Validation of ERA5 and ERA5-Land Reanalysis Data for Daily Air Temperature and Extreme Indices over Bangladesh
Paper is Under Review at Theoretical and Applied Climatology, 2026
Abstract
Reanalysis datasets are often used as an ideal alternative to weather station observations because they provide temporally and spatially consistent data coverage. However, these reanalysis products need validation against weather station readings, as their performance varies regionally. In this study, we compared daily maximum (Tmax) and minimum air temperature (Tmin) from the ERA5 and ERA5-Land reanalysis datasets with 31 ground meteorological station observations of Bangladesh from 1989 to 2019. Both products reproduced observed temperatures well, with overall correlations exceeding 0.89 for Tmax and 0.95 for Tmin. Wilcoxon signed-rank tests revealed that ERA5 significantly (p < 0.05) outperformed ERA5-Land for Tmax across multiple performance metrics, with large effect sizes (> 0.90), whereas the two datasets exhibited comparable performance for Tmin. Both products showed a cold bias in Tmax and a warm bias in Tmin. Seasonal analyses revealed the lowest Tmin correlations during the monsoon season (ERA5 = 0.648; ERA5-Land = 0.531), despite relatively low RMSE and MAE values, suggesting difficulties in reproducing day-to-day variability under relatively uniform monsoon conditions. For absolute extreme temperature indices, ERA5 generally exhibited significantly (p < 0.05) lower values, whereas ERA5-Land more accurately reproduced observed trends in percentile-based extreme indices. These findings indicate that both ERA5 and ERA5-Land provide reliable temperature data for Bangladesh, with ERA5 performing better for daily temperatures and absolute extremes. At the same time, ERA5-Land showed greater agreement with observed long-term changes in percentile-based extremes.
Recommended citation: Anwoy Barua, Sanjib Ghosh, Shyamal Karmakar et al. Validation of ERA5 and ERA5-Land Reanalysis Data for Daily Air Temperature and Extreme Temperature Indices over Bangladesh, 14 October 2026, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-11138659/v1]
View paper
