Wataru KunimiShinji TakadaYuya ItoLuis Andrés GuillénToru AbeTakuo Suganuma2026-03-222026-03-22202510.1145/3737611.3776627https://doi.org/10.1145/3737611.3776627https://andeanlibrary.org/handle/123456789/79114Urban heat-island (UHI) mitigation requires fine-grained intra-urban meteorology, yet existing interpolation methods rely on fixed search radii or linear time corrections, which blur microclimatic heterogeneity, while dense sensor deployments remain costly. We propose Spatio-Temporal Kriging (ST-Kriging), a probabilistic framework that fuses asynchronous fixed and mobile measurements within a single space–time covariance model, eliminating explicit time correction and reducing dependence on dense sensor grids. Field experiments demonstrate that ST-Kriging achieves high predictive accuracy, with a MAE of 0.29 °C and an RMSE of 0.24 °C—reducing errors by up to 27% compared with existing approaches. These results highlight ST-Kriging as a practical foundation for affordable UHI countermeasures and more broadly for fine-scale environmental monitoring in smart-city applications.KrigingMicroclimateInterpolation (computer graphics)CovarianceEnvironmental scienceRemote sensingProbabilistic logicField (mathematics)Computer scienceLinear interpolationarticle