Comment on hess-2021-476

Abstract

Understanding the impacts of land use changes (LUCC) on the dynamics of water quantity and quality is necessary to identify suitable mitigation measures that are needed for sustainable watershed management. Lowland catchments are characterized by a strong interaction of streamflow and near-surface groundwater that intensifies the risk of nutrient pollution. This study aims to reveal the relationship between long-term land use change and the water and nutrient balance in a typical lowland catchment in northern Germany. A hydrologic model (Soil and Water Assessment Tool, SWAT) and partial least squares regression (PLSR) were used to quantify the impacts of different land use types on the variations in actual evapotranspiration (ET), surface runoff (SQ), base flow (BF), and water yield (WYLD) as well as on sediment yield (SED), total phosphorus (TP) and total nitrogen (TN) loads. To this end, the model was calibrated and validated with daily streamflow data (30 years) as well as sediment and nutrient data from two water quality measurement campaigns (3 years in total). Three model runs over thirty years were performed using land use maps of 1987, 2010, and 2019, respectively. Land use changes between those years were used to explain the modelled changes in water quantity and quality on the subbasin scale applying PLSR. SWAT achieved a very good performance for daily streamflow values (calibration: NSE = 0.79, KGE = 0.88, PBIAS = 0.3 %; validation: NSE = 0.79, KGE = 0.87, PBIAS = 7.2 %), a satisfactory to very good performance for daily TN (calibration: NSE = 0.64, KGE = 0.71, PBIAS = −11.5 %; validation: NSE = 0.86, KGE = 0.91, PBIAS = 5 %), a satisfactory performance for daily sediment load (NSE = 0.54–0.65, KGE = 0.58–0.59, PBIAS = −22.2 %–12 %), and an acceptable performance for daily TP (calibration: NSE = 0.56, KGE = 0.65, PBIAS = −4.7 %; validation: NSE = 0.29, KGE = 0.22, PBIAS = −46.2 %) in the Stör Catchment. The variations in ET, SQ, BF, WYLD, SED, TP, and TN could be explained to an extent of 61 %–88 % by changes in the area, shape, dominance, and aggregation of individual land use types. They were largely correlated with the major LUCC in the study area i.e. a decrease of arable land, and a respective increase of pasture and settlement. The change in the areal percentage of arable land positively affected the dynamics of SED, TP, TN and negatively affected BF, indicated by a Variable Influence on Projection (VIP) > 1.16 and large absolute regression coefficients (RCs: 0.6–0.88 for SED, TP, TN; −1.65 for BF). The change in pasture area was negatively affecting SED, TP, and TN (RCs: −0.69–−0.12, VIPs > 1) while positively affecting ET (RC: 0.09, VIP: 0.92). The change in settlement percentage had a VIP of up to 1.17 for SQ and positively and significantly influenced it (RC: 1.16, p-value < 0.001). PLSR helped to identify the key contributions from individual land use changes on water quantity and quality dynamics. These provide a quantitative basis for targeting most influential land use changes to mitigate impacts on water quality in the future.