Assessment of uncertainties in soil erosion and sediment yield estimates at ungauged basins: an application to the Garra River basin, India

Abstract

Abstract. High soil erosion and excessive sediment load are serious problems in several Himalayan river basins. To apply mitigation procedures, precise estimation of soil erosion and sediment yield with associated uncertainties are needed. Here, the revised universal soil loss equation (RUSLE) and the sediment delivery ratio (SDR) equations are used to estimate the spatial pattern of soil erosion (SE) and sediment yield (SY) in the Garra River basin, a small Himalayan tributary of the River Ganga. A methodology is proposed for quantifying and propagating uncertainties in SE, SDR and SY estimates. Expressions for uncertainty propagation are derived by first-order uncertainty analysis, making the method viable even for large river basins. The methodology is applied to investigate the relative importance of different RUSLE factors in estimating the magnitude and uncertainties in SE over two distinct morphoclimatic regimes of the Garra River basin, namely the upper mountainous region and the lower alluvial plains. Our results suggest that average SE in the basin is very high (23 ± 4.7 t ha−1 yr−1) with higher values in the upper mountainous region (92 ± 15.2 t ha−1 yr−1) compared to the lower alluvial plains (19.3 ± 4 t ha−1 yr−1). Furthermore, the topographic steepness (LS) and crop practice (CP) factors exhibit higher uncertainties than other RUSLE factors. The annual average SY is estimated at two locations in the basin – Nanak Sagar Dam (NSD) for the period 1962–2008 and Husepur gauging station (HGS) for 1987–2002. The SY at NSD and HGS are estimated to be 6.9 ± 1.2 × 105 t yr−1 and 6.7 ± 1.4 × 106 t yr−1, respectively, and the estimated 90 % interval contains the observed values of 6.4 × 105 t yr−1 and 7.2 × 106 t yr−1, respectively. The study demonstrated the usefulness of the proposed methodology for quantifying uncertainty in SE and SY estimates at ungauged basins.