Variation of net radiation over heterogeneous surfaces: measurements and simulation in a juniper–sagebrush ecosystem

As part of a larger study of carbon dioxide and energy exchange, energy components in an open-canopied juniper–sagebrush ecosystem located in the semi-arid region of Eastern Oregon were measured with the eddy covariance technique. Daytime net radiation averaged 20–30% greater than the sum of sensible, latent and soil heat fluxes. On cloudless days several days after a rain event the imbalance was ∼200–250 W m−2. At such times, differences between the surface radiation temperatures of soil and foliage were large, and we investigated whether such differences may generate systematic errors in the measurement of net radiation. A point measurement of net radiation above an open-canopied forest ecosystem is uncertain, because vegetation structure around the measurement location can be highly variable. Depending on location, various fractions of the upwelling radiation from the soil are intercepted by vegetation and do not reach the radiometer. To determine the magnitude of this uncertainty, we measured tree locations and dimensions, and surface radiation temperature (Tr) and shortwave reflection coefficients (α) of soils and vegetation in a 100 by 100 m area. Geometrical models generated by ray tracing and rendering software were used to calculate the upwelling radiation that would reach radiometers placed at random locations above the surface. In summer, under cloudless skies the measured radiative surface temperatures of soil and vegetation varied considerably, from a mean of 56°C for sunlit soil to 25°C for shaded soil, and 27–29°C for sunlit and shaded vegetation (trees and shrubs). The mean shortwave reflection coefficient varied little between components (with αv=0.10 for vegetation and αs=0.13 for soil). Spatial variability in upwelling radiation (Ru) arises mainly from component variability at viewing angles from ∼30 to ∼60°, where contributions to Ru are large and variation in fractional cover between radiometer locations is large. Our measurements and modeling suggest that a radiometer deployed from a tower in a small clearing will only be affected slightly by the clearing since only about 10% of Ru arises from viewing angles less than 15° (directly below the radiometer). The spatial variation in the upwelling radiation reaching a sensor above the canopy increases with increasing differences between the radiation temperatures and reflection coefficients of the various ecosystem components. For the radiative properties found at our site, where the radiative temperature of sunlit soil was ∼30°C larger than the temperature of vegetation and shaded components, the spatial variability in the longwave upwelling radiation (Rlu) was less than 20 W m−2. The spatial variation in the shortwave upwelling radiation (Rsu) for the small differences in the reflection coefficient of the ecosystem components was less than 10 W m−2. Consequently, the uncertainty associated with estimating the available energy from a single point measurement of net radiation is not enough to explain the lack of energy closure (200–250 W m−2) in this complex open-canopy ecosystem.