COMPARISON OF SPATIAL INTERPOLATION METHODS FOR ATMOSPHERIC CARBON MONITORING

Spatial interpolation is an important tool for prediction of unobserved points in space in the earth and environmental sciences. Three methods for spatial interpolation were atmospheric compared. The first two methods are ordinary kriging and Empirical Bayesian Kriging (EBK). The third method is the Bayesian Transformed Gaussian (BTG) model. The three methods are applied to remotely sensed satellite data of atmospheric carbon dioxide (XCO2) provided by NASA’s Orbiting Carbon Observatory (OCO-2) mission. Cross validation on the data shows that the methods are close in terms of mean squared error (MSE) when applied to XCO2 data.