Strengths and limitations of the Gálvez-Davison Index in forecasting tropical and subtropical convection over South America

Abstract

The Gálvez-Davison Index (GDI) was developed to enhance forecasting capabilities of tropical convection. This study evaluates the applicability of the GDI in forecasting deep tropical and subtropical convection, particularly in synoptic scale systems, such as the South Atlantic Convergence Zone (SACZ), and mesoscale phenomena like Mesoscale Convective Complexes (MCC). A hybrid methodology was applied, integrating retrospective and prospective analyses. Initially, a statistical evaluation of the GDI was conducted over the historical period from 1994 to 2024, using convection proxies including omega (vertical velocity) at 500 hPa and net longwave radiation flux (LWR). Subsequently, a brief diagnostic of SACZ and MCC events, which occurred respectively in January 2024 and October 2016, was performed. Based on these events, GDI forecasts derived from the Global Forecast System (GFS) model were compared against satellite observations and ERA5 reanalysis data within the affected regions. Key results demonstrated that, as expected, the GDI performed better in tropical than subtropical regions, showing stronger correlations with convection proxies. Statistical analysis for the SACZ (tropical region) indicated a strong negative correlation between the GDI and omega at 500 hPa ($R=-0.93$). Conversely, the correlation between GDI and omega at 500 hPa was weaker for MCC events (subtropical region; $R=-0.21$), highlighting a greater influence of dynamic factors in these systems. The GDI successfully identified regions prone to deep convection, particularly within the SACZ domain, where high GDI values coincided with regions of instability and intense precipitation. For MCC events, the index effectively captured convective intensification in subtropical regions. Forecasts of the GDI were reliable in regions where SACZ and MCC systems were active, as validated by satellite imagery. However, high index values were also recorded in areas without actual convection, indicating potential limitations. Comparison between GFS forecasts and ERA5 reanalysis revealed a systematic tendency of the global model to underestimate GDI values across all analyzed regions. The root mean square error (RMSE) varied regionally, reaching a maximum of 6.75 in the oceanic portion of the SACZ. Overall, the GDI proved valuable in forecasting deep convection, capturing both intensification and dissipation stages of convective systems. Its use as a complementary tool is promising, although systematic underestimation by the GFS highlights the need for model improvements. While originally designed for tropical regions, caution is advised when applying the GDI to subtropical areas due to its generally lower performance and weak correlations with convection proxies, particularly in the MCC domain, emphasizing the necessity for methodological refinements.