Continuous thunderstorm monitoring: retrieval of instantaneous rainfall and precipitating ice rates from lightning and GOES IR observations adjusted with TRMM Precipitation Radar

Carlos Augusto Morales

Department of Civil and Environmental Engineering, University of Connecticut

A new algorithm is presented to estimate rainfall and precipitating ice rates based on lightning and GOES IR images observations adjusted with the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) over the north and south Americas. This new procedure assumes that lightning is associated with ice particles. As a result, a better convective rain/area definition is obtained. Parameterizations of rain area and convective rain area are obtained for both electrified and non-electrified clouds. The rainfall rates and precipitating ice rates are adjusted to resemble the observed TRMM-PR measurements. During the calibration period we have used 631 TRMM-PR orbits and a total of 366 electrified and 3103 non-electrified clouds have been used to extract the rainfall and precipitating-ice relationships, during December 1997 and January 1998. Error statistics are computed from independent TRMM-PR measurements on February 1998 and rain gauge measurements from a network in Florida, US, during December 1997 through February 1998.

Comparisons of rain area and rain volume show that our algorithm underestimates the rain area for both cloud types: electrified (-21.55%) and non-electrified clouds (-29.14%), while for the rain volume we overestimate by 19.54% for electrified and 12.52% for non-electrified clouds. Mean hourly rainfall rates at selected regions show that we are able to represent the same diurnal cycle as observed by TRMM-PR and no diurnal bias is observed. Histograms of observed convective and stratiform rainfall and precipitating ice distributions from our method show similar distributions as observed by TRMM-PR. Finally comparisons with the rain gages network has shown that our proposed algorithm is able to represent the same observed precipitation distribution at 100 km² up to 4 degree². Our algorithm has a tendency to overestimate the precipitation by 5.77% for 100 km² up to 35.43% for 4 degree².