The closeness of dependence level between growth environment (abiotic predictors) and forest ecosystem (receptor) indicates accordance or discrepancy between site and forest state. Our forest ecosystem analysis was focused on static model approximation between abiotic predictors with the closest dependence and properties of the receptor at 1×1 km grid in the Czech Republic (Central Europe). The predictors have been selected from natural abiotic quantities sets of temperatures, precipitation, acid deposition, soil properties and relative site insolation. The receptor properties have been selected from remote sensing data, density and volume of above-ground biomass of forest stands according to the forest management plans, and from surface humus chemical properties. A selection of the most dependent quantities was made by combining factor analysis and cluster analysis. The static modelling of the dependences between selected predictors and receptor properties was conducted by canonical correlation analysis. Average temperature, annual precipitation, total potential acid deposition, soil base saturation, CEC, total acid elements and site insolation index closely corresponded to NDVI and surface humus base saturation, C_org and acid elements content at 30% of the analysed grid of forest soils and it indicated forest state within the confidence interval at 69% of the forest soil grid (r_CCA = 0.79; P < 0.00001). The forest ecosystem state that corresponds to the selected abiotic predictors was demonstrated in hilly altitudes. The tested procedure is inconvenient for forest state analysis in floodplains and moorlands. Based on approximation deviations, highland and mountain forests were divided into areas with non-optimum or more optimum ecosystem state than as corresponds to the values of the predictors.

Keywords: forest state monitoring; EMEP-LRTAP; floodplain; mountain forests; canonical correlation analysis.