Hello all. Im working on developing high resolution meteorological surfaces for mountainous environments for use in ecological modeling. Ive briefly reviewed the literature on the topic and it seems that people primarily use interpolation techniques in conjunction with environmental lapse rates particularly for temperature and precip (e.g. daymet, anuspline, etc..). Ive seen some more recent work by Lookingbill and Urban at fine scales that simply use a regression approach taking into account drivers such as potential radiation and topographic position. What are peoples thoughts about the relative strenghs and weaknesses of these two approaches? I recognize that interpolation approaches can handle complex spatial pattern with sufficient ground data and are perfect at measured sites but it is somewhat difficult to incorporate drivers such as radiation into their formulation. On the other hand, regression, particularly non-parametric forms such as GAM seem well suited for incorporating covariables and can include northing and easting to account for spatial pattern. Am I missing something obvious here? Thoughts and comments are appreciated.
Solomon Dobrowski Tahoe Environmental Research Center (TERC) John Muir Institute of the Environment University of California, Davis 530 752 5092