Performance Evaluation of Level-2 TRMM Rain Profile Algorithms by Intercomparison and Hypothesis Testing

View More View Less
  • a NASA Goddard Space Flight Center Greenbelt, Maryland
  • b Department of Meteorology, Florida State University, Talahassee, Florida
© Get Permissions
Restricted access

Abstract

Currently, satellite algorithms are the methodology showing most promise for obtaining more accurate global precipitation estimates. However, a general problem with satellite methods is that they do not measure precipitation directly, but through inversion of radiation–rain relationships. Because of this, procedures are needed to verify algorithm-generated results. The most common method of verifying satellite rain estimates is by direct comparison with ground truth data derived from measurements obtained by rain gauge networks, ground-based weather radar, or a combination of the two. However, these types of comparisons generally shed no light on the physical causes of the differences. Moreover, ground validation measurements often have uncertainty magnitudes on the order of or greater than the satellite algorithms, motivating the search for alternate approaches. The purpose of this research is to explore a new type of approach for evaluating and validating the level-2 Tropical Rainfall Measuring Mission (TRMM) facility rain profile algorithms. This is done by an algorithm-to-algorithm intercomparison analysis in the context of physical hypothesis testing.

TRMM was launched with the main purpose of measuring precipitation and the release of latent heat in the deep Tropics. Its rain instrument package includes the TRMM Microwave Imager (TMI), the Precipitation Radar (PR), and the Visible and Infrared Scanner (VIRS). These three instruments allow for the use of combined-instrument algorithms, theoretically compensating for some of the weaknesses of the single-instrument algorithms and resulting in more accurate estimates of rainfall. The focus of this research is on the performance of four level-2 TRMM facility algorithms producing rain profiles using the TMI and PR measurements with both single-instrument and combined-instrument methods.

Beginning with the four algorithms' strengths and weaknesses garnered from the physics used to develop the algorithms, seven hypotheses were formed detailing expected performance characteristics of the algorithms. Procedures were developed to test these hypotheses and then applied to 48 storms from all ocean basins within the tropical and subtropical zones over which TRMM coverage is available (∼35°N–35°S). The testing resulted in five hypotheses verified, one partially verified, and one inconclusive. These findings suggest that the four level-2 TRMM facility profile algorithms are performing in a manner consistent with the underlying physical limitations in the measurements (or, alternatively, the strengths of the physical assumptions), providing an independent measure of the level-2 algorithms' validity.

Abstract

Currently, satellite algorithms are the methodology showing most promise for obtaining more accurate global precipitation estimates. However, a general problem with satellite methods is that they do not measure precipitation directly, but through inversion of radiation–rain relationships. Because of this, procedures are needed to verify algorithm-generated results. The most common method of verifying satellite rain estimates is by direct comparison with ground truth data derived from measurements obtained by rain gauge networks, ground-based weather radar, or a combination of the two. However, these types of comparisons generally shed no light on the physical causes of the differences. Moreover, ground validation measurements often have uncertainty magnitudes on the order of or greater than the satellite algorithms, motivating the search for alternate approaches. The purpose of this research is to explore a new type of approach for evaluating and validating the level-2 Tropical Rainfall Measuring Mission (TRMM) facility rain profile algorithms. This is done by an algorithm-to-algorithm intercomparison analysis in the context of physical hypothesis testing.

TRMM was launched with the main purpose of measuring precipitation and the release of latent heat in the deep Tropics. Its rain instrument package includes the TRMM Microwave Imager (TMI), the Precipitation Radar (PR), and the Visible and Infrared Scanner (VIRS). These three instruments allow for the use of combined-instrument algorithms, theoretically compensating for some of the weaknesses of the single-instrument algorithms and resulting in more accurate estimates of rainfall. The focus of this research is on the performance of four level-2 TRMM facility algorithms producing rain profiles using the TMI and PR measurements with both single-instrument and combined-instrument methods.

Beginning with the four algorithms' strengths and weaknesses garnered from the physics used to develop the algorithms, seven hypotheses were formed detailing expected performance characteristics of the algorithms. Procedures were developed to test these hypotheses and then applied to 48 storms from all ocean basins within the tropical and subtropical zones over which TRMM coverage is available (∼35°N–35°S). The testing resulted in five hypotheses verified, one partially verified, and one inconclusive. These findings suggest that the four level-2 TRMM facility profile algorithms are performing in a manner consistent with the underlying physical limitations in the measurements (or, alternatively, the strengths of the physical assumptions), providing an independent measure of the level-2 algorithms' validity.

Save