Modelling stream flow with a discrete rainfall–runoff model and 37 GHz PDBT microwave observations: the Xiangjiang River basin case study
Haolu Shang1,2, Massimo Menenti2,1, and Li Jia11Stake Key Laboratory of Remote Sensing Science, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China 2Geoscience and Remote Sensing, Delft University of Technology, Delft, 2628 CN Delft/2600 GA Delft, The Netherlands
Received: 18 Mar 2016 – Accepted for review: 15 Jun 2016 – Discussion started: 23 Jun 2016
Abstract. A discrete rainfall–runoff model has been developed, which uses retrievals of Water Saturated Soil (WSS) and inundation area from 37 GHz microwave observations. The model was implemented at three levels of increasing complexity using field-measured ground water table, WSS and inundated area, and precipitation data. The three levels, defined by the key-variables are: (1) precipitation and base flow; (2) overland flow, infiltrated flow and base flow; (3) overland flow, potential subsurface flow and base flow. The base flow is estimated from observed ground water table depth, while overland and infiltrated flows are estimated from precipitation and the WSS and inundated area. A linear scaling method is developed to estimate the potential subsurface flow. The three model implementations are calibrated with the gauge measurements of 10-day average river discharge in 2002 and 2005 respectively at Changsha station, downstream of Xiangjiang River basin, China. The discrete rainfall–runoff model assumes that specific runoff is determined by antecedent precipitations over a variable period of time. This duration is a model parameter varying between 10 and 150 days. The performance of the discrete rainfall–runoff model increased with the duration of antecedent precipitation for all three implementations in both years. With a duration of 150 days, the model reaches its best performance: Nash–Sutcliffe Efficiency, NSE, for the 1st implementation was ≥ 0.90 with relative RMSE ≤ 22 %; NSE ≈ 0.99 with relative RMSE ≤ 5 % for the 2nd implementation, and NSE ≥ 0.99 with relative RMSE ≤ 4 % for the 3rd one. These good performances prove that the retrievals of WSS and inundated area clearly improve model accuracy, thus justifying the choices of parameters and the method to estimate the potential subsurface flow. The set of parameters driving each implementation is an indication of dominant hydrological processes, particularly water storage, in determining the catchment response to rainfall. Significant differences in the annual water yield have been observed across the three implementations. The relative RMSE in each season demonstrates the possible recharge period of the ground water in Xiangjiang River basin.
Shang, H., Menenti, M., and Jia, L.: Modelling stream flow with a discrete rainfall–runoff model and 37 GHz PDBT microwave observations: the Xiangjiang River basin case study, Hydrol. Earth Syst. Sci. Discuss., doi:10.5194/hess-2016-129, 2016.