Relationships between lightning flash rates and radar reflectivity vertical structures in thunderstorms over the tropics and subtropics

[1] Relationships between the vertical profile of radar reflectivity and lightning flash rates are investigated using 13 years of Tropical Rainfall Measuring Mission (TRMM) observations during 1998–2010. First the Radar Precipitation Features (RPFs) are defined by grouping raining areas detected by the TRMM Precipitation Radar (PR). Then the characteristics of radar reflectivity and lightning flash rate are calculated in each RPF using PR and Lightning Imaging Sensor (LIS) observations. Using these RPFs, temperatures at 20, 30, and 40 dBZ radar echo tops, used as proxies of the maximum convective intensity of precipitation systems, are examined as indicators of the probability of lightning. Although 30 and 40 dBZ echo top temperatures are better indicators of the probability of lightning than the 20 dBZ echo top temperature, there is a large regional variation in the temperature thresholds, especially between land and ocean. In general, oceanic thunderstorms have higher 20 dBZ echo top and larger horizontal extent than those over land. However, radar reflectivity is more likely to exceed 30 and 40 dBZ at cold temperatures over land than over ocean. The correlations between flash rates and radar echo top temperatures, areas and volumes of radar reflectivity, and ice water contents in the mixed phase region are analyzed using RPFs with at least one flash. In agreement with previous studies, the correlations with the echo top temperatures are low, but the correlations between flash rates and areas and volumes of high radar reflectivity in the mixed phase region are much higher. There is a high correlation between the flash rates and the volumes with radar reflectivity greater than 30, 35, or 40 dBZ in the mixed phase region, but the correlation coefficient varies significantly between thunderstorms over different regions, especially between land and ocean. These results are confirmed by repeating the analysis for regions of the storms defined as convective, thus eliminating the contribution from large areas of stratiform radar echo that have much less lightning.