Geographia Technica, No. 1/2009, pp. 50 - 56

USING LOW-PASS FILTERS TO IMPROVE REGRESSION MODELS FOR CLIMATE VARIABLES

 R. Lhotellier, C.V. Patriche

ABSTRACT - Our study presents a simple G.I.S-based technique, which can be successfully used to improve the regression spatial models of climate variables, especially in the case of air temperature. This technique is a low-pass filter which calculates the means of the predictors within moving windows of different sizes. In this way, we are able to account for more general terrain characteristics, depicted from a larger area surrounding the meteorological stations. This information is sometimes more relevant for the spatial variation of some climatic elements then the strictly local terrain information. Our study proves this point in the case of mean monthly minimum and maximum winter temperatures in the French Alps.

Keywords: low-pass filter, climate, regression, G.I.S, French Alps

Full article:lhotellier_using_lowpass_filters_to_improve.pdf