Assessing Drought Vulnerability in the Brazilian Atlantic Forest Using High-Frequency Data

Abstract

This research investigates the exposure of plant species to extreme drought events in the Brazilian Atlantic Forest, employing an extensive dataset collected from 205 automatic weather stations across the region. Meteorological indicators derived from hourly data, encompassing precipitation and maximum and minimum air temperature, were utilized to quantify past, current, and future drought conditions. The dataset, comprising 10,299,236 data points, spans a substantial temporal window and exhibits a modest percentage of missing data. Missing data were excluded from analysis, aligning with the decision to refrain from using imputation methods due to potential bias. Drought quantification involved the computation of the aridity index, the analysis of consecutive hours without precipitation, and the classification of wet and dry days per month. Mann–Kendall trend analysis was applied to assess trends in evapotranspiration and maximum air temperature, considering their significance. The hazard assessment, incorporating environmental factors influencing tree growth dynamics, facilitated the ranking of meteorological indicators to identify regions most exposed to drought events. The results revealed consistent occurrences of extreme rainfall events, indicated by positive outliers in monthly precipitation values. However, significant trends were observed, including an increase in daily maximum temperature and consecutive hours without precipitation, coupled with a decrease in daily precipitation across the Brazilian Atlantic Forest. No significant correlation between vulnerability ranks and weather station latitudes and elevation were found, suggesting that geographical location and elevation do not strongly influence observed dryness trends.