Statistical and Hydrological Evaluations of Multi-Satellite Precipitation Products over Fujiang River Basin in Humid Southeast China

Weiwei Sun,Gang Yang,Jun Ma,Weiyue Li

doi:10.3390/rs10121898

Abstract

The purpose of the paper is to evaluate the quality and hydrological utility of four popular satellite precipitation products, including the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) product (3B42V7), near real-time product (3B42RT), and the Climate Prediction Center (CPC) MORPHing technique (CMORPH) satellite–gauge merged product (CMORPH BLD) and bias-corrected product (CMORPH CRT) over Fujiang River basin, China. First, we provided a statistical assessment of the four precipitation products at multiple spatiotemporal scales. The results show that: (1) all the products except 3B42RT capture the spatial pattern of annual precipitation fairly well; (2) in general, CMORPH BLD benefits from the application of the probability density function-optimal interpolation (PDF-OI) gauge adjustment algorithm and performs best among all the products with Pearson correlation coefficients (CC) of 0.84 and 0.94, equitable threat score (ETS) of 0.56 and 0.63 in grid and basin scales, respectively, followed by 3B42V7 and CMORPH CRT; whereas 3B42RT performs worst across all the metrics; (3) according to the occurrence frequencies of rainfall, satellite estimates mainly fall into the bin of 0–1 mm/day and tend to underestimate light precipitation. In addition, the performance of all the products in warm season is much better than in cold season in both grid and basin scales. Subsequently, a physically based distributed model is established to further evaluate the hydrological utility of different precipitation products. The results reveal that: (1) the errors in precipitation products mainly propagate into hydrological simulations, resulting in the best hydrological performance in CMORPH BLD in both daily and monthly scales after recalibrating the model, while 3B42RT shows limited skills in reproducing the daily observed hydrograph; (2) after recalibrating the model with the respective satellite data, significant improvements are observed for all the products; (3) CMORPH BLD no longer shows its superiority during near-real-time monitoring of floods. There is still a great challenge for the application of current satellite-based estimates into local flood monitoring. This study could be used as guidance for choosing alternative satellite precipitation products for hydrological applications in a local community, particularly in basins in which rainfall gauges are scarce.

Highlights

Precipitation is the key forcing variable in atmospheric, hydrological and energy cycles at local, regional and global scales, which is of great importance to agricultural irrigation, water resources management and disaster monitoring linked to floods, droughts, and landslides [1,2,3]
One thing to be noted is that the rainfall gauges used for the generation of CGDPA are independent of those used for the generation of Tropical Rainfall Measuring Mission (TRMM) and CMORPH (Table 1), and the reliability of the evaluation results is guaranteed
The results shows that the overall accuracy of all the products declined (POD and equitable threat score (ETS) were reduced and False alarm ratio (FAR) was increased) with increasing rainfall rate, which indicated the limited capability of satellite sensors for depicting intense precipitation [54]

Summary

Introduction

Precipitation is the key forcing variable in atmospheric, hydrological and energy cycles at local, regional and global scales, which is of great importance to agricultural irrigation, water resources management and disaster monitoring linked to floods, droughts, and landslides [1,2,3]. There are three ways to measure precipitation, i.e., rain gauges, weather radars, and satellite-based sensors [4,5]. A conventional rain gauge is a straightforward way to provide precipitation observations. It is always insufficient for precipitation monitoring due to its sparse and uneven distributions in some regions [5]. A weather radar can monitor precipitation accurately, but it may suffer from the problem of high cost and limited coverage in mountainous regions in terms of distorted signals [6]. Satellite-based rainfall products have recently been accepted as a result of wide spatial coverage, fine spatio-temporal resolutions and easy data acquisition, providing an alternative and effective way to obtain large-scale precipitation measurement [7,8]. It is recognized and recommended to integrate IR and MW sensors to improve the accuracy and time resolution of satellite precipitation products

Objectives

Methods

Results

Discussion

Conclusion