheds was found at 48.6%, 43.1%, and 14.4% for Anjeni, Andit Tid, and Maybar respectively. There were only slightly higher weighted RCs for Anjeni and Andit Tid watersheds. The weighted RCs in relation to slope, land use/cover and soil type representation were 51%, 49.2%, and 45.5%, at Anjeni; 48%, 41%, and 40.4% at Andit Tid and 13.9% 15.5%; and 13.7% at Maybar watersheds. In general, the relative contribution of factors to the magnitude of runoff coefficient is in the order of slope followed by land use/cover and then soil type. The scaling factor of the runoff coefficient of area-weighted observed RC to the observed RC at watershed scale was 1.61, 1.13, and 0.51 respectively for Anjeni, Andit Tid, and Maybar. Overestimation
of weighted RC at Anjeni and Andit Tid and underestimation at Maybar compared to the measured RC might arise due to lack of complete watershed representation by the four runoff plots and the relative location specific importance of plot characteristics. Cultivated areas in Anjeni and Maybar watersheds cover 80% and 60%, respectively. These extents were consistent with the order of extent of representativeness of plots’ characteristics signifying the influence of land use/cover (Li & Wang, 2010). The estimated area-weighted runoff coefficient increased for runoff plots’ slopes up to 50% and reduced for steeper slopes in Andit Tid and Maybar watersheds. Similar trends for the same ranges of slopes were reported (Patin et al., 2012). However, the slope-runoff coefficient trend in Anjeni was found to be random across runoff plots. In Anjeni, about 65% of the area is covered with deep to moderately deep, drained to moderately drained soils (SCRP, 2000a). In Andit Tid most of the areas are covered with shallow, stony, up to 2 m deep and excellent water storing capacity soils (SCRP, 2000b). About 75% areas of Maybar was steep to very steep, covered by shallow (average of 15 cm) soils that are excessively drained and well structured (SCRP, 2000c). The remaining, from gentle to the flat valley area, is covered with deep to very deep and poorly drained soil. In these flatter and valley floor areas, the water table can reach from 20 to 50 cm. According to Bayabil, Tilahun, Collick, Birru, and Steenhui (2010), and Engda et al. (2012), in Maybar watershed rainfall infiltrated into large steep areas and reached the downslope area through interflow resulted in the rise of the water table. As explained earlier, runoff test plots share similar characteristics with their watersheds reflecting lower runoff coefficient in the watershed than the plots. In spite of the average 83.67% representation of runoff plots in Maybar, much of the runoff is generated in the quarter saturated valley floor, and closer to the stream area. This area is not represented by the runoff plots. This situation substantiates the influence of spatial scale on runoff generation (Li & Wang, 2010). Besides, it necessitates further investigation of relationships between the spatial distribution of
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
7
Table 4 Representation of plots to the watershed (SF) in slope, soil, and land use characteristics. Plot features
Slope Land use Soil type Weighted Avg. RCobs Watershed Avg. RCobs
Anjeni
Andit Tid
Maybar
Weighted RCobs (%)
SFobs
Weighted RCobs (%)
SFobs
Weighted RCobs (%)
SFobs
51.03 49.26 45.55 48.6 30.15
1.69 1.63 1.51 1.61
47.98 41.01 40.36 43.10 38.0
1.26 1.08 1.06 1.13
13.86 15.48 13.73 14.4 28.3
0.49 0.55 0.59 0.51
SF is Scaling Factor.
plots and potential runoff generating areas within the watershed. In Anjeni and Andit Tid, plots are situated where much of the runoff is produced. For example, unlike in Maybar, 60% of the area in Anjeni contributes to base flow and the runoff is not generated in some specific part (Elias, 2009). This situation dictates that runoff generation is not affected by topography rather by soil type or land use. In Maybar, although the three surface characteristics represented nearly in equal share, much of the runoff is generated on specific topographic positions. That topographic position is the small valley floor where the runoff coefficient of the watershed is determined. The variability of runoff coefficient in plots within a watershed revealed that the variation is associated with the probable differences in the opportunity of flow to move extra distance and infiltrate into the ground (Gomi et al., 2008). Consequently, research results showed that variations in runoff coefficient are implications of lost/gained opportunities for transfer of flow through differences in soil, land use, and slope. Rainfall-runoff process in a watershed is governed by lots of known and unknown factors requiring special reasoning for special responses (Ramana, 2014; Shi, Chen, Fang, Qin, & Cai, 2009). The connectivity of surface flow is directly linked to the dominant surface characteristics i.e., soil, vegetation (land use), and slope conditions that control the hydrological response (Gomi et al., 2008; Lin & Wang, 2010). Unless interventions are proposed, designed, and implemented in reference to the proper transfer of hydrological information from small scale to larger scale, negative externalities could emerge within and out of the area intervened. This is because that result from small-scale experiments usually considers only part of the dominant hydrologic processes. As such, recommended measures could not have a proportional effect on the watershed level. Although scale effect is related to the scale of interest, undermining its influence that impacts runoff generation capacity of an area could cause problems reaching to downstream impacts (Hunink, Droogers, Kauffman, Mwaniki, & Bouma, 2012). To promote watershed development through up-scaling of BSWCPs to the areas where they can best fit with minimized negative externalities, scaling factor (SF) needs inferences in hydrologic processes. The dramatic change in runoff coefficients resulted because of the high correlation of surface characteristics with runoff coefficient at plot scale than at watershed scale (Merze et al., 2006). In cases where pronounced variability in surface condition between small and large areas occurred, the customary consistent pattern of runoff generation (decreasing runoff with increasing area) could be altered (Gomi et al., 2008). This inverse relation between runoff and drainage area is prevalent in humid and sub-humid areas (Stone et al., 2008) with the exception of watersheds like Maybar. Therefore, the results of this study dictate that assessment should be done at watershed scale or else runoff plot results are used with appropriate scaling factors. Otherwise, under or overestimation of extrapolation of runoff in plots experiments might lead to erroneous management decisions (Leys et al., 2010).
4. Conclusion and recommendation The purpose of the study was to analyze the effects of spatial scale on hydrological processes amongst which runoff coefficient is of critical interest. Long-term rainfall and runoff data from three different watersheds and four equal sized runoff plots in each of the watershed were used for the analysis. In all the three study sites, the runoff coefficient scaling factor from runoff plot to the watershed was determined. Area-weighted runoff coefficients estimated by considering the heterogeneity of watersheds represented by the extent of plot characteristics (slope, land use/cover and soil type). The results showed that runoff coefficient was reduced from plot to the watershed in Anjeni and Andit Tid, while it was the reverse in Maybar. Lower runoff coefficient at the plot than watershed values in Maybar challenges the common results of high runoff coefficient from a small area and low runoff coefficient from the larger area. Thus, taking the variable response of spatial scale on the runoff coefficient into account, the proportional effect of small plots to large watersheds is not always expected. Therefore, extrapolation of results from a small area to the enclosing larger area for up-scaling purposes shall be done with caution. The scale of representativeness of plots to the particular watershed with marked variability and the combined effect of surface characteristics need to be areas of future research by considering local conditions. Furthermore, it is necessary to investigate relationships between the spatial location of plots and the potential runoff generating areas within a watershed.
Acknowledgments The thoughtful supply of raw data and information was from Water and Land Resources Center of Ethiopia. Dr. Daniel Bekele from Melkasa Agricultural Research Center has helped on the manuscript structure and Amsalu Weldei (Ph.D. candidate) provided me support in editing the English.
References Abebe, S., Hurni, H., & Gete, Z. (2013). Long-term changes in soil-based ecological services at three sites in Ethiopia. Journal of Ecology and the Natural Environment, 5(8), 172–180. Abrham, T. G., Tilashwork, C., Tesfaye, F., & Abdlesemed, J. (2016). Impact assessment of land use /change on soil erosion and rural livelihood in Andit Tid watershed, North Shewa, Ethiopia. Asia Pacific Journal of Energy and Environment, 2, 125–132. Ajmal, M., Khan, T. A., & Kim, T. (2016). A CN-based ensembled hydrological model for enhanced watershed runoff prediction Muhammad. Water, 8, 20. Asadzadeh, F. M., Gorji, A., Vaezi, Sokouti R., & Shorafa, M. (2012). Scale effect on runoff from filed plots under natural rainfall. American-Eurasian Journal of Agriculture and Environmental Science, 12(9), 1148–1152. Bayabil, H. K., Tebebu, T. Y., Stoof, C. R., & Steenhuis, T. S. (2015). Spatial and temporal runoff processes in the degraded Ethiopian Highlands: The Anjeni Watershed. Journal of Hydrology and Earth System Sciences, 12, 4387–4411. Bayabil, H. K., Tilahun, S. A., Collick, A. S., Birru, Y., & Steenhui, T. S. (2010). Are runoff processes ecologically or topographically driven in the (sub) humid Ethiopian
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i
8
T.M. Lemma et al. / International Soil and Water Conservation Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎
highlands? The Case of the Maybar Watershed. Ecohydrology(3), 457–466. Binyam, B. (2009). SWAT to identify watershed management options: (Anjeni Watershed, Blue Nile Basin, Ethiopia) (M.Sc. Thesis). USA: Cornell University. Bosshart, U. (1996). Measurement of river discharge for the SCRP research catchments: Methodology and theoretical background. Switzerland: University of Berne (Research Report 29). Cerdan, Olivier, Le Bissonnais, Yves, Govers, Gerard, Lecomte, V., Oost, Kristof, Couturier, A., … Dubreuil, N. (2004). Scale effect on runoff from experimental plots to catchments in agricultural areas in Normandy. Journal of Hydrology, 299, 4–14. http://dx.doi.org/10.1016/j.jhydrol.2004.02.017. Elias, S., L. (2009). Modeling rainfall-runoff relationships for the Anjeni Watershed in the Blue Nile Basin (M.Sc. Thesis). USA: Cornell University. Engda, T., A., Bayabil, H., Legesse, E., Ayana, E., Tilahun, S., Collick, Amy S., … Tammo, S. Steenhui (2012). Watershed hydrology of the (Semi) humid Ethiopian highlands In: Nile River Basin (pp. 145–162). Springer Science. EWRI (Environmental and Water Resources Institute) (2009). Curve number hydrology: State of the practice. Alexander Bell Drive Reston, Virginia, USA: American Society of Civil Engineers. Feng, P., & Li, J. Z. (2008). Scale effects on runoff coefficient in mesoscale and largescale sub-basins in the Luanhe River Basin. Journal of Hydrology and Earth System Sciences, 5, 1511–1531. Gebirrye, T. D. (2004). Investigating scale, rainfall-runoff sequences and bmp effects on phosphorus, runoff and sediment yield (Doctoral dissertation). USA: Oklahoma State University. Giovanoli, L. (2014). Linking land use mapping and runoff estimation in the Gerda Watershed in the Ethiopian Highlands (M.Sc. Thesis). Switzerland: University of Bern. Gizaw, D. (2003). Estimation of runoff coefficients at different growth stages of crops in the highlands of Amhara Region (M.Sc. Thesis). Ethiopia: Alemaya University. Gomi, T., Sidle, R. C., Miyata, S., Kosugi, K., & Onda, Y. (2008). Dynamic runoff connectivity of overland flow on steep forested hillslopes: Scale effects and runoff transfer. Water Resources Research, 44, W08411. http://dx.doi.org/ 10.1029/2007WR005894. Guzman, C. D., Tilahun, S. A., Zegeye, A. D., & Steenhuis, T. S. (2013). Suspended sediment concentration-discharge relationships in the (sub-) humid Ethiopian highlands. Journal of Hydrology and Earth Science System, 17, 1067–1077. Hawkins, R. H., Ward T. J., Woodward, D. E., and Mullen J. A. (2010). Continuing Evolution of Rainfall-Runoff and the Curve Number. 2nd Joint Federal Interagency Conference, June 27 - July 1, 2010, Las Vegas, NV USA. Heras, M. M., Nicolau, J. M., Martín, L. M., & Wilcox, B. P. (2010). Plot-scale effects on runoff and erosion along a slope degradation gradient. Water Resources Research, 46, W04503. http://dx.doi.org/10.1029/2009WR007875. Hughes, D. A., Kapangaziwiri, E., & Tanner, J. (2013). Spatial scale effects on model parameter estimation and predictive uncertainty in ungauged basins. Journal of Hydrology Research, 44(3), 441–453. http://dx.doi.org/10.2166/nh.2012.049. Hunink, J. E., Droogers, P., Kauffman, S., Mwaniki, B. M., & Bouma, J. (2012). Quantitative simulation tools to analyze up- and downstream interactions of soil and water conservation measures: Supporting policymaking in the Green Water credits program of Kenya. Journal of Environmental Management, 111, 187–194. Hurni, H., Kebede, Tato, & Gete, Zeleke (2005). The implications of changes in population, land use, and land management for surface runoff in the upper Nile Basin Area of Ethiopia. Mountain Research and Development, 25(2), 147–154. Junker, L. (2012). Evaluation of the AnnAGNPS model used for river discharge and sediment yield predictions in a microscale catchment in the Highlands of Ethiopia (Master of Science Thesis). Switzerland: University of Berne. Lemann, T., Roth, V., & Gete, Z. (2015). Impact of precipitation and temperature on hydrological responses in the Ethiopian highlands. Hydrological Sciences, 62(2), 270–282. Leys, A., Govers, G., Gillijns, K., BeRCkmoes, E., & Takken, I. (2010). Scale effects on runoff and erosion losses from arable land under conservation and conventional tillage: The role of residue cover. Journal of Hydrology, 390(2010), 143–154. Lin, Y., & Wang, G. X. (2010). Scale effect on runoff in alpine mountain catchments on China's Gongga Mountain. Hydrology Earth System Science Discussion, 7, 2157–2186. Liu, B. M., Collick, A. S., Gete, Z., Enyew, A., Easton, Z. M., & Steenhuis, T. S. (2008). Rainfall-discharge relationships for a monsoonal climate in the Ethiopian Highlands. Journal of Hydrological Processes, 22, 1059–1067. Meher, J. (2014). Rainfall and runoff estimation using hydrological models and ANN techniques (Doctoral Dissertation). India: National Institute of Technology Rourkela. Merz, R., Blö schl, G., & Parajka, J. (2006). Spatio-temporal variability of event runoff coefficients. Journal of Hydrology, 331, 591–604. Mounirou, L. A., Hamma, Y., Harouna, K., Jean-Emmanuel, P., & Gil, M. (2012). Measuring runoff by plots at different scales: Understanding and analyzing the sources of variation. Hydrology and, Environment(20), 865–890. Nyssen, J., Clymans, W., Descheemaeker, K., Poesen, J., Vandecasteele, I., VanmaeRCke, M., & Walraevens, K. (2010). Impact of soil and water conservation measures on watershed hydrological response – a case in North Ethiopia. Journal of Hydrological Processes, 24, 1880–1895. Patin, J., Mouche, E., Ribolzi, O., Chaplot, V., Sengtahevanghoung, O., Latsachak, K. O., & Valentin, C. (2012). Analyses of runoff production at the plot scale during a
long-term survey of a small agricultural catchment in Lao PDR. Journal of Hydrology, 426–427(2012), 79–92. Penna, N., D. Mantese, N., Gobbi, A., & Borga, M. (2011). Runoff response at different spatial scales: Moving from small experimental areas to mesoscale watersheds. Research Support, Innovation & Technology Transfer, 62, 1–4. Prats, S. A., Wagenbrenner, Joseph W., Martins, Martinho A. S., Malvar, Maruxa Cortizo, & Keizer, Jan Jacob (2015). Hydrologic implications of post-fire mulching across different spatial scales. Land Degradation and Development, 27, 1440–1452. Raghunath, H. M. (2006). Hydrology: Principles. Analyses. Design. Revised ((2nd ed.). New Delhi, India: New Age International (P) Limited, Publishers. Ramana, G. V. (2014). Regression analyses of rainfall and runoff process of a typical watershed. International Journal of Sciences and Applied Information Technology, 3(1), 16–26. Sadeghi, S. R., Gholami, L., Sharifi, E., Khaledi, A. D., & Homaee, M. (2015). Scale effect on runoff and soil loss control using rice straw mulch under laboratory conditions. Solid Earth, 6, 1–8. SCRP (Soil Conservation Research Program) (1982). SCRP – First Progress Report (the year 1981)Volume 2. University of Bern, Switzerland, in association with The United Nations University, Tokyo, Japan. SCRP (Soil Conservation Research Program) (1984). Compilation of phase –I progress Reports (the year 1981, 1982, and 1983) – by Hans Hurni – 1984. The University of Bern, Switzerland in association with The United Nations University, Tokyo, Japan. SCRP (Soil Conservation Research Program) (2000a). Area of Anjeni, Gojam, Ethiopia: Long-term Monitoring of the Agricultural Environment 1984 – 1994. Soil Conservation Research Program (SCRP). Center for Development and Environment, University of Berne, Switzerland, in Association with the Ministry ofAgriculture, Ethiopia. SCRP (Soil Conservation Research Program) (2000b). Area of Andit Tid, Shewa, Ethiopia: Long-term Monitoring of the Agricultural Environment 1982 – 1994. Soil Conservation Research Program (SCRP). Center for Development and Environment, University of Berne, Switzerland, in Association with the Ministry of Agriculture, Ethiopia. SCRP (Soil Conservation Research Program) (2000c). Area of Maybar, Wello, Ethiopia: Long-term Monitoring of the Agricultural Environment 1981 – 1994. Soil Conservation Research Program (SCRP). Center for Development and Environment, University of Berne, Switzerland, in Association with the Ministry ofAgriculture, Ethiopia. Shi, Z., Chen, L., Fang, N., Qin, D., & Cai, C. (2009). Research on the SCS-CN initial abstraction ratio using rainfall-runoff event analyses in the three Gorges area, China. Catena, 77, 1–7 (2009). Soulis, K. X., & Valiantzas, J. D. (2012). Identification of the SCS-CN parameter spatial distribution using rainfall-runoff data in heterogeneous watersheds. Journal of Water Resource Management, 27, 1737–1749. Stewart, R. D., Liu, Z., Rupp, D. E., Higgins, C. W., & Selker, J. S. (2015). A new instrument to measure plot-scale runoff. Geoscientific Instrumentation Methods and Data System, 4, 57–64. Stone, J. J., Nichols, M. H., Goodrich, D. C., & Buono, J. (2008). Long-term runoff database, Walnut Gulch Experimental Watershed, Arizona, United States. Journal Water Resources Research (p. 44), 44. Tebebu, T. Y., Steenhuis, T. S., Dagnew, D. C., Guzman, C. D., Bayabil, H. K., Zegeye, A. D., & Tilahun, S. A. (2015). Improving efficacy of landscape interventions in the (sub) humid Ethiopian highlands by improved understanding of runoff processes. Frontiers of Earth Science, 3, 49. Tegenu, A., E. (2009). Modeling rainfall, runoff and soil loss relationships in the northeastern highlands of Ethiopia, Andit Tid Watershed (M.Sc. Thesis). USA: Cornell University. Temesgen, G. (2015). The implications of watershed management for reversing land degradation in Ethiopia. Research Journal of Agriculture and Environmental Management, 4(1), 5–12. Woods, R. (2016). The Search for Catchment -Scale Physics: Spatial Scaling & Similarity with Hydrological Heterogeneity. Symposium in Honor of Eric Wood: Observations and Modeling across Scales, June2-3, 2016 University of Bristol, Princeton, USA. Woubet, G. Alemu, Tadele, A., Birru, Y., Selassie, Yihenew G., Wolfgramm, Bettina, & Hurni, H.( (2013). Impacts of soil and water conservation on land suitability to crops: The case of Anjeni watershed, Northwest Ethiopia. Journal of Agricultural Science, 5(2). Wu, J., Jones, K. B., Li, H., & Loucks, O. L. (2006). Scaling and Uncertainty Analysis in Ecology: Methods and Applications ((1st ed.). . Springer (Printed in the Netherlands). Yair, A., & Raz-Yassif, N. (2004). Hydrological processes in a small arid catchment: Scale effects of rainfall and slope length. Geomorphology, 61(1), 155–169. Yakob, Mohammed (2009). Climate change impact assessment on soil water availability and crop production in Anjeni Watershed Blue Nile Basin (M.Sc. Thesis). Ethiopia: Arbraminch University. Yildiz, O., & Barros, A. P. (2009). Evaluating spatial variability and scale effects on hydrologic processes in a midsize river basin. Scientific Research and Essay(4), 217–225.
Please cite this article as: Lemma, T. M., et al. International Soil and Water Conservation Research (2018), https://doi.org/10.1016/j. iswcr.2018.08.002i