Relations between surface conductance and spectral vegetation indices at intermediate (100 m2 to 15 km2) length scales

Abstract

The theoretical analysis of Sellers et al. (1992) indicates that the relative response of the unstressed canopy conductance (g*c) to changes in incident (nonsaturating) PAR flux (F0) should be proportional to some spectral vegetation indices (SVI), specifically the simple ratio (SR) vegetation index, for vegetation covers of similar physiology and physiognomy; or ∇F ≡ (∂g*c/∂F0) ∝ SR. This relationship was tested using the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE) flux station data set (g*c) and the FIFE Landsat thematic mapper image data (SVI). The flux station data were used to invert a soil‐plant‐atmosphere model (the simple biosphere model (SiB) of Sellers et al., 1986) to derive estimates of g*c separate from the soil evaporation contribution and corrected for the “stress” effects of vapor pressure deficit and soil moisture deficit. The Landsat imagery was sampled to produce SR vegetation index values for small areas (90 × 90 m) centered on each flux station. The derived ∇F and SR values were found to be near‐linearly related on a site‐by‐site basis. Differences between sites are thought to be related to the fractional cover of C3 versus C4 vegetation so that ∇S,F ≡ (∂∇F/∂(SR)) ∝ V3, where V3 is the fractional cover of C3 vegetation. The above equations form the basis for a simple biophysically based model of canopy‐scale conductance. The model was applied on the flux station scale (100 m)2 and was also used to calculate fluxes for the entire FIFE site (15 × 15 km)2; the latter results were compared with airborne flux measurements. It is demonstrated that because the proposed relationship between ∇F and SR is near‐linear, the calculation of evapotranspiration rates for large areas using this model is effectively scale‐invariant.