Assignment of the kriging technique with external derivative for a reasoned cartography of the equivalent in snow water:: Application to the basins of the Gatineau river

The geostatistical algorithm of kriging with external drift (KED) is applied to the spatial estimation of snow water equivalent measured at single points. A digital elevation model with a 10-km resolution is used as external drift. Over the dense network of the period of interest (mid-March 1982), which corresponds to the maximum snow accumulation and the beginning of the snow melt in the Gatineau River basin, the KED technique is compared to the univariate ordinary kriging (OK). The results indicate a significant estimation precision improvement when the KED technique is used, notably in the under-sampled and extrapolated zones. A quantitative performance barometer - the root-mean-square (RMS) error - of this method with regards to the various degradation levels of the snow depth measurement network is proposed.