Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Hydrology and Remote Sensing Laboratory » Research » Publications at this Location » Publication #370441

Research Project: Integrating Remote Sensing, Measurements and Modeling for Multi-Scale Assessment of Water Availability, Use, and Quality in Agroecosystems

Location: Hydrology and Remote Sensing Laboratory

Title: The effect of spatiotemporal resolution degradation on the accuracy of IMERG products over the Huai River basin

item SU, J. - Hohai University
item LU, H. - Hohai University
item ZHU, Y. - Hohai University
item Crow, Wade
item CUI, Q. - Hohai University

Submitted to: Journal of Hydrometeorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/20/2019
Publication Date: 5/22/2020
Citation: Su, J., Lu, H., Zhu, Y., Crow, W.T., Cui, Q. 2020. The effect of spatiotemporal resolution degradation on the accuracy of IMERG products over the Huai River basin. Journal of Hydrometeorology.

Interpretive Summary: Satellite-based estimates of surface rainfall accumulations are of major benefit for efforts to track agricultural drought worldwide. In recent years, the accuracy of such products has improved significantly over short accumulation time scales (i.e., less than one day). However, agricultural drought monitoring typically requires assessments of drought (and related precipitation deficits) over longer time periods (e.g., within a 5-day pentad or a 30-day month). It this case, the temporal autocorrelation of retrieval errors becomes critically important because it determines the impact of aggregation on error levels (e.g., how daily errors in accumulation estimates are related to monthly errors in the same estimates). This paper provides the first complete description of error autocorrelation for satellite-based rainfall accumulation estimates and, therefore, the impact of temporal and spatial aggregation on uncertainty in these products. This research will eventually be used by drought monitors and water resource managers to assess the significance of observed precipitation deficits and guide drought management decision support.

Technical Abstract: The rapid development of satellite observation and retrieval techniques has prompted the continuous improvement of satellite-based precipitation estimations (SPEs) - providing new opportunities for potential applications. Recently, Version 05 (V05) of the Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG) products with high spatial (0.1° × 0.1°) and temporal (half-hourly) resolutions have become publicly available. It is important to comprehensively evaluate these new products before their widespread application. Therefore, taking the Huaihe River basin as the study case, three hourly IMERG products including the near-real-time “Early” and “Later” run products (IMERG-E and IMERG-L, respectively), and the post-real-time “Final” run product (IMERG-F) are statistically assessed against the China Merge Precipitation Analysis Hourly V1.0 product for the period April 2014 to March 2017. Subsequently, the effects of spatial and temporal resolution degradation are explored. Results indicate that all three IMERG V05 products accurately capture the spatial distribution of precipitation accumulation in the basin - despite a large positive bias in total rainfall accumulation (with grid-averaged relative biases of 39.79%, 35.75% and 29.43% for IMERG-E, IMERG-L and IMERG-F, respectively). This overestimation is caused mainly by relatively high false alarm ratios (FARs). Grid-average FARs exceeding 0.57 are found for all three IMERG products). As expected, the post-real-time IMERG-F outperforms the near-real-time IMERG products in capturing total precipitation accumulation and representing grid-scale accumulation patterns. IMERG-E demonstrates comparable performance in comparison with IMERG-L. All three IMERG products provide high probability of detection values, indicating high accuracy in detecting real precipitation events, but at same time, they also have high FARs. In addition, degrading the temporal and spatial resolution of IMERG products improves their accuracy and thus generates better performance metrics, especially at finer resolutions. However, this improvement is significantly less than that expected from a theoretical case of assuming purely uncorrelated error. Therefore, error autocorrelation clearly weakens the mutual canceling of error during averaging. Overall, temporal aggregation is more effective than spatial aggregation – suggesting that errors are more highly auto-correlated in space than in time. In summary, this evaluation provides useful guidelines for the potential applications of IMERG V05 products at suitable temporal-spatial resolutions and aids in unifying existing validation assessments obtained at different temporal and spatial scales.