MODELING LOCAL-SCALE FROST VARIATIONS USING MOBILE TEMPERATURE-MEASUREMENTS WITH A GIS

A stepwise linear regression model for local-scale temperature climate s for clear and calm summer nights is presented for a heterogeneous an d forested area in south Sweden. The spatial variations in measured ai r temperature 2 m above 5.5 km of roads were modelled from the followi ng independent variables retrieved from a geographic information syste m (GIS): sky view factor, altitude, relative relief and presence of pe at soils. The model explains 88% of the measured variations in air tem perature and 87% of the variations in air temperature of a separate va lidation data set collected along 2.4 lan of roads. Distributed model results for a 7.5-km(2) area generates a map with areas delineated tha t are prone to low temperature and frost after clear-cutting. The map gives useful information for efficient implementation of frost protect ion measures in forestry and is based on a simple methodology that can be applied in practical forestry. (C) 1998 Elsevier Science B.V.