Statistical modeling of sea ice concentration in the northwest region of the Antarctic Peninsula

摘要: Sea ice is one of the main components cryosphere that modifies exchange heat and moisture between ocean atmosphere, regulating global climate. In this sense, it important to identify concentration sea in different regions Antarctica order measure impact environmental changes on region’s ecosystem. The objective study was evaluate performance multiple linear regression Box–Jenkins methods for predicting along northwest coast Antarctic Peninsula. data from May November period 1979–2018 were extracted passive remote sensors including a scanning multichannel microwave radiometer, special sensor imager, imager/sounder. Meteorological variables atmospheric reanalysis model ERA5 European Center Medium-Range Weather Forecasts used as predictor variables, leave-one-out cross-validation technique calibrate validate models. It found both statistical models have similar when analyzing residual analysis results, root mean square error cross-validation, final accuracy standard deviation, these responses being related regionalization area presents strong, homogeneous, stable correlations time series modeled each pixel.