SPACE-TIME INTERPOLATION OF SURFACE TEMPERATURE USING SPDE-INLA: A CASE STUDY WITH ERA5 DATA (2000–2024) IN MATO GROSSO, BRAZIL

INLA, a well-established methodology, was used to interpolate low-resolution data into a higher-resolution map. This approach is combined with stochastic partial differential equations (SPDE), enabling inference of the underlying spatial process in the data, assessment of covariate effects, and modeling of additional sources of variability. The space-time SPDE-INLA model was able to account for the effects of interaction of the spatial and temporal dependencies in estimating uncertainties, enabling the generation of maps for the monthly mean temperature at 2 meters above the surface in Mato Grosso over the years.