Hydrologic Evaluation of Multi-Source Satellite Precipitation Products for the Upper Huaihe River Basin, China

Zhiyong Wu,Zhenchen Liu,Zhengguang Xu,Fang Wang,Jianhong Zhou,Hai He,Xiaotao Wu

doi:10.3390/rs10060840

Abstract

To evaluate the performance and hydrological utility of merged precipitation products at the current technical level of integration, a newly developed merged precipitation product, Multi-Source Weighted-Ensemble Precipitation (MSWEP) Version 2.1 was evaluated in this study based on rain gauge observations and the Variable Infiltration Capacity (VIC) model for the upper Huaihe River Basin, China. For comparison, three satellite-based precipitation products (SPPs), including Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) Version 2.0, Climate Prediction Center MORPHing technique (CMORPH) bias-corrected product Version 1.0, and Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) 3B42 Version 7, were evaluated. The error analysis against rain gauge observations reveals that the merged precipitation MSWEP performs best, followed by TMPA and CMORPH, which in turn outperform CHIRPS. Generally, the contribution of the random error in all four quantitative precipitation estimates (QPEs) is larger than the systematic error. Additionally, QPEs show large uncertainty in the mountainous regions, with larger systematic errors, and tend to underestimate the precipitation. Under two parameterization scenarios, the MSWEP provides the best streamflow simulation results and TMPA forced simulation ranks second. Unfortunately, the CHIRPS and CMORPH forced simulations produce unsatisfactory results. The relative error (RE) of QPEs is the main factor affecting the RE of simulated streamflow, especially for the results of Scenario I (model parameters calibrated by rain gauge observations). However, its influence on the simulated streamflow can be greatly reduced by recalibration of the parameters using the corresponding QPEs (Scenario II). All QPEs forced simulations underestimate the streamflow with exceedance probabilities below 5.0%, while they overestimate the streamflow with exceedance probabilities above 30.0%. The results of the soil moisture simulation indicate that the influence of the precipitation input on the RE of the simulated soil moisture is insignificant. However, the dynamic variation of soil moisture, simulated by precipitation with higher precision, is more consistent with the measured results. The simulation results at a depth of 0–10 cm are more sensitive to the accuracy of precipitation estimates than that for depths of 0–40 cm. In summary, there are notable advantages of MSWEP and TMPA with respect to hydrological applicability compared with CHIRPS and CMORPH. The MSWEP has a greater potential for basin–scale hydrological modeling than TMPA.

Highlights

Precipitation is an important component of the hydrological cycle and the most primary forcing data of hydrological models [1,2,3]
CHIRPS performs unsatisfactorily compared with TMPA and Center MORPHing technique (CMORPH), which is probably due to the fact that CHIRPS is mainly based on IR data, while CMORPH and TMPA combined IR and MV data
The evaluation of the hydrological simulation capability in this study partly indicates that the merged precipitation product, Multi-Source Weighted-Ensemble Precipitation (MSWEP), has great potential to be a reliable dataset for conducting long-term hydrological studies compared with the three satellite-based precipitation products (SPPs)

Summary

Introduction

Precipitation is an important component of the hydrological cycle and the most primary forcing data of hydrological models [1,2,3]. The SPPs have a wide spatial coverage and high spatiotemporal resolution, effectively making up deficiencies of the conventional rain gauge observations and greatly enriching alternative precipitation data sources, especially in data-scarce or ungauged regions [17,18]. Benefitting from these advantages, the SPPs have been extensively applied in many fields such as hydrological simulations [2,14,17,19], extreme events analysis [20,21,22,23], and water resource management [24]. The SPPs are inevitably subject to errors resulting from sampling uncertainties and retrieval algorithms; the error characteristics change depending on the climate regions, seasons, altitudes, and other factors [3,7,25,26]

Objectives

Methods

Results

Discussion

Conclusion