engleski

Spatio-temporal interpolation of daily temperatures for global land areas at 1 km resolution

Around 9000 stations from merged GSOD and ECA&D daily meteorological data sets were used to build spatio-temporal geostatistical models and predict daily air temperature at ground resolution of 1 km for the global land mass. Predictions were made for the mean, maximum and minimum temperature using spatio-temporal regression-kriging with a time series of MODIS 8 day images, topographic layers (DEM and TWI) and a geometrical temperature trend as covariates. The model and predictions were built for the year 2011 only, but the same methodology can be extended for the whole range of the MODIS LST images (2001–today). The accuracy of predicting daily temperatures has been assessed using leave-one-out cross-validation ; the standard approach is extended with block approach. The values were aggregated for blocks of land of size 500×500 km 13 to account for geographical point clustering of station data. All computations 14 were implemented in the R environment for statistical computing by combin- 15 16 ing functionality of the gstat package (geostatistical modelling), rgdal and raster packages (raster data loading and analysis), and snowfall package (cluster computing). - removed The results show that the average accuracy for predicting mean, maximum and minimum daily temperatures is RMSE=2°C for areas densely covered with stations, and between 2°C and 4°C for areas with lower station density. The lowest prediction accuracy was observed in highlands (>1000 m) and in Antarctica with a RMSE around 6°C. This automated geostatistical framework could be used to produce global archives of daily temperatures (new generation WorldClim repository) and to feed various global environmental models.

air temperature ; global ; spatio-temporal prediction

nije evidentirano