The spatial interpolation of point temperature measurements in Amalia pilot site

Bachelor of Science Rodrigo Manzanero Villoria

Master's thesis

Abstract:

Temperature fields exhibit strong spatial variability driven by environmental and topographic factors. Accurate spatial interpolation of high-frequency temperature observations is therefore essential for understanding local thermal dynamics. This study evaluates geostatistical approaches for interpolating temperature measured at 15-minute intervals from a sensor network deployed at the Amálie pilot farm in the Czech Republic. Two empirical variogram estimators were compared and four theoretical variogram models were fitted to characterize the spatial covariance structure. The spatial domain was divided into four landscape units to account for differences in monitoring network density and environmental conditions. In some landscapes, seasonal and diurnal stratification was applied to define regime-specific subsets, while others were treated as stationary. Two geostatistical interpolation methods—Ordinary kriging (OK) and kriging with external drift (KED)—were implemented to reconstruct spatial temperature fields. Elevation and other terrain attributes derived from a Digital Elevation Model (DEM) were tested as auxiliary predictors. Results indicate that incorporating topographic information improves interpolation accuracy in some landscapes, while OK performs better in others. The optimal neighborhood size was selected using leave-one-out cross-validation (LOOCV), and model performance was evaluated using the mean absolute error (MAE) and root mean square error (RMSE). The optimal variogram models and interpolation strategies were identified for each landscape unit, enabling the reconstruction of high-resolution temperature fields across the study area at 15-minute temporal resolution.

Keywords

kriging with external drift, high-frequency temperature observations, ordinary kriging, spatial interpolation, variogram modeling

Language used: English

Date on which the thesis was submitted / produced: 30. 3. 2026

Thesis defence

• Date of defence: 20. 5. 2026
• Supervisor: doc. Ing. Petr Máca, Ph.D.
• Reader: Luděk Bureš, Ph.D.

Czech University of Life Sciences Prague

Faculty of Environmental Sciences

Master programme:
Environmental Modelling