Empirical Models for Estimating Air Temperature Using MODIS Land Surface Temperature (and Spatiotemporal Variables) in the Hurd Peninsula of Livingston Island, Antarctica, between 2000 and 2016

In this article, we present empirical models for estimating daily mean air temperature (Ta) in the Hurd Peninsula of Livingston Island (Antarctica) using Moderate Resolution Imaging Spectroradiometer (MODIS) Land Surface Temperature (LST) data and spatiotemporal variables. The were obtained validated Ta from three Spanish situ meteorological stations (AEMET stations), Juan Carlos I (JCI), Johnsons Glacier (JG), (HG), our team’s monitoring sites, Incinerador (INC), Reina Sofía (SOF), Collado Ramos (CR), as well daytime nighttime Terra-MODIS LST Aqua-MODIS between 2000 2016. Two types multiple linear regression (MLR) obtained: each individual station (for JCI, INC, SOF, CR—not JG HG due to a lack data) global all stations. study period, JCI INC relocated, so analyzed both locations separately (JCI1 JCI2; INC1 INC2). general, best Terra data, results CR being followed by JCI2, INC2 (R2 = 0.5–0.7 RSE 2 °C). Model cross validation (CV) yielded similar those data: R2CV 0.4–0.6, RMSECV 2.5–2.7 °C, bias −0.1 0.1 model was also 0.6 °C; its validation: 0.5, 3, −0.03), along with significant (p < 0.05) variables: time (t) two harmonics (sine-cosine), distance coast (d), slope (s), curvature (c), hour observation (H). carefully corrected filtered, respectively, prior analysis comparison. series revealed different cooling/warming trends locations, indicating complex climatic variability at spatial scale Peninsula. variation Locally Weighted Regression (LOESS) method. that not “good quality” usually underestimated which drastically reduced (<5% studied days). Despite shortage “good” MODIS these cold environments, months represented final dataset, demonstrating through are useful long-term generating maps level (1 km2 resolution) Island.