Spatiotemporal analysis of soil loss in cold climate upland farming watersheds using SWAT: case study of Tokoro River watershed, Hokkaido, Japan

Abstract

Severe soil erosion can cause both loss of ecosystem services and riverine environmental contaminations. Thus, understanding the spatiotemporal characteristics of soil loss at the watershed scale is critical for sustainable management of land and water resources. The primary objective of this study was to estimate the spatiotemporal variation in water runoff and soil loss in the Tokoro River watershed (TRW), Hokkaido, Japan. Assuming that agricultural land, which becomes bare depending on the season, is the main source of erosion and that large soil losses occur especially during the snowmelt/freeze-thaw period, we quantified and analyzed the mechanism of soil loss using the Soil and Water Assessment Tool 2012 (SWAT2012). Simulation results based on a publicly available dataset showed that croplands were the primary source of stream sediment load, contributing to 86% of the total soil loss. Furthermore, the soil loss rate was high (241.1–270.9 Mg km−2 yr−1) in sub-basins with a higher percentage of cropland, especially sloped croplands. These sub-basins had a higher risk of productivity loss due to intense erosion and river contamination by muddy runoff. Soil loss mainly occurred in early spring (March and April) and late summer (August–October), accounting for 36% and 45% of the annual soil loss, respectively. Vegetation removal by harvest exacerbated soil loss in October. In March, soil drainage was suppressed by frozen soil, resulting in high surface runoff and soil loss. Therefore, surface runoff when the soil was frozen was considered an important driver of soil loss. The estimation error of the stream sediment loads out of the 95% prediction uncertainty accounted for 26% of the total absolute error. This error could be attributed a lack of certainty in the observation data of high stream sediment loads and the model’s underestimation of water runoff during intense snowmelt and rainfall events.