AN ANALYSIS OF GPM IMERG AND TRMM TMPA PRECIPITATION ESTIMATES IN VARYING ENVIRONMENTS

Satellite-based estimates of precipitation have evolved and improved over the last twenty years. Merged satellite products which utilize multiple satellite intercalibrations were developed to improve the accuracy as well as the temporal and spatial resolution. These include the TRMM Multi-Satellite Precipitation Analysis (TMPA) product and more recently, the GPM Integrated Multi-Satellite Retrievals for GPM (IMERG) which offers a higher temporal and spatial resolution along with a presumed increase in accuracy. Unfortunately, due to both scientific and sensor constraints, precipitation estimates in arid environments is still problematic.

We investigated the accuracy of multiple TMPA products in varying environments (e.g., rainfall, elevation) by performing a statistical analysis using over 200 rain gauges spanning a ten year period. In addition, we provide a first look at the skill of the GPM IMERG products within these environments compared to the field gauges and previous generation product (e.g., TMPA). We analyzed the overall correlation using standard statistical measures (e.g., Pearson Correlation Coefficient, Percent Bias, Mean Absolute Error) and precipitation specific indices (e.g., Probability of Detection, False Alarm Ratio, Heidke Skill Score).

Results demonstrate the accuracy of TMPA precipitation products and highlight the opportunities and challenges of their use in data scarce regions (e.g. Morocco; MENA). Specifically, the research products outperform the real-time products in all environments within the study areas, and the newest algorithm development (3B42 V7) outperforms the previous version (V6), particularly in the low rainfall and high elevation environments. Most notably, TMPA products continue to overestimate precipitation in arid environments and underestimate in high elevation areas. Moreover, a statistically significant temporal discrepancy was found in all of the analyzed products. The GPM IMERG product showed higher correlations though only one year of data was analyzed at this time. Overall, the results corroborated findings from previous studies, highlighted the difficulty of TMPA products in varying conditions, and presented preliminary research for future algorithm development for the GPM mission.