Modelling local-scale frost variations using mobile temperature measurements with a GIS

A stepwise linear regression model for local-scale temperature climates for clear and calm summer nights is presented for a heterogeneous and forested area in south Sweden. The spatial variations in measured air temperature 2 m above 5.5 km of roads were modelled from the following independent variables retrieved from a geographic information system (GIS): sky view factor, altitude, relative relief and presence of peat soils. The model explains 88% of the measured variations in air temperature and 87% of the variations in air temperature of a separate validation 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 that are prone to low temperature and frost after clear-cutting. The map gives useful information for efficient implementation of frost protection measures in forestry and is based on a simple methodology that can be applied in practical forestry. (C) 1998 Elsevier Science B.V.