Datasets and Python scripts supporting the analysis of extreme temperature events and regional heat waves in Pernambuco, Brazil (1961–2024)

This database is part of the master’s research entitled “Extreme Temperature Events: Understanding Heat Waves and Their Impacts in the State of Pernambuco, Brazil.” It contains the algorithm files used for the following procedures: Filling missing data in daily temperature time series (Filling_Gaps_v1.py) Short data gaps of one to two days are filled using linear interpolation based on the values immediately preceding and following the missing interval. Gaps lasting between three and seven days are treated using a centered seven-day moving average (±3 days), in order to minimize error propagation and preserve the fundamental statistical properties of the time series (Martins et al., 2021). Data gaps longer than seven days are not filled. This procedure requires a CSV file with two columns, where the first column contains the observation date and the second column contains the daily temperature value. Identification of seasonal heat wave occurrences from temperature time series (Heatwaves_90thPercentile_15days_v1.py) This algorithm identifies temperature occurrences exceeding the 90th percentile, classifying all continuous periods ranging from one day to the maximum observed duration. The output includes the season, start and end dates of each event, event duration, maximum temperature, minimum temperature, and mean temperature (for events lasting more than one day). The procedure requires a CSV file with four columns containing DAY, MONTH, YEAR, and TEMPERATURE. Seasonality is defined according to the austral seasons (Southern Hemisphere), and each event is assigned to a season based on the start date of the temperature exceedance above the 90th percentile. Trend, magnitude, and homogeneity analysis of observed heat waves (ManKendall_Pettit_SenSlope_v1.py) This algorithm evaluates whether there is a statistically significant trend in the observed heat wave occurrences, estimates the magnitude of the trend, and detects potential abrupt changes in the time series. It provides the necessary statistical parameters for analysis, including p-values. The procedure requires a CSV file with two columns: the first representing the time period (e.g., annual, monthly) and the second representing the number of heat wave occurrences for the corresponding analyzed period (e.g., annual, seasonal). Output files of nocturnal (CTN90pct) and diurnal (CTX90pct) heat wave occurrences These datasets correspond to the analysis of minimum and maximum air temperature time series from the meteorological stations of Cabrobó (Cabrobo_CTN90pct.csv and Cabrobo_CTX90pct.csv), Surubim (Surubim_CTN90pct.csv and Surubim_CTX90pct.csv), Recife (Recife_CTN90pct.csv and Recife_CTX90pct.csv), and Garanhuns (Garanhuns_CTN90pct.csv and Garanhuns_CTX90pct.csv), covering the period from 1961 to 2024. The data were obtained from the National Institute of Meteorology (INMET). The datasets indicate periods in which air temperatures exceeded the 90th percentile for one or more consecutive days, enabling subsequent analysis of heat wave occurrences depending on the definition of continuous time periods adopted for heat wave identification.