H075-04
Exploration on the Applicability of Different Rainfall Products for Hydrological Simulations
Abstract:
Rainfall is considered as the most essential component in the hydrological cycle (e.g., its impact on soil moistures, Wang et al, 2020) and, thus of great significance in hydrological simulations. Nowadays, different rainfall products are broadly used by meteorologists and hydrologists in research as there exist many circumstances of small catchments with sparsely available ground observations. However, due to the emerging of newly developed datasets, it is still lack of understanding as to how to select and what are the pros and cons. The research hence seeks to investigate the performance features of different rainfall products for hydrological simulation.
In this study, a humid and natural area (about 635 km2) which suffers from flooding frequently is taken as a case study. The performance of nine rainfall datasets, encompassing rain gauges, weather radars (NEXRAD), satellite-based products (GPM IMERG and PERSIANN-CCS), reanalysis products (NLDAS, GLDAS, ERA5 and ERA5-Land), as well as high-resolution rainfall estimates downscaled by Weather Research and Forecasting (WRF) model, are compared at the catchment scale in space and hourly scale in time (Ying et al, 2020), with 24 scenarios of WRF configurations to work out the most appropriate parameterisations in the study catchment. Then the well-known Xin’anjiang (XAJ) hydrological model was applied for evaluating the reliability of the datasets selected in streamflow simulation. In particular, seven commonly-used evaluation indicators (Nash-Sutcliffe Efficiency (NSE), Pearson’s correlation coefficient (PCC), root mean square error (RMSE), Percent Bias (PBIAS) and Kling–Gupta Efficiency (KGE), Percentage Error of Peak (PEP) and Error of Peak time (EPT)) are employed to appraise the performance of different rainfall estimates and the corresponding streamflow forecasts.
The results show that KGE is the most useful indicator in evaluating the rainfall products, whilst NSE is more suitable in assessing the performance of flood simulations. Moreover, the NLDAS product outperforms other rainfall datasets when compared with the gauged observations, whilst NEXRAD rainfall exhibits the best performance in streamflow simulation through the XAJ hydrological modelling, attributed to its high spatial and temporal resolution. The findings also show that with the suitable selected parameter configurations, the rainfall downscaled by WRF is capable of generating reliable streamflow. It is hoped this study will provide useful guidance to the community on selecting suitable rainfall products for hydrological applications.
Keywords: Rainfall data sources, Assessment criterion, Weather Radar precipitation, Satellite precipitation, Reanalysis precipitation, Rainfall, WRF, XAJ.
References
Wang, J., Chen, O., Chen, Y., Liu, Y., Zhuo, L., Rico-Ramirez, M., and Han, D. Flood inundation mapping with multi-satellite soil moisture observations, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-9225, https://doi.org/10.5194/egusphere-egu2020-9225, 2020.
Liu, Y., Chen, Y., Chen, O., Wang, J., Zhuo, L., and Han, D. Exploration of WRF simulations of extreme rainfall in Egypt, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-10538, https://doi.org/10.5194/egusphere-egu2020-10538, 2020.