Rainfall-Runoff Processes and Spatial Variability of Soil Properties of the Koupendri Catchment in North-West Benin, West Africa under different Land Use/Land Cover

Abstract

Soil is an important input data in most hydrological modeling. A detailed (1:25000) soil survey and soil classification was done using SOTER (SOil and TERrain) approach. Data on both morphological (color, structure, mottles e.t.c), physical (texture, bulk density (BD) and saturated hydraulic conductivity (Ksat)), and chemical properties (soil organic carbon (SOC), cation exchange capacity (CEC), total nitrogen (TN), available phosphorus (Avail.P), carbon-nitrogen ratio (C:N) e.t.c) were obtained. 180 surface soil (0-20 cm) samples were collected in a grid resolution of 25 m x 25 m and supplemented with additional 116 samples at 5 m x 5 m resolution within the grid across three land use types (maize-sorghum cropland (MS), rice field and fallow-shrub-grassland (FSG)) to evaluate spatial variability of soil properties. These samples were analyzed using standard laboratory procedures. Further data analysis using simple statistical and geostatistical tools and Pearsons's correlation analysis were performed. The soil map of the catchment showed five distinct soil types, belonging to three major soil orders namely: Ultisols, Inceptisols and Alfisols (USDA), and four reference soil groups: Plinthosols, Cambisols, Luvisols and Gleysols (WRB). More than 55 % of the soils were mainly Plinthosols, highly degraded, acidic in nature and low in nutrient with low soil water retention capacity. The variability was high for Ksat, moderate for SOC, Avail.P and C:N, but low for TN, BD. SOC showed the strongest spatial dependence (0-5.3%) across the three land use. The spatial dependence, pattern and distribution of the soil properties were influenced by land use and soil management practices but independent of individual property or geology. Six microplots (1m x 1m) for measuring surface runoff were installed on each of the two slope positions and each of the two land use types. Moreso, infiltration runs, soil samples and soil moisture measurements were done in triplicates. The suitability of a conceptual model (NRCS-CN) to capture plot-scale rainfall-runoff processes was tested. To understand rainfall-runoff processes at the catchment scale, WaSiM model was calibrated and validated using the most sensitive and the optimized soil, land use and unsaturated zone parameters. Surface runoff coefficient for the microplots and the catchment were computed and compared. Soil hydrologic properties (infiltration rate, soil moisture, surface runoff and saturated hydraulic conductivity) were influenced by land use and slope position though little disparity exists between the laboratory and the in situ determined values. Both total and event-based plot scale surface runoff were high and significantly (p < 0.05) influenced by the land use. Runoff coefficient at plot scale (0.3-0.9 %) compared to the catchment scale (0-0.6 %) revealed strong scale issues. NRCS-CN simulation captured the dynamics of the plot-scale rainfall-runoff processes (R2= 0.84 -0.97; ME= 0.56-0.91) with slight over-and-underestimation of peak and low flows. The model WaSiM was successfully calibrated (NSE=0.61, R2=0.61, RMSE= 0.63) with acceptable uncertainty range or value (P-factor = 94%; r-factor = 0.93), and also validated (NSE=0.68, R2=0.78, RMSE= 0.57) for the Koupendri catchment. The result was satisfactory and showed strong agreement between the modeled and the observed data. Direct flow was the dominant runoff (more than 85 % of the total runoff) component in the catchment. The water balance reflected the hydro-climatic condition of the catchment with real evapotranspiration ranging between 68 % and 75 % of the total annual rainfall.