I'm comparing some different methods in temperature estimation, i.e. idw, spline, ordinary and universal kriging and cokriging. I'm doing it in geostatistical analyst (ARCVIEW 8.1), nevertheless the low errors (by x-val) presented by all the methods, some give better results. The universal cokriging (covariable altitude) is the one with low errors and r nearest to 1, but the map is very close to the other methods's maps. It doesn't reveal the conexion with altitude like it should (the regression coeficcient between temperature and altitude is high 0.88). The DEM has 1km resolution, but the cokriging's map is smooth like the ones from other methods. What can be wrong? I have already reduced the search elypse in the variable altitude to the minimum and the points to include also (so i'm doing something like colocalized cokriging), but the map doesn't reveal much spacial variability.
Thanks for your attention
�lvaro
Jos� �lvaro Mendes Pimp�o Alves Silva
Ge�grafo - T�cnico de SIG Geographer - GIS Technician
Departamento de Clima e Ambiente Atmosf�rico Climate Department
Instituto de Meteorologia Portuguese Meteorological Institute
Rua C do Aeroporto
1749 - 047 Lisboa
Portugal
Tel: (+351) 218483961
Fax: (+351) 218402370
Email: [EMAIL PROTECTED]
