Capitão , Rodrigo José Coelho2026-02-102026-02-102025http://hdl.handle.net/10400.5/116969Tese de mestrado, Ciências Geofísicas, 2025, Universidade de Lisboa, Faculdade de CiênciasAccurate estimates of surface temperature are essential for understanding climate processes and polar feedback, yet Antarctica remains one of the most challenging regions for such measurements. Harsh environmental conditions and the limitations of both satellite retrievals and reanalysis datasets often lead to large uncertainties. This thesis addresses the problem of evaluating how well reanalysis and remote sensing products capture surface temperature in Antarctica, and whether recent model improvements can reduce known biases. Previous studies have shown that reanalysis such as ERA-Interim and ERA5 systematically misrepresent Antarctic surface temperatures, often displaying warm nighttime biases and cold daytime biases, while satellite products are constrained by cloud masking difficulties and overpass timing. Building on this knowledge, the present study additionally investigates whether the newly developed ERA6-Land prototype, which incorporates a multi-layer snow scheme, offers measurable improvements over ERA5. The analysis made in this thesis uses data from 2015–2024 and combines multiple sources: satellite-derived land surface temperature (LST) from AVHRR/Metop (EPS) and MODIS, reanalysis skin temperature (Tskin) from ERA5 and ERA6-Land (IKTR), and in situ observations from Baseline Surface Radiation Network (BSRN) stations. Methods include spatial and temporal aggregation, matchup procedures across datasets, and evaluation of seasonal, diurnal, and spatial variability. Results show that ERA5 performs reasonably well overall but exhibits systematic biases, such as, the underestimation of daytime warming under strong solar forcing and overestimation of nighttime cooling, especially on the Antarctic Plateau. IKTR improves the representation of interior conditions, reducing errors and better capturing diurnal variability, while in coastal regions the differences compared to ERA5 remain modest. The temperature diurnal cycles of reanalysis and in situ datasets show the importance of satellite acquisition time in dataset comparisons, with ERA5 aligning more closely with EPS (morning overpass) and IKTR with MODIS (mid-day overpass). This work demonstrates that IKTR represents a meaningful step forward in modelling Antarctic surface temperatures, while highlighting the complementary role of satellite datasets as independent references.application/pdfengAntarcticaRemote sensingReanalysisLand surface temperatureSnow schememaster thesis