Abstract
The main aim of the study was to compare the different geostatistical methods used to map the spatial distribution of temperature and to select the optimal method using the example of maximum temperatures in Africa. An exploratory data analysis was carried out, including statistical tests, distribution verification, and identification of outliers. Maps were produced in ArcGIS Pro using deterministic (IDW, LPI) and stochastic (Kriging Ordinary, Kriging Simple, Kriging Universal, EBK) methods with different parameters. Validation was then carried out using the GIS Data Modelling Validator software, analysing the interpolation error parameters. The validation made it possible to compare the different methods and select the most effective one. Prediction, probability and error maps were developed for the selected method. The analysis carried out showed that the appropriate selection of the interpolation method is crucial for the accuracy and usability of the resulting maps. Several key research questions were posed to identify the optimal geostatistical method for spatial mapping of temperature distribution, using data from Africa as a case study. The effectiveness of various methods in this process was investigated, along with the parameters during the data modelling phase that have the greatest impact on the accuracy of temperature predictions. Additionally, an attempt was made to determine to what extent the presence of outliers influences interpolation results. A critical issue was also to establish which method provides the highest predictability and concordance with actual temperature measurements.