A new approach to ensemble precipitation Estimation: Coupling satellite hydrological products with backward water balance models in Large-Scale

Abstract

The estimation accuracy of global precipitation datasets has been enhanced using a committee of various sources. However, the existing merging methods mainly rely on gauge-based measurements, which may have significant uncertainty across data-scarce regions. The unique approach of Triple Collocation (TC) has been widely used in various geographical information systems since it allows quantifying the uncertainties of three datasets with independent errors. Recent studies have confirmed the use of the TC for consolidating disparate triple datasets. However, merging techniques based on Modified Collocation (MC), which can combine more than three datasets, have not yet been applied to precipitation data. This study develops the backward water balance as a new approach to ensemble precipitation estimation via large-scale hydrological products without access to ground information. The proposed methodology is validated using monthly observations from more than 580 rain gauge stations in the Central Plateau watershed of Iran. In the proposed method, the error of each water balance component is statistically estimated using different methods, like MC. Furthermore, the residuals of the water balance were minimized using three techniques with different complexity levels, such as the Constrained Kalman Filter, to enforce water balance closure; hence, the error is smoothed. The precipitation products are optimally merged by employing various methods in the ensemble precipitation estimation framework. As stated in the metrics, the best-recorded estimated ensemble precipitation data improved the accuracy of the used raw precipitation products over the study region compared with the benchmark values in 10.8%, 2.3%, and 0.2% greater KGE, NSE, and TS values. Additionally, this data resulted in a 31.2% smaller RMSE. The proposed modification in the CKF method (CKF2) showed better results to close the water balance.