Long term trends of hydrology, sediment yield and crop productivity in Andit Tid watershed central highlands of Ethiopia

Background: Previously in Ethiopia reliable climatic and hydro-meteorological data are not available and not maintained properly but the long-term database is needed for the assessment and planning of resource dynamics. To minimize the lack of reliable database, the Soil Conservation Research Program (SCRP) established observatory model watersheds since 1981. Andit tid watershed is one of these watersheds established for monitoring the long term trends of climatic, hydrologic, sediment loss and crop production system as a representative site for central high land parts of Ethiopia at 1982. This research paper compiles the analysis of spatial and temporal distribution of the rain fall; trends of run off and sediment loss and their relation and the influence of position of terraces on crop production. The rainfall spatial and temporal distribution trend analysis results conclude that the watershed is not vulnerable for future drought. The highest sediment concentration occurred in June was because of the reason that the lands are plowed and prepared for crop growth; following this small rainfall can carry much soil and can contribute for high suspended sediment concentration. The highest grain yield obtained from above bunds is because of the trapped and accumulated soil and plant nutrient could contribute for better performance and production of crops. To obtain better crop yield; to minimize sediment loss and improve the stream flow it is better to maintain the existed soil and water conservation structures and apply the new interventions.

In the watershed different soil and water conservation structures were constructed by different programs since 1994. Even though the function of the constructed structures is not validated, almost a half of the watershed was treated by campaign soil and water conservation structures. To examine and estimate the importance of these conservation structures sediment samples were taken based on the rainfall and runoff events. These long-term collected sediment samples expected to give reliable evidence on the performance of soil and water conservation structures than the erosion empirical models. So in this trend analysis progress we add the analysis of sediment loss with relating other climatic, crop production and hydrologic parameters.
Precipitation directly affects the availability of water resources and is one of the most important climatic factors and hydrological parameters (De Luis et al., 2011). Investigating the temporal variations of precipitation in previous time periods is critical for making reliable predictions of future climate changes.
With mentioned justification the aim of this study were: (1) to analyze the long term data and establish trends of climate (rainfall and temperature), discharge, sediment yield and crop productivity impacts of soil and water conservation practices, and (2) to provide a hydro-sedimentological characterization of the watershed, so that information has been available for longer time series monitoring since 1982..

Climatic data collection and analysis
A-class meteorological instruments were installed near the outlet of the watershed. The rainfall data was recorded using four manual rain gauges, distributed throughout the watershed to collect catchment-scale representative data. From continuous readings of the automatic rain gauge, rainfall characteristics including amount, intensity, and time intervals between storm events were determined.
We used precipitation concentration index (PCI) which is the ratio of square of the rainfall amount of the specific month to the square of the total rainfall to show the distribution of rainfall in the watershed.
According to the , the PCI value of less than 10% represents a uniform rainfall distribution (i.e. low rainfall concentration); PCI values between 11-15 denote a moderate rainfall concentration; values from 16-20 denote irregular rain fall distribution, and values above 20% represents the irregularity(i.e. high rainfall concentration) of rainfall distribution (De Luis, 2011).
Precipitation coefficient (PC) was calculated as the ratio between the mean monthly rainfall and one twelfth of the mean annual rainfall. When the PC greater than one, the month is wet month that can contribute more than one twelfth of the mean annual rainfall and dry months contribute less than one twelfth of the mean annual rainfall. A month is rainy if the rainfall coefficient is greater than 0.6. The expression "small rains" are used to refer to months with rain fall coefficient of 06-0.9; and the expression "big rains" refer to months with rainfall coefficient of 1 and above. Big rainy months are further classified in to three groups: months with "moderate concentration" (coefficient of 1 to 1.9); months with "high concentration" (coefficient of 2-2.9); and months with "very high concentration" (coefficient of 3 and above) (WLRC, 2015).
The climatic data at the station also include minimum and maximum air temperature, minimum and maximum soil temperature. Air and soil temperature were measured using 1.5 and 0.1 meter above ground thermometers that were installed in the station under shelter respectively.

River discharge and sediment data collection
The river gauge stage was monitored continuously using limuniograph accompanied with manual water level measurements during storm events. Whenever there was runoff events, one-liter grab samples were taken every 10 minutes interval as soon as the water turned brown for sediment measurement. When the water level decreased and the runoff water returned to its original color, sampling rate decreased to 30 minutes intervals and then hour intervals. Together sediment sampling, the river water level was measured manually to determine the total stream flow and to estimate the suspended sediment carried by the flow at that specific time interval. The amount of sediment load within the sample was determined by oven drying the one liter samples then weighing the oven dried soil. Total soil loss for that sampling interval was then calculated by multiplying total water flow per time by the sediment concentration determined from the one liter sample.
The reliability of all collected rainfall, stream flow and suspended sediment load raw data were checked before the analysis. Data events at which the river height was beyond the rating equation were also avoided. Wrongly written starting and ending times for stream flow recordings and sediment samplings were also adjusted.
We used the rating equation developed by Bosshart (1997) (2) Where Q is the runoff discharge in l/s and H is the true water level (height of stage) in cm??? Drainage ratio which is the ratio of runoff to rainfall was calculated to identify when the rainfall and runoff reaches maximum and minimum. Time of concentration is useful in predicting flow rates that would result from hypothetical storms, which are based on statistically derived return periods through IDF curves (Monjo, R. (2016). The time of concentration (TC) and time of peak discharge (TP) were also generated as: Where, TC is time of concentration; L is the length of the largest stream; S is the slope variation between the upper stream and the lower stream and TP is the time of peak discharge.

Crop yield data collection
Catchment harvest data is the representative yield and biomass sample including data on management practices, inputs, soil depth, slope, tillage, precursor crop, and crop type data taken from 35 fixed and 50 non-fixed plots from the farmers' land. In the fixed plot the samples were taken from 'a' (above terrace (zone of deposition)), 'b' (between terrace) and 'c' (below terrace (zone of transportation)) to represent the soil erosion gradient effect. This data was used to show the productivity impacts of SWC applied in the watershed.

Rainfall characteristics
The PCI value of the watershed is 1.07% which means the rainfall in the watershed have uniform distribution (i.e. low rainfall concentration) as shown in Table 1. In fact the distribution of the rainfall can also be verified by the rain fall recorded from four different rain gauges distributed in different location of the watershed. Based on the recorded rainfall from these four rain gauge sites of the watershed there was insignificant variation among the rainfall amount. The precipitation coefficient (%) value of the watershed indicates that July (PC=2.551) and August (PC=2.395) have big rain with high concentration; September (PC=1.068) have big rain with moderate concentration and these three months could contribute more than one twelve of total rainfall amount. The long-term average annual rainfall based on 20 years of observation (1995-2017) is 1581mm. The maximum and minimum annual rainfall ever received was 2183.6 mm recorded in 1999 and 1069 mm which were recorded in 1995. The coefficient of variation for annual rainfall is 16.7%, which means the rainfall amount of each year is fairly scattered around the mean. Derib (2005) states annual rainfall with CV > 30% is an indication of high vulnerability to drought. Regardless of the higher monthly rainfall variability; the low variability of total annual rainfall minimizes the risk of drought in the study area. Hurni and Grunder (1986) verified that drought is not a problem in Andit tid because of low variability of annual rainfall. Similar to the previous study; this study also confirms that the watershed will not be vulnerable to drought and dry spell according to the low annual variability of the rainfall. The seasonal rainfall analysis result implied that; since 2008 the rainfall falling at Bulg (February to May) season is less than the long term average rainfall of the Bulg season in most years as shown in Table 2. Holden and Shiferaw (2000) also mentioned that the short rainy seasons have recently become more unreliable. The standardized climatic diagram of Andit Tid (according to Walter, 1964) is characterized by a bimodal rainfall regime with one dryer month (June) between Belg (first, small rainy season) and Kremt (second, main rainy season). Even though the rainfall amount of the Bulg season is smaller than the long term average of the Bulg rain fall; the rain fall regime of the watershed has still bimodal characteristics as mentioned in Table 2. During six months (April, May, June, July. August and September) mean monthly rainfall exceeds 100 mm; with similar result of Walter (1964); which states the four months (May, July, August and September) mean monthly rainfall exceeds 100mm. The air and soil surface temperature of the watershed did not have a continuous declining or increasing trend. Based on the long term time series air temperature data November and December were the coldest months with average value of 11.5 °C and 11.4°C respectively; whereas May and June are the hottest months with average value of 13.7°C and 13.6°C respectively. In more than 99% of the records, daily soil surface temperature was higher than daily air temperature. The difference between air temperatures versus soil surface temperature was greater during the dry season than during the rainy season. With a few exceptions the daily range of soil surface temperature (difference between minimum and maximum temperature) was larger than that of air temperature. Soil surface temperature is more sensitive to seasonal weather variations than air temperature.

Trends of discharge and sediment yield
Two small rivers, Gudibado and Wadyat, drain the catchment from east to west. Their confluence is approximately 150 m above the gauging station, which is just upstream of the asphalt bridge crossing of the Hulet Wenz. Both rivers originate from the protected perennial grass lands on a wide plateau located in the upper portion of the watershed where water accumulates and saturates then drains to the two streams. Wadyat River is a perennial river, while Gudibado is mostly seasonal. The stream flow data in the period between 1994 and 2017 shows that the discharge varied between 93.6 mm (2014)    The relationship between river discharge and sediment loss in Andit Tid watershed is 67.27%. The result indicated that as the river discharge increased by 10m 3 the sediment loss will increase by 6.7kg of sediment. This result confirms that 67 percent of the agent of sediment loss is runoff and the remaining 33 percent is the result of other factors (edaphic, topographic, and land cover).

Time of concentration (T C )
The

Trends of crop yield under conserved lands
Crop yield samples were collected on cultivated land along the existing conservation structures, i.e. within the open area between terraces/bunds. Each cropping season sampling was done permanently on various farmers' cultivated fields in the entire catchment. Three comparable samples within a terrace were taken on different locations: one immediately above (zone of deposition), one in between, and one immediately below (zone of transportation) the conservation structures.
Table5. Statistical variation of crop yield over different position of terrace ("a": above, "b": in between terrace and "c": below terraces with α=0.05) The result of on-farm yield data in relation to its positions on terraces is shown in Table5. The table shows the impact of conservation structures on the crop productivity. All crops delivered statistically highest yield immediately above bunds and the lowest yield immediately below bunds. This is the result in which the bunds could trap the soil which comes from the upper parts and the accumulated soil expected to have plant nutrients that helps for better performance and production of crops.

Figure3.
The long term mean response of major crops yield for different locations of terraces The long term average grain yield result of crops in the study watershed was slightly greater than the grain yield reported by Hurni (2000) from the period (1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994). In general, the situation in Andit Tid is difficult for peasants. It is characterized by relatively high population and livestock densities, a high degree of land degradation, low crop yield and production as well as drastically reduced fallow periods (Hurni, 2000). Beside shortage of land, the lack of manure for fertilizer is a main problem for the farmers.

Conclusion and Recommendation
This research paper compiles the analysis of temporal distribution of the rain fall, temperature, run off sediment yield and the influence of the position of terraces on crop production. Based on the rainfall analysis the mean annual rainfall of the watershed is 1581mm and the distribution has bimodal characteristics which concentrated from March to May and July to September. The PCI value of the watershed is 1.07% which means the rainfall in the watershed have uniform spatial distribution. From the rainfall trend analysis result the rainfall of the watershed has insignificant variation (CV%=16.7) which is an indicator for less vulnerability to drought. The precipitation coefficient (%) value of the watershed indicates that July (PC=2.551) and August (PC=2.395) have big rain with high concentration; September (PC=1.068) have big rain with moderate concentration and these three months could contribute more than one twelve of total rainfall amount and January, February, October and November are dry months. The long term average runoff and sediment loss of the watershed is 417.7 mm and 1613.7 ton respectively.
There was the decreased quantity of run off and sediment loss since 2000; it was expected to be the result of soil and water conservation intervention works applied in the watershed; whereas the increased trend of sediment loss starting from 2008 was the result of the oldness and destruction of applied soil and water conservation structures due to need of maintenance and other biological strengthen methods. The average soil surface and atmospheric temperature of the watershed is 14.2 o C and 12.6 o C. The result of on-farm yield data shows the impact of conservation structures on the crop productivity. All crops delivered statistically highest yield immediately above bunds and the lowest yield immediately below bunds. This is the result in which the bunds could trap the soil which comes from the upper parts and the accumulated soil expected to have plant nutrients that helps for better performance and crop production.

Ethics approval and consent to participate
I agree with the ethics of the journal and I consent to participate in the publication process of this journal.

Consent of publication
Not applicable

Availability of data and materials
The data and materials used in this manuscript are found with the hand of corresponding author.