A Modified ATI Technique for Nowcasting Convective Rain Volumes over Areas

Abstract

Digital radar data from the North Dakota Cloud Modification Project (NDCMP)?the 1981 and 1982 summer experiments?are used to further investigate the relationship between convective rain volumes and area-time-integrals (ATI). The ATI technique provides a means of estimating total rain volumes using area covered by rain events (for reflectivities ≥ 25 dBz) integrated over the cluster duration (Doneaud et al., 1984a). The purpose of this investigation is twofold: (a) to estimate ATIs only for the growth portion of a convective storm (while the rain volume is computed using the entire life history of the convective event); and (b) to nowcast the total rain volume of a convective system at the stage of its maximum development. For the aforementioned purpose, the ATIs were computed using the maximum echo area ≥ 25 dBz (ATIA), the maximum reflectivity (ATIR), and the maximum echo height (ATIH) as the end of the growth portion of the convective event. A simple linear regression analysis demonstrated that correlations between total rain volume (TVR) or the maximum rain volume (MYR) versus ATIA were the strongest. In a log-log plot, the correlation coefficient and the standard error of estimates of total rain volume versus ATIA were 0.98 and 0.23 for the summer 1982 data, and 0.96 and 0.24 for the summer 1981 data, respectively. In percentage terms, the corresponding range of variation of the rain volume for a given ATIA lies between 70% and ? 41% (1982 data) and between 74% and ? 44% (1981 data). That is comparable to the uncertainties which typically occur in rain volume estimates obtained from radar data employing Z-R conversion followed by space and time integration. This demonstrates that the total rain volume of a storm can be nowcasted at its maximum stage of development (max ATIA). The scatter in the rain volume and in the maximum volumetric rain rate estimates are somewhat smaller if a multiple linear regression instead of a simple linear regression is considered, but the improvement is of little significance. The tests with independent data confirmed the consistency of the results for the region considered.