Satellite Precipitation Characterization, Error Modeling, and Error Correction Using Censored Shifted Gamma...

Area of Science:

• Hydrology
• Remote Sensing
• Atmospheric Science

Background:

• Satellite multisensor precipitation products (SMPPs) are valuable but limited by accuracy issues.
• Systematic biases and random errors affect precipitation occurrence and magnitude estimates from SMPPs.

Purpose of the Study:

• To characterize the Tropical Rainfall Measurement Mission Multi-Satellite Precipitation Analysis (TMPA) using the Censored Shifted Gamma Distribution (CSGD).
• To compare TMPA precipitation data against the North American Land Data Assimilation System Phase 2 (NLDAS-2) reference dataset.
• To develop and apply a CSGD-based error modeling framework to quantify and reduce TMPA errors.

Main Methods:

• Utilized the Censored Shifted Gamma Distribution (CSGD) to model precipitation occurrence and magnitude.
• Compared TMPA with the NLDAS-2 reference precipitation dataset across the conterminous United States.
• Employed a CSGD-based error modeling framework to quantify TMPA errors and merge TMPA with MERRA-2 atmospheric data.

Main Results:

• Climatological CSGD characterization revealed significant regional differences between TMPA and NLDAS-2 in precipitation magnitude and occurrence probability.
• The error modeling framework successfully quantified TMPA errors relative to NLDAS-2.
• Merging TMPA with MERRA-2 data demonstrated robust reductions in random error, particularly for stratiform precipitation.

Conclusions:

• The CSGD provides a robust method for characterizing and correcting satellite precipitation estimation errors.
• Merging satellite precipitation data with atmospheric reanalysis products like MERRA-2 offers a promising approach to improve precipitation accuracy.
• Improvements were more substantial in the near-realtime version of TMPA compared to the research version.