Evaluation of ERA5, FLDAS and AIRS temperature data in Central Java and Its utilization as a basis for climate-based policy

Konten Artikel

Increasing global temperatures over recent decades are evidence of global warming. Central Java Province, Indonesia, located in the equatorial region, may also be impacted by global warming. To investigate the rise in air temperature, long-term climate analysis is necessary, utilizing accurate air temperature data. Therefore, this study examines the accuracy of gridded surface air temperature datasets, ERA5, FLDAS, and AIRS, against observational data obtained from six meteorological stations representing the northern coastal areas, central highlands, and southern coasts of Central Java. The Bilinear interpolation method was applied over the period 2009-2024. Statistics evaluation used to prove bias, mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE). Results show that FLDAS has the highest accuracy, with MAE, MAPE, and RMSE of 0.79°C, 2.92%, and 0.91°C, for all regions. ERA5 and AIRS show comparable accuracy. MAE, MAPE, and RMSE of ERA5 are 1.54°C, 5.59%, and 1.76°C, respectively while for AIRS are 1.33°C, 5.19%, and 1.75°C, respectively. The best agreement with observations data in the highland and southern coast regions with the MAE and RMSE of 1.54°C and 1.61°C, respectively. However, AIRS data has a large error in highland areas with MAE, MAPE, and RMSE of 3.35°C, 14.45% and 3.46°C, respectively. FLDAS and ERA5 tend to underestimate, as shown by the bias of -0.77°C and -1.52°C, respectively. Scatter plot, annual cycle, and Taylor diagram analyses consistently supported the statistical findings, highlighting that FLDAS and ERA5 more accurately represented seasonal and spatial temperature variations compared to AIRS for all area conditions. Furthermore, ERA5 offers the longest and most continuous historical record, making it more suitable for studies of long-term climate change and global warming. The findings highlight the importance of selecting datasets based on both accuracy and data availability for climate research and policy development in Indonesia.