Downscaling latent and sensible heats from ERA5 to Eddy Covariance measurements for a tropical coastal environment

This study was conducted at the Muka Head Station (MY-MKH), on the northwestern tip of Penang Island, Malaysia, within a tropical monsoon coastal environment shaped by land sea thermal contrasts. Eddy covariance (EC) observations were collected from 1 March to 30 September 2022 to characterize latent heat flux ( LE ), sensible heat flux ( H ), air temperature ( T a ), wind speed ( U ), global radiation ( R G ), and sea surface temperature ( SST ). The study evaluated the capability of the ERA5 reanalysis dataset to represent LE and H relative to high-frequency EC observations measured with a 3-D sonic anemometer and open-path infrared gas analyzer. ERA5 fields were downscaled using Statistical Linear Scaling, Quantile Mapping (QM), and Quantile Mapping with Temporal Shift (QM+TS). Performance was assessed using RMSE, MBE, MAPE, correlation, and lag analysis derived from daily averaged EC and ERA5 variables. Overall, QM+TS provided the greatest improvement, reducing systematic biases and strengthening agreement across EC observations compared with raw ERA5 fields. Improved surface flux representation advances understanding of evaporation, moisture availability, and boundary layer development along the Penang coastal margin. Correcting magnitude and timing biases enhances the use of ERA5 for estimating coastal evaporation, diagnosing surface atmosphere coupling, and assessing moisture exchange processes linked to rainfall. These refinements support more reliable hydrological modeling and climate impact evaluations in monsoon-influenced coastal environments. • ERA5 faces challenges in representing coastal surface energy fluxes accurately. • Eddy covariance data benchmarked ERA5 for latent (LE) and sensible heat (H) fluxes. • Quantile mapping and temporal shift downscaling reduced ERA5 bias significantly. • RMSE reduced by 80 % for LE, and MBE was near-zero after bias corrections.