Modeling surface temperature distributions in forest landscapes

Frequency distributions of the spatial variabilityin surface temperature are examined and modeled.Surface temperature data were obtained from ther-mal images, acquired using NASA's Thermal In-frared Multispectral Scanner (TIMS). Two daytimeand two nighttime overflights, from different forestmanagement units on the H. J. Andrews Experi-mental Forest in the Cascade mountains of westernOregon, were made during August 1985. Prior toanalysis, TIMS temperature values were correctedfor atmospheric attenuation and emission of ther-mal radiation using the LOWTRAN-6 algorithm.Surface temperatures ranged from 10 and 60° C.Frequency distributions of these spatial tempera-ture fields exhibited patterns that are closely asso-ciated with the nature of the surface coverage onthe unit. The frequency distribution patterns ofsurface temperature were modeled using a two-parameter beta probability density distribution.The fit of the model was evaluated against theTIMS dataset. The parameters represent a mathe-matical description of the thermal signature of thedistribution. The model's parameters provide a ba-sis for identifying and classifying surface types. Model discrimination among the various surfaceswas good for thermal images taken during the day.The model yielded a comparable, but less distinct,nighttime ranking.