DeepDyve requires Javascript to function. Please enable Javascript on your browser to continue.
Long-term trends of river flow, sediment yield and crop productivity of Andit tid watershed, central highland of Ethiopia Long-term trends of river flow, sediment yield and crop productivity of Andit tid watershed,...
Woldemarim, Ayele; Getachew, Tilahun; Chanie, Tilashwork
2023-12-31 00:00:00
ALL EARTH 2023, VOL. 35, NO. 1, 3–15 https://doi.org/10.1080/27669645.2022.2154461 Long-term trends of river flow, sediment yield and crop productivity of Andit tid watershed, central highland of Ethiopia a a b Ayele Woldemarim , Tilahun Getachew and Tilashwork Chanie a b Department of soil and water management, Debre Brihan Agricultural Research Center, Debre Brihan, Ethiopia; Department of soil and water management, Adet Agricultural Research Center, Adet, Ethiopia ABSTRACT ARTICLE HISTORY Received 27 July 2022 Andit tid watershed is part of Blue Nile basin located in the central highlands of Ethiopia. The Accepted 29 November 2022 lack of data and information at watershed level resulted in different conclusions from trend studies of river flow, sediment yield and crop productivity at a basin scale. There is an KEYWORDS opportunity to improve water and land if it can be underpinned by a better scientific under- River flow; sediment yield; standing of trends of flow, sediment yield and crop production at the basin level. This research crop production; trends; is carried out using descriptive statistics, Mann–Kendall (MK) and Pettit’s test to determine the Mann–Kendall test potential trends of river flow, sediment yield and crop productivity using Andit tid watershed case. The result showed that there was high variability of interannual river flow with CV >30%. The Pettitt test showed a significant abrupt change in monthly (March, July, August, September and October) and seasonal (summer and winter) river flow. The Pettitt test result of sediment yield and crop production showed no change. MK test showed a significant (P < 0.05) decreas- ing trend in March, August, September and October river flow. The other MK values showed no significant trends for all parameters. Researchers should consider representative watershed- based information and data for the analysis and interpretation of large basins. studies that were conducted in the same year resulted 1. Introduction in different trends and variability output. Nile (‘Abbay’ in its local name) is the longest river in the Studies that estimate annual sediment load from world and one of the most water-limited basins. Eighty- the Upper Blue Nile basin also reported different sedi- five percent of the total amount of water entering Lake ment yield results. For instance, as reviewed by Nasser at the Aswan dam originates from the Ethiopian Gebremicael et al. (2013) the annual sediment yield Highlands (Sutcliffe & Parks, 1999). The Upper Blue Nile of the Blue Nile basin ranged from 111 × 106 to River basin which contributes over 60% of the Nile’s 140 × 106 tons/year. To minimise this type of research water (Conway, 2000) is crucial for the socio-economic output difference, basin-representative, experimental development and environmental stability of Ethiopia, watersheds offer essential knowledge in recognising Sudan and Egypt. These countries have experienced the hydrological and erosive processes including serious problems in their storage reservoirs and irriga- trends of crop production. The only long-term river tion canals due to excessive sediment loads (Betrie et al., flow, sediment and crop yield data monitoring has 2011). Seleshi et al. (2011) did research at the border of been carried out in the small Soil Conservation Sudan and reported that sediment concentrations were Research Project (SCRP) watersheds which were estab- −1 as high as 12.3 g L . In this circumstance, studying the lished in 1981 and conserved with soil and water con- trend and variability of river flow and sediment yield is servation structures in the upper reaches of large crucial to exactly put measures to lengthen the lifespan basins (Steenhuis et al., 2014). For example, discharge of reservoirs and canals. assessment at the watershed outlet is the standard The long-term trend analysis of runoff in the Blue Nile analysis and forms the basis for the development of basin was studied by many scholars, for instance, many fundamental theories of runoff (Blume et al., Gebrehiwot et al. (2010), Kebede (2009), Legesse et al. 2007; Erturk, 2010). Andit tid watershed is one of the (2003) and Tesemma et al. (2010). However, the conclu- SCRP watersheds, which is representative of the upper sions of these studies have not shown a common con- Blue Nile basin. In this watershed, two rivers (Gudi sensus on the trends and variability of flow. Legesse et al. Bado and Wani Gedel) confluence 150 m above the (2003) reported an increasing trend; Tesemma et al. gauging station and create the river called ‘Hulet (2010) reported no change and Kebede (2009) and Wenz’. In the gauging station of this river, daily river Gebrehiwot et al. (2010) reported a decreasing trend of flow and event-based sediment samples data have the annual flow of the Blue Nile basin. Furthermore, been collected since July 1982. CONTACT Ayele Woldemarim ayeledesalegn5@gmail.com Debre Brihan Agricultural Research Center, Po Box 112, Debre Brihan, Ethiopia © 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 4 A. WOLDEMARIAM ET AL. In Ethiopian highlands including Andit tid for further studies and for deciding other watershed- watershed in the upper Blue Nile basin, soil erosion based decisions rather they used simulated model and sedimentation become the major threat to crop output. production by washing the top fertile soil. The hillside and mountainous nature, and shape of this watershed 2. Methodology foster the formation of flooding and soil erosion. Numerous studies have been conducted in Andit tid 2.1. Description of the watershed over the years, including ones on suitability mapping Andit tid watershed is situated at 39°43’ E longitudes for major crops (Yohannes & Soromessa, 2018); land and 9°48’ N latitudes apart 180 km northeast of the capability classification (Yohannes & Soromessa, 2019); capital city of Ethiopia, Addis Ababa (Figure 1). The land use and land cover change impact on livelihoods altitude of the catchment ranges between 3040 (at and soil erosion (Abrham Tezera et al., 2016); soil ero- the outlet) to 3550 (at the elevated area of the sion estimation and mapping (Desalegn et al., 2018); watershed) m.a.s.l. The mean annual rainfall is 1585.2 comparison of CFSR and conventional weather data mm, and the minimum and maximum temperatures (Roth & Lemann, 2015) and rainfall-runoff modelling are 7°C and 17°C, respectively. The minimum and max- (Engda, 2009). While the very crucial watershed char- imum soil surface temperatures are 8°C and 20°C, acteristics of how are the monthly, seasonal and respectively. The agro-climatic zone of the watershed annual trends of the river flow and sediment yield is Wet Dega/Wet high Dega. The dominant soil types of were not addressed. Around 50 fixed plots in different the watershed are Humic and ochric Andosols, positions of bunds were set up in the watershed to Fluvisols, Regosols and Lithosols. It is characterised by assess the effects of soil and water conservation on severe soil degradation, especially in the lower part of agricultural productivity, and for the past 24 years, we the catchment. Soil fertility is limited through low pH have been gathering data on crop yield and agronomic and N- and P-deficiency. Smallholder mixed farming parameters from these plots. The effects of these soil systems including grain production (barley) and Ox- and water conservation structures particularly Fanya plough farming oriented with free grazing practices is juu and soil bunds on agricultural yield have not been the major practice of the watershed community. thoroughly studied. The productivity trends of the major crops grown in the watershed were not also studied. Studying these issues is therefore anticipated 2.2. River flow and sediment yield data collection to increase the watershed community’s understanding of the importance of soil and water conservation for The river gauging stage was monitored continuously increasing crop productivity, preparing for seasonal using a limnigraph accompanied by manual water- changes in peak river flow, and using various mulching level measurements during storm events. The stage techniques or green cover in the early rainy seasons to height (water level) which was monitored using limni- reduce bulk erosion. As an objective, this study was graph and the manual ruler was converted into dis- focused to (i) analyse the long-term trend and varia- charge using the following equation (Bosshart, 1997): bility of runoff, sediment yield and crop productivity of 2:749 QðH< ¼ 67Þ ¼ 0:03� H (1) Andit tid watershed using statistical methods and (ii) characterise the watershed based on river flow, sedi- 1:35 Qð67< H � 250Þ ¼ 10:846� H (2) ment yield and crop production using the data from 1994 to 2017. where Q is the runoff discharge in l/s and H is the true In this paper, we addressed the monthly, seasonal water level (height of the stage) in cm. and annual trends of runoff and sediment loss and Every 10 min during runoff events, one-litre grab the year-to-year changes in crop productivity in the samples were taken to measure the amount of sedi- watershed. Therefore, this study tells how the river ment immediately when the colour of the water turned flow, sediment yield and crop production have chan- brown. The sampling rate was reduced to 30 min and ged over time in the watershed; whether or not these hourly intervals once the water level dropped and the parameters are increasing or decreasing. From the colour returned to light brown. The overall stream flow result of this study, researchers can go further about and an estimate of the suspended material carried by the factors causing any abrupt happened in all studied the flow at that particular time interval were calculated parameters. It can be used as a ground truth reference using sediment samples and manual measurements of to evaluate the model simulation outputs of different the river’s water level. By oven-drying the one-litre studies that have been conducted in the Blue Nile samples and weighing the oven-dried soil, the quantity basin. Policymakers can use the result of this research of sediment load within the sample was determined. as the benchmark for making appropriate land man- The sum of the total water flow per time and the agement decisions, improving local land productivity sediment concentration as obtained from the and enhancing livelihood. The researchers can use it 1-L sample was then multiplied to get the total soil ALL EARTH 5 Figure 1. The location map of the study watershed. loss for that measurement period. Suspended sedi- tillage, predecessor crops and crop types was included. ment concentration was also determined by dividing Zone A (above the terrace or zone of deposition), Zone the weight of dry sediment by the volume of water B (between terraces) and Zone C (below terraces or (Miller et al., 2015; Womber et al., 2021). zone of transportation) were the locations from which For the seasonal analysis, the data have been samples were taken. The reason for different positions divided into four seasons based on the local situation, sampling was to identify the impact of soil conserva- which are winter (Bega) (December–February), Belg tion on crop production. The samples were taken from (March–May), summer (kiremt) (June–September) and 4 m of land for each plot position and we extrapolate spring (Tsedey) (October–November). The collected to kg/ha. The crop types, crop management and time time-series data have some missing values. The miss- of cropping are decided by the land owners (farmers). ing value is interpolated during data processing. We In this data, all the crop types may not be sown in the used the Auto-Regressive Integrated Moving Average fixed and non-fixed plots all over the study period (ARIMA) function to fill in the missing data. ARIMA (1994–2017). As a result, we only looked at the data model is considered a powerful and extensively used from a single crop’s growing years when analysing the statistical tool to analyse and predict time-series data. trend in crop production. This information was primar- The main advantages of the model are that it can ily used to assess how crop production in the detect seasonal changes and consider serial correlation watershed would be affected by soil and water con- within the time series (Yurekli et al., 2007). servation and other new agronomic practices. 2.3. Crop yield data collection 2.4. Homogeneity test of the data Crop yield samples were taken from 35 fixed and 50 The Pettitt test (Pettitt, 1979) was chosen to detect non-fixed plots located throughout the watershed dur- inhomogeneity in the time-series data. This test ing each cropping season. During samplings, informa- detects shifts in the average and calculates their sig- tion on crop management, inputs, soil depth, slope, nificance (Liu et al., 2012) in a hypothesis test. In the 6 A. WOLDEMARIAM ET AL. Pettitt test, the null hypothesis stated that the data are Xi is taken as a reference point which is compared with homogeneous, as against the alternative hypothesis the rest of the data point’s Xj so that tells that the data have abrupt change. The empirical If S > 0, then later observations in the time series significance level (p-value) was computed using tend to be larger than those that appear earlier in the XLSTAT 2020 v.3. In this study, this test was performed time series and it is an indicator of an increasing trend, at a significance level of 5%. while the reverse is true if S < 0 and this indicates a decreasing trend. Under the null hypothesis of no trend, the statistic S follows an approximately normal distribution with 2.5. Mann–Kendall (MK) test mean zero and variance (Kendall, 1975) statistic is Mann–Kendall trend test is an extremely important given as parameter for watershed modelling, and studying nðn 1Þð2nþ 5Þ t ðt 1Þð2t þ 5Þ 1 1 1 catchment characteristics which are very important to t¼1 varðSÞ ¼ determine water resources planning strategies in the (5) long term for any region (Kothawale & Rupa Kumar, 2005). The application of the Mann–Kendall non- where n is the number of observations and ti are the parametric trend test is recommended by the World ties of the sample time series. And m is the number of Meteorological Organization to detect statistically sig- tied groups. nificant tendencies in environmental datasets When the sample size n ≥ 10, as used in this study, (Irannezhad et al., 2016). The use of the Mann– the test statistic Z is calculated (Kendall, 1975). Kendall trend test is widespread in the analysis of 8 9 S 1 < if S> 0 = climatological and hydrological time series, because it Z ¼ 0 if S ¼ 0 (6) is simple and robust, and can cope with missing values : ; Sþ1 if S< 0 and values falling beneath the detection limit (Gavrilov et al., 2016). This non-parametric test is commonly where Z follows a normal distribution, a positive Z and used to detect monotonic tendencies in a series of a negative Z depict an upward and downwards trend environmental data, too (Pohlert, 2016). Mann– for the period, respectively. Kendall trend test is also used by many scholars to The presence of positive autocorrelation in the data analyse crop production trends. For instance, Ahmad increases the chance of detecting trends when actually et al. (2017) and Polisetty and Paidipati, Fentaw and none exists, and vice versa (Hamed & Rao, 1998). In the Dereje Hailu (2017) used Mann–Kendall and other non- present study, autocorrelation has been taken into parametric statistics to analyse the trends of Maize and account using the Hamed and Rao method to avoid Wheat production in Pakistan and India, respectively. the above uncertainty. This version is based on the The Mann–Kendall trend test is based upon Kendall modified variance (Hamed & Rao, 1998; Taxak et al., (1975) and is closely related to Kendall’s rank correla- 2014). tion coefficient. To determine the presence of MK calculates Kendall’s statistics (S), the sum of the a monotonic trend in a time series, the null hypothesis difference between data points and a measure of asso- (H ) of the Mann–Kendall test is that there is no mono- 0 ciations between two samples (Kendall’s tau) to indi- tonic trend in the series, while the alternative hypoth- cate an increasing or decreasing trend. Positive values esis (H ) is that the data follow a monotonic trend over a of those parameters indicate a general tendency time. The MK test is a rank-based method for trend towards an increasing trend while negative values analysis of time-series data (Burn et al., 2004; Tesemma show a decreasing trend. Finally, a two-tailed probabil- et al., 2010). The normalised test statistics Z for the MK ity (p-value) was computed and compared with the test is computed using Equations (3)–(6) (Yu et al., user-defined significance level (5%) to identify the 1993). trend of variables. X X n 1 n S ¼ sgn x x (3) j i i¼1 j¼iþ1 2.6. Sen’s slope estimation where 8 � 9 Sen’s Slope estimation test is also another non- <þ1if x x > 0 = j i � � parametric trend analysis method for climatic, hydrolo- sgn x x ¼ 0 if x x ¼ 0 (4) j i j i : ; gic and sediment yield studies (Sen, 1968). It computes 1if x x < 0 j i both the slope (i.e. the linear rate of change) and inter- sgn is the signum function and xi and xj are the annual cepts according to Sen’s method. The magnitude of the values in the years i and j, i > j, respectively, trend is predicted by (Sen, 1968; Theil, 1950) slope The application of the trend test is done to a time estimator methods. A positive value of β indicates an series Xi that is ranked from i = 1, 2 . . . n-1 and Xj, which ‘upward trend’ (increasing values with time), while is ranked from j = i + 1, 2 . . . . n. Each of the data points a negative value of β indicates a ‘downward trend. ALL EARTH 7 Here, the slope T of all data pairs are computed as (Sen, is shown in Figure 3. Statistically significant change 1968). In general, the slope T between any two values of points cannot be detected in all positions of bunds. a time series x can be estimated from That means the crop production data were not inhomogeneous. x x k j Ti ¼ (7) j k where x and x are considered as data values at time j j k 3.2. River flow (m ) trend analysis result and k (j >k) correspondingly. The median of these The results of the descriptive statistics; MK trend test N values of T is represented as Sen’s estimator of and Sen’s slope are presented in (Table 1). The result slope which is computed as showed the maximum and minimum river flows from 8 9 TNþ1 if N is odd 3 < = 2 the watershed were recorded in 1996 (5242418.1 m ) � � T þT N ðNþ1Þ Q ¼ (8) 2 and 2015 (444779.5 m ) respectively with a mean of : if N is even; 1958943.1 m and a standard deviation of 1287910.5 m . The coefficient of variation (CV %) indicated high A positive value of Qi indicates an upward or increas- variability of monthly, seasonal and annual river flow ing trend and a negative value of Qi gives a downward (CV > 30) (Table 1). The tendencies were also analysed or decreasing trend in the time series using an auto-correlated Mann–Kendall trend test, as modified according to Hamed and Rao (1998) to take 3. Results and discussion into account the possibility of autocorrelation in the meteorological data. If an auto-correlated trend test 3.1. Homogeneity test of the data was used, a significant monotonic tendency could be 3.1.1. Pettitt’s test of monthly, seasonal and yearly detected in the annual dataset. The monthly MK and river flow and sediment yield Sen’s slope trend test result implied there is The Pettitt tests of river flow and sediment yield were a significantly decreasing trend in March, August, applied to the monthly, seasonal and annual patterns. September and October at (P < 0.05) level of signifi - The empirical significance level of the Pettitt test cance (Figure 4). (p-value) is shown in Figure 2. Statistically significant The rainfall distribution pattern of Andit tid change points can be detected in March, July, August, watershed is bimodal and concentrated from March September, October, summer, winter and even in to May (Belg season) and June to September (summer mean annual river flow. The sediment yield data have season). Based on the distribution pattern of rainfall, no significant change points rather it has constant these months are the main rainy months of the sequences for the months and seasons which have watershed. So, the significant decrement in river flow zero sediment yield records. Months and seasons in these months and the insignificant decrement of the have constant sequences with zero records in other months have a relation with either development December, January, February and winter. In the time of the water holding capacity of the soil or decreasing series of river flow amounts, the significant shift of the trend of the rainfall in the watershed. Even though mean is downward over all the months and seasons. these all months showed abrupt changes, by using Hamed and Rao autocorrelation testing we detected 3.1.2. Pettitt’s test of crop production data the sharpness of the data. The seasonal and annual MK The Pettitt tests for homogeneity of crop yield data and Sen’s slope trend results showed no significant were applied to all sampling points (A, B and C). The change in river flow for the last 24 years (1994–2017), empirical significance level of the Pettitt test (p-value) while it was insignificantly decreasing. Similar to our 0.4 1.2 0.35 0.3 0.8 0.25 0.2 0.6 0.15 0.4 0.1 0.2 0.05 (a) (b) Figure 2. The Pettitt’s test result of monthly, seasonal and annual (a) river flow and (b) sediment yield. p-values Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual Belg Summer Spring Winter p-values Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual Belg Summer Spring Winter 8 A. WOLDEMARIAM ET AL. 0.5 0.25 0.4 0.2 0.3 0.15 0.2 0.1 0.1 0.05 A B C A B C Barley Horse Baen 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 A B C Lentil Wheat Figure 3. The Pettitt’s test result of the crop production data in different position terraces (A, B and C). Table 1. Monthly, seasonal and annual trend and variability result of river flow (m ). Variable Min Max Mean Std. deviation CV (%) Auto correlated MK value Sen’s slope Jan 176.7 386153.6 19463.6 78364.0 402.6 −0.182 −61.1 Feb 159.6 223193.3 12066.5 45320.1 375.6 −0.196 −43.0 Mar 176.7 530881.6 41086.9 114416.0 278.5 −0.317 * −134.7 Apr 650.7 519234.0 52437.7 111165.9 212.0 −0.152 −532.5 May 440.0 718838.5 76863.6 156037.2 203.0 −0.072 −351.6 Jun 405.0 235257.0 30448.6 51939.2 170.6 −0.007 −2.9 Jul 768.8 1097059.9 486088.3 277947.7 57.2 −0.290 −16361.3 Aug 146895.1 1484077.1 711755.7 329228.4 46.3 −0.398** −30692.0 Sep 56181.9 1223821.6 287340.3 307614.6 107.1 −0.39** −10781.3 Oct 1005.4 958392.5 154254.0 256500.3 166.3 −0.304* −2692.8 Nov 171.0 359716.5 54510.8 97395.5 178.7 −0.207 −477.5 Dec 176.7 481175.8 32627.0 100025.8 306.6 −0.186 −89.2 Belg 2821.5 1768954.1 170388.2 364775.3 214.1 −0.174 −2609.2 Summer 337438.1 3555128.0 1515633.0 789769.4 52.1 −0.384 −56588.0 Spring 1418.5 1260204.0 208764.9 309568.7 148.3 −0.203 −4518.4 Winter 513.0 618263.2 64157.1 161234.1 251.3 −0.254 −265.2 Annual Q (m ) 444779.5 5242418.1 1958943.1 1287910.5 65.7 −0.326 −67192.0 3 3 3 Min, minimum river flow in m ; Max, maximum river flow in m ; std., standard deviation of river flow in m ; CV, coefficient of variation in percent; MK, Mann–Kendall. result, Tesemma (2009) reported that there was no studies reported that the simulated annual runoff trend in the mean annual and flood season runoff in increased by 10% over the 40 years as a result of the the Upper Blue Nile basin. The trend analysis study increase in degraded soils (Steenhuis et al., 2014). performed using indicators of hydrologic alteration (IHA) at the Tekeze river basin showed a significant 3.3. Sediment yield (ton) trend analysis result decreasing trend in maximum flow duration and rise rate (Fentaw and Dereje Hailu, 2017). The study con- The results of some descriptive statistics, MK trend ducted at Awash River showed a significantly decreas- test and Sen’s slope are presented in Table 2. The ing trend of discharge from 1980 to 2016 (Gedefaw result showed the maximum and minimum sediment et al., 2018). The result of our research is in contrast yield from the watershed were recorded in 1994 with the studies conducted in the Blue Nile basin (3064.7 tons) and 2007 (392.1 tons) respectively which reported a significantly increasing trend of run- with a mean of 1693.8 tons and a standard deviation off during the wet season, the short rainy season and of 744.4 tons. The coefficient of variation (CV %) the annual period (Gebremicael et al., 2013). The other indicated the high variability of the monthly, p-values p-values p-values p-values C ALL EARTH 9 Figure 4. The long-term trends of monthly river flow (m /s) (March, August, September and October showed a significantly decreasing trend). seasonal and annual sediment yield (CV >30%). The trend of sediment yield at (P < 0.05) level of signifi - tendencies were also analysed using an auto- cance for the last 24 years (1994–2017), while it was correlated Mann–Kendall trend test, as modified insignificantly decreasing (Figure 5). according to Hamed and Rao (1998) to take into Even though it was insignificant; the annual MK and account the possibility of autocorrelation in the Sen’s slope indicated decreasing trend in sediment meteorological data. The monthly, seasonal and yield. The decreasing trend in the sediment yield is annual MK and Sen’s slope trend test results showed due to the application of different soil and water con- that there was no significant increasing or decreasing servation measures in the watershed. Fanya juu and 10 A. WOLDEMARIAM ET AL. bench terraces were applied in the watershed as an most significant decrease, with sediment yield decreas- intervention to protect the soil from erosion. Currently, ing up to 90% in the 2000s compared with the 1950s due to a lack of maintenance, the constructed soil and (Zhongbao et al., 2015). Studies in Blue Nile basin water conservation measures are dimensioned and reported by Steenhuis et al. (2014) revealed that the consequently, it will affect river flow and crop produc- sediment loads have been highly increasing, but this tivity if they are not maintained properly. Similar to our needs further validation as data availability is limited. finding, the trend analysis study conducted in the east- The other studies conducted in the Blue Nile basin ern part of the middle Yellow River basin showed the outlet also revealed statistically increasing sediment ALL EARTH 11 Table 2. Monthly, seasonal and annual trends of sediment yield (ton). Minimum Maximum Mean Std. deviation CV (%) MK test Sen’s slope Jan 0.0 0.0 0.0 0.0 CS CS Feb 0.0 0.0 0.0 0.0 CS CS Mar 0.0 489.3 49.4 116.1 234.7 −0.224 0 Apr 0.0 593.7 97.8 136.4 139.5 −0.140 −1.445 May 0.0 886.0 179.9 246.1 136.8 −0.034 0.000 Jun 0.0 868.5 162.1 281.3 173.5 −0.126 0.000 Jul 137.9 1378.9 634.6 329.0 51.9 −0.036 −2.933 Aug 74.9 938.6 411.9 217.5 52.8 −0.138 −6.627 Sep 0.0 342.8 123.8 94.6 76.4 0.077 1.558 Oct 0.0 303.9 33.7 69.9 207.5 0.008 0.000 Nov 0.0 7.0 0.6 1.9 339.1 −0.224 0.000 Dec 0.0 0.0 0.0 0.0 CS CS Belg 0.0 1500.8 327.1 401.5 122.8 −0.004 −0.171 Summer 337.0 2315.5 1332.4 578.6 43.4 −0.080 −9.805 Spring 0.0 303.9 34.2 69.7 203.7 0.008 0.000 Winter 0.0 0.0 0.0 0.0 CS CS Annual SY(ton) 392.1 3064.7 1693.8 744.4 43.9 −0.036 −5.441 SY, sediment yield; CS, sequences are constant or ‘0’ ton records. Table 3. The long-term monthly suspended sediment concentration (SSC) of Andit tid watershed. Month Mean Median Minimum Maximum Standard deviation January 0 0 0 0 0 February 0 0 0 0 0 March 3.32 0.04 0.00 17.93 5.05 April 2.74 1.32 0.00 10.43 3.47 May 2.77 1.76 0.00 10.23 3.28 June 3.25 0.00 0.00 22.03 6.05 July 1.76 1.32 0.45 5.19 1.42 August 0.59 0.52 0.03 1.85 0.46 September 0.70 0.48 0.00 3.79 0.91 October 0.30 0.04 0.00 3.72 0.81 November 0.02 0.00 0.00 0.31 0.07 December 0 0 0 0 0 Annual 1.29 Regression of Sediment yield (ton) by River flow (m ) (R²=0.947) -500000 0 500000 1000000 1500000 2000000 2500000 -500 River flow (m ) Model Conf. interval (Mean 95%) Conf. interval (Obs 95%) Figure 6. The regression graph of river flow (m ) and sediment yield (ton) of Andit tid watershed. Sediment yield(ton) 12 A. WOLDEMARIAM ET AL. 2.0 Different from mean Annual SSC (g/l) Linear (Annual SSC (g/l)) 1.5 1.0 y = 0.0413x + 0.5211 R² = 0.2573 0.5 0.0 1994 1995 1996 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2012 2013 2014 2016 2017 -0.5 Year -1.0 Figure 7. The long-term annual Suspended Sediment Concentration (SSC) of Andit tid watershed. Table 4. Trends of crop production in different positions of bunds). Sample position Sampled year Min (kg/ha) Max (kg/ha) Mean (kg/ha) Std. deviation CV (%) MK value Sen’s slope Barley production A 17 years 455.7 1750.0 940.2 369.5 39.3 0.015 1.718 B 17 years 418.2 1443.8 827.7 324.0 39.1 0.206 14.921 C 17 years 360.9 1375.0 741.8 271.7 36.6 0.250 13.542 Wheat production A 14 years 328.1 1325.0 775.7 328.3 42.3 - - B 14 years 237.5 1241.1 644.6 322.6 50.0 −0.077 −10.056 C 14 years 131.3 1450.0 588.1 341.3 58.0 0.022 6.5 Lentil production A 16 years 250 1593.8 691.2 347.8 50.3 −0.183 −9.263 B 16 years 250 1462.5 608.9 292.4 48.0 −0.067 −5.221 C 16 years 300 1175.0 563.0 236.7 42.0 −0.1 −5.477 Horse bean production A 8 years 300 2250 927.3 607.1 65.5 0.327 35.655 B 8 years 300 1750 839.6 470.3 56.0 0.429 47.13 C 8 years 200 1875 734.5 534.5 72.8 0.429 41.563 A, above bund; B, between bunds; C, below bunds; Min, Minimum; Max, Maximum. 6 6 load from 91 × 10 tons/year in 1980–1992 to 147 × 10 3.4.1. Suspended sediment concentration (SSC) tons/year in 1993–2009 (Gebremicael et al., 2013). The annual average long-term suspended sediment concentration (SSC) for the study watershed was 1.29 g/L, ranging from 0.34 g/L in 2007 to 1.55 g/L in 2013 3.4. Relation between sediment yield and river flow and 2017 as shown in Figure 6. As seen in Figure 7, there are no noticeable increasing or decreasing pat- Great care must be applied when studying the rela- terns in the watershed’s annual SSC; however, it has tionship between river flow and sediment yield in been slightly rising since 2008. watersheds where biological and physical factors are According to the monthly SSC analysis, the months subjected to change during the rainy season. In this with the highest SSC are March, April, May, June and study, we have seen the correlation and regression of July (3.32, 2.74, 2.77, 3.25 and 1.76 g/l, respectively), river flow and sediment yield by taking the long-term while December, January and February have SSC levels monthly, seasonal and annual average data of both of 0 g/l (Table 3). As previously noted, the river flow parameters. The correlation matrix implied that the reaches its peak in August compared to the other river flow and sediment yield have significantly corre- months, although August’s suspended sediment con- lated (R = 97.3%) at (P < 0.01). The regression analysis centration was only 0.7 g/L. The watersheds’ loose soil, clearly shows that the river flow and sediment yield which is easily washed away at the beginning of the have a linear relation with the equation (Sediment rainy season, causes the concentrations of sediment to −4 3 yield (ton) = 3.41+ (8.55 × 10 )*River flow (m )). The peak before the rainy season starts. The months with graph showing the regression between river flow and the highest sediment concentration are those in which there is no cover over the ground, rendering the soil to sediment yield is in Figure 6: SSC(g/l) ALL EARTH 13 easily eroded. Additionally, beginning in the middle of disease and pests affecting Lentil in the watershed. The March, all cultivated fields are ploughed and ready for increasing (even though it was insignificant) trend of crop growth; subsequently, modest rainfall can carry other crop types might be the result of new crop vari- a lot of soil and can contribute to a high concentration eties and agronomic practices demonstrated in the of suspended sediment. All agricultural land and other watershed by the Debre Brihan Agricultural Research land uses get covered with green vegetation after mid- Centre and Bureau of Agriculture. The average barley, summer, and the soil also develops a high cohesive wheat and horse bean grain yield of the watershed force; this results in a decreased concentration of sus- from 1987 to 1994 were 112.8, 57.14 and 139.3 kg/ha, pended sediment in August and September compared respectively (Hurni, 2000). When compared to crop pro- to early summer. The occurrence of SSC peak before duction before 1994, the mean crop yield of the flow peak is attributed to the soil properties that are watershed is highly increased. loose erodible sediment at the beginning of the rainy season because of ploughing (Zimale et al., 2018). Sediment concentrations in rivers at the beginning of 4. Conclusion and recommendation the monsoon season are high and then decrease gra- dually (Tilahun et al., 2011, 2013). In this paper, we analysed the variability and trend of river flow, sediment yield and crop productivity of Andit tid watershed from 1994 up to 2017. To analyse 3.5. Crop production (kg/ha) trends the variability and trend; we used descriptive statistics and non-parametric statistical tests. To detect the In this trend analysis, we analysed the crop production abrupt changes in the data, autocorrelation has trends of major growing crops of the watershed with been taken into account using the Hamed and Rao soil and water conservation structures applied in the method. With this analysis, the coefficient of variation watershed. The major growing crops of the watershed of inter-monthly, inter-seasonal and inter-annual river are Barley, Wheat, Horse Bean and Lentils. So we ana- flow and sediment yield indicated high variability (CV lysed the trends by relating the production to soil and >30) of both parameters. Similar to river flow and water conservation. Table 4 shows some descriptive sediment yield; crop production also showed high statistics and Mk test value and Sen’s slope for the inter-annual variability. The annual and seasonal crop yield in the watershed. The descriptive statistics trend analysis result revealed that there were no sta- indicated that all the plots immediately above the tistically significant (P < 0.05) changes in the trend of bunds (zone A) delivered higher crop yields, while the river flow and sediment yield. Monthly trend analysis plots below bunds (zone C) gave lower crop yields. This showed statistically significant decreasing trends for implies the positive impacts of soil and water conserva- river flow at March, August, September and October tion practices applied in the watershed. Even though river flow. the conservation structures are partially diminished due The correlation and regression analysis of river flow to lack of maintenance, they are still contributing to and sediment yield indicated that they have a statistically crop productivity. The lowering of crop yield in the strong linear relation (R = 97.3%). This study determined plot below the bunds (zone C) is expected to be the through the SSC analysis that the early summer runoff result of nutrient depletion as a result of the loss of transported more sediment than the mid and late- topsoil with erosion and lack of water storage capacity summer runoff. This is a result of the early-summer bare- of the soil. In the other report by Hurni (2000) of Andit ness of the ground and the soil’s tendency to quickly tid watershed from 1982 to 1994, all the crops delivered erode, whereas in the mid- and late-summer, the green the highest yield immediately above bunds and the cover and the formation of cohesive soil force help to lowest yield immediately below bunds. The crop pro- minimise the sediment carried by runoff. duction performance of the watershed for the last 24 This study confirmed the positive effects of soil and years (1994–2017) was generally low. The watershed is water conservation measures on crop productivity. All characterised by relatively high population and live- crops from plots above soil bunds delivered the high- stock densities, a high degree of land degradation, est yield and the lowest yield was harvested from plots low crop yield and production as well as drastically below bunds. The overall crop productivity trend in the reduced fallow periods (Hurni, 2000). The coefficient of watershed showed an increasing trend except for variation (CV) indicated that there was inter-annual Lentil. Lastly, for large-scale river flow, sediment yield variability of crop yield in all crop types and all positions and crop production analysis, we suggest for the of bunds (CV > 30%). The MK value and Sen’s slope researchers to focus on the representative watershed value showed statistically insignificant trends for all (s) with required data monitoring. Researchers can find crop types and positions of bunds. Even though it was out more about the causes triggering any changes in insignificant; Lentil production in the watershed all analysed parameters. showed decreasing trend. This might be the result of 14 A. WOLDEMARIAM ET AL. Fentaw, Fikru, & Hailu, Dereje (2017). Trend and variability Acknowledgments analysis of rainfall & stream flow series at Tekeze River The authors would like to acknowledge the Amhara Regional Basin, Ethiopia. International Journal of Scientific & Agricultural Research Institute (ARARI) and Water and Land Engineering Research, 8(11), 665–680 http://www.ijser.org ). Resource Center (WLRC) for their financial support during the Gavrilov, M. B., Tosic, I., Markovic, S. B., Unkasevic, M., & whole life of this research. We duly thank all the data enco- Petrović, P. (2016). Analysis of annual and seasonal tem- ders Mr Temesgen Yilma and data collectors and guards of perature trends using theMann- Kendall test in Vojvodina. Andit tid watershed specifically Mr Amara Belachew; Mr Időjárás, 120(2), 183–198. Zewdie Assefa and Mr Solomon Alemu. Gebrehiwot, S., Taye, A., & Bishop, K. (2010). Forest cover and stream flow in a headwater of the blue nile: Complementing observational data analysis with commu- nity perception. AMBIO: A Journal of the Human Disclosure statement Environment, 39(4), 284–294. https://doi.org/10.1007/ s13280-010-0047-y No potential conflict of interest was reported by the Gebremicael, T. G., Mohamed, Y. A., Betrie, G. D., van der author(s). Zaag, P., & Teferi, E. (2013). Trend analysis of runoff and sediment fluxes in the Upper Blue Nile basin: A combined analysis of statistical tests, physically-based models and landuse maps. Journal of Hydrology, 482, 57–68. https:// ORCID doi.org/10.1016/j.jhydrol.2012.12.023 Ayele Woldemarim http://orcid.org/0000-0002-3253-9532 Gedefaw, M., Wang, H., Yan, D., Song, X., Yan, D., Dong, G., Wang, J., Girma, A., Aijaz Ali, B., Batsuren, D., & Abiyu, A. (2018). Trend analysis of climatic and hydrological vari- References ables in the Awash River Basin, Ethiopia. Water, 10(1554), 1554. https://doi.org/10.3390/w10111554 Abrham Tezera, T. C., Jemal, T., & Feyisa, A. (2016). Impact Hamed, K. H., & Rao, A. R. (1998). A modified Mann–Kendall assessment of land use/land cover change on soil erosion trend test for autocorrelated data. Journal of Hydrology, and rural livelihood in Andit tid Watershed, North Shewa, 204(1–4), 182–196. https://doi.org/10.1016/S0022- Ethiopia. Archives of Current Research International, 3(1), 1694(97)00125-X 1–10. https://doi.org/10.9734/ACRI/2016/22268 Hurni, H. (2000). Area of Andit Tid, Shewa, Ethiopia: Ahmad, S., Israr, M., & Asif Yaseen, N. A. (2017). Climate Long-term monitoring of the agricultural environment change trend analysis on selected food crops at central 1982 - 1994. Khyber Pakhtunkhwa of Pakistan. International Journal of Irannezhad, M., Marttila, H., Chen, D., & Kløve, B. (2016). Development and Sustainability, 6(12), 2082–2093. Century-long variability and trends in daily precipitation Betrie, G., Mohamed, Y., van Griensven, A., Srinivasan, R., & characteristics at three Finnish stations. Advances in Mynett, A. (2011). Sediment management modelling in Climate Change Research, 7(1–2), 54–69. https://doi.org/ Blue Nile Basin using SWAT model. Hydrology and Earth 10.1016/j.accre.2016.04.004 System Sciences, 15 (3) , 807–818 doi: 10.5194/hess-15-807- Kebede, E. (2009). Hydrological responses to land cover changes in Gilgel Abbay catchment of Ethiopia. Blume, T., Zehe, E., & Bronstert, A. (2007). Rainfall—runoff Kendall, M. G. (1975). Rank correlation methods (4th ed.). response, event-based runoff coefficients and hydrograph Charles Griffin. separation. Hydrological Sciences Journal, 52(5), 843–862. Kothawale, D. R., & Rupa Kumar, K. (2005). On the recent https://doi.org/10.1623/hysj.52.5.843 changes in surface temperature. Bosshart, U. (1997). Measurement of River Discharge for SCRP Legesse, D., Vallet-Coulomb, C., & Gasse, F. (2003). Research Catchments; University of Bern, Switzerland in Hydrological response of a catchment to climate and land- association with the Ministry of Agriculture, Ethiopia use changes in Tropical Africa: Case study South Central (Vol. 31). Ethiopia. Journal of Hydrology, 275(1–2), 67–85. https://doi. Burn, D., Cunderlik, J., & Pietroniro, A. (2004). Hydrological org/10.1016/S0022-1694(03)00019-2 trends and variability in the Liard River basin / Tendances Liu, L., Xu, Z. X., & Huang, J. X. (2012). Spatio-temporal varia- hydrologiques et variabilité dans le basin de la rivière tion and abrupt changes for major climate variables in the Liard. Hydrological Sciences Journal, 49(1), 53–68. https:// Taihu Basin, China. Stochastic Environmental Research and Risk Assessment, 26(6), 777–791. https://doi.org/10.1007/ doi.org/10.1623/hysj.49.1.53.53994 s00477-011-0547-8 Conway, D. (2000). The climate and hydrology of the Upper Miller, J. R., Sinclair, J. T., & Walsh, D. (2015). Controls on Blue Nile River. The Geographical Journal, 166(1), 49–62. suspended sediment concentrations and turbidity within https://doi.org/10.1111/j.1475-4959.2000.tb00006.x a reforested, Southern Appalachian headwater basin. Desalegn, A., Tezera, A., & Tesfay, F. (2018). Developing Water (Switzerland), 7(6), 3123–3148. https://doi.org/10. GIS-based soil erosion map using RUSLE of Andit Tid 3390/w7063123 Watershed, Central Highlands of Ethiopia. Journal of Pettitt, A. (1979). A non-parametric approach to the Scientific Research and Reports, 19(1), 1–13. https://doi. org/10.9734/jsrr/2018/40841 change-point problem. Applied Statistics, 28(2), 126–135. Engda, T. A. (2009). Modeling rainfall, runoff and soil loss https://doi.org/10.2307/2346729 relationships in the northeastern highlands of Ethiopia, Pohlert, T. (2016). Non-parametric trends and change-point Andit tid watershed. Cornell University. detection. Erturk, A. (2010). Water quality assessment and meta model Roth, V., & Lemann, T. (2015). Comparing CFSR and conven- development in Melen watershed – Turkey. Journal of tional weather data for discharge and sediment loss mod- Environmental Management, 91(7), 1526–1545. https:// elling with SWAT in small catchments in the Ethiopian doi.org/10.1016/j.jenvman.2010.02.021 Highlands. Hydrology and Earth System Sciences ALL EARTH 15 Discussions, 12, 11053–11082. https://doi.org/10.5194/ Steenhuis, T. S. (2011). A saturation excess erosion hessd-12-11053-2015 model. ASABE - International Symposium on Erosion and Seleshi, B. A., Wubet, F. D., McCartney, M., & Yilma, S. (2011). Landscape Evolution 2011, May 2014, 381–389. https:// Hydrological water availability, trends and allocation in the doi.org/10.13031/2013.42675 blue Nile river basin. Hydrology, climate and land use. Womber, Z. R., Zimale, F. A., Kebedew, M. G., Asers, B. W., Springer Science. Deluca, N. M., Guzman, C. D., Tilahun, S. A., Sen, P. K. (1968). Estimates of the regression coefficient based Zaitchik, B. F., & Kaloop, M. (2021). Estimation of sus- on Kendall’s tau. Journal of the American Statistical pended sediment concentration from remote sensing Association, 63(324), 1379–1389. https://doi.org/10.1080/ and in situ measurement over Lake Tana, Ethiopia. 01621459.1968.10480934 Advances in Civil Engineering, 2021, 1–17. https://doi. Steenhuis, T. S., Tilahun, S. A., Tesemma, Z. K., Tebebu, T. Y., org/10.1155/2021/9948780 Moges, M., Zimale, F. A., Worqlul, A. W., Alemu, M. L., Yohannes, H., & Soromessa, T. (2018). Land suitability assess- Ayana, E. K., & Mohamed, Y. A. (2014). Soil erosion and ment for major crops by using GIS-based multi-criteria discharge in the Blue Nile Basin: Trends and challenges. approach in Andit Tid watershed, Ethiopia. Cogent Food Nile River Basin, 133–147. https://doi.org/10.1007/978- & Agriculture, 4(1), 1470481. https://doi.org/10.1080/ 3-319-02720-3_8 23311932.2018.1470481 Sutcliffe, J. V., & Parks, Y. P. (1999). The hydrology of the Nile. Yohannes, H., Soromessa, T., & Hamere Yohannes & Teshome IAHS Special Publication IAHS Press, Institute of Hydrology. Soromessa. (2019). Integration of remote sensing, GIS and Taxak, A. K., Murumkar, A. R., & Arya, D. S. (2014). Long term MCDM for land capability classification in Andit Tid spatial and temporal rainfall trends and homogeneity Watershed, Ethiopia. Journal of the Indian Society of analysis in Wainganga basin, Central India. Weather and Remote Sensing, 47(5), 763–775. https://doi.org/10.1007/ Climate Extremes, 4, 50–61. https://doi.org/10.1016/j.wace. s12524-019-00949-z 2014.04.005 Yurekli, K., Simsek, H., Cemek, B., & Karaman, S. (2007). Tesemma, Z. K. (2009). Long term hydrologic trends in the Nile Simulating climatic variables by using stochastic basin. Cornell University. approach. Building and Environment, 42(10), 3493–3499. Tesemma, Z. K., Mohamed, Y. A., & Steenhuis, T. S. (2010). https://doi.org/10.1016/j.buildenv.2006.10.046 Trends in rainfall and runoff in the Blue Nile Basin: 1964– Yu, Y. -S., Zou, S., & Whittemore, D. (1993). Non-parametric 2003. Hydrological processes, 24(25), 3747–3758. https:// trend analysis of water quality data of rivers in Kansas. doi.org/10.1002/hyp.7893 Journal of Hydrology, 150(1), 61–80. https://doi.org/10. Theil, H. (1950). A rank-invariant method of linear and poly- 1016/0022-1694(93)90156-4 nomial analysis, part 3. Nederlandse Akademie van Zhongbao, X., Bofu, Y., & Yuguo, H. (2015). Spatiotemporal Wettenschappen. Proceedings, 1397–1412. Variations in Annual Sediment Yield from the Middle Tilahun, S. A., Guzman, C. D., Zegeye, A. D., Engda, T. A., Yellow River, China, 1950–2010. Journal of Hydrologic Collick, A. S., Rimmer, A., & Steenhuis, T. S. (2013). An Engineering, 20(8). https://doi.org/10.1061/(ASCE)HE. efficient semi-distributed hillslope erosion model for the 1943-5584.0001113 subhumid Ethiopian Highlands. Hydrology and Earth Zimale, F. A., Moges, M. A., Alemu, M. L., Ayana, E. K., System Sciences, 17(3), 1051–1063. https://doi.org/10. Demissie, S. S., Tilahun, S. A., & Steenhuis, T. S. (2018). 5194/hess-17-1051-2013 Budgeting suspended sediment fluxes in tropical monsoo- Tilahun, S. A., Mukundan, R., Demisse, B. A., Guzman, C., nal watersheds with limited data: The Lake Tana basin. Tarakegn, B. C., Engda, T. A., Easton, Z. M., Collick, A. S., Journal of Hydrology and Hydromechanics, 66(1), 65–78. Zegeye, A. D., Schneiderman, E. M., Parlange, J. Y., & https://doi.org/10.1515/johh-2017-0039
http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png
All Earth
Taylor & Francis
http://www.deepdyve.com/lp/taylor-francis/long-term-trends-of-river-flow-sediment-yield-and-crop-productivity-of-4Y2UraGID0
