Intra-annual characterization of the mean temperature of soil covered at 5 and 10 cm depth based on remote sensing information, for Uruguay.

Soil temperature (ST) is an important physical property that influences all soil processes; it is a relevant component in the climate system and impacts terrestrial ecological, hydrological, biogeographical, and biogeochemical processes. The measurement of its variability poses difficulties that have limited studies on the spatial-temporal distribution and predictions of ST. This thesis documents the calibration of a model that utilizes remote sensing information (MODIS LST and NDVI) and solar declination for predicting the mean temperature of soil covered at 5 cm depth (TMSc5cm) and 10 cm depth (TMSc10cm), as well as the application of two validated models for the intra-annual characterization of TMSc5cm and TMSc10cm in Uruguay. During the validation stage, TMSc5cm models with R2 of 0.84 and RMSE of 2.3, and TMSc10cm models with R2 ranging from 0.87 to 0.89 and RMSE ranging from 2.1 to 1.8, were selected. When compared to each other, the predictions of TMSc5cm and TMSc10cm maintained a similar relationship to those presented by in-situ observations at the same depth, with slight differences during months of thermal crossovers. It was observed that there is room for improving the quality of in-situ observations and the criteria for selecting satellite information. Preliminary observations highlighted the relevance of studying the effects of forests, soil and subsoil characteristics, and other sources of variability on ST.