Tuesday, November 14, 2006

Estimating Net Ecosystem Exchange of Carbon Dioxide using the Simple Ratio Vegetation Index.

Denise Gutzmer, Univ. of Nebraska, 3100 Holdrege, 803 Hardin Hall, School of Natural Resources, Lincoln, NE 68583-0968 and Elizabeth Walter-Shea, University of Nebraska-Lincoln, 3310 Holdrege, 803 Hardin Hall, University of Nebraska, Lincoln, NE 68583-0968.

Estimating net ecosystem exchange (NEE) of CO2 is essential in understanding processes controlling CO2 fluxes. Remote sensing provides a means of extending flux measurements beyond the local scale and can provide a means of monitoring surface-atmosphere fluxes.   The objectives of this study were to demonstrate (1) that the change in NEE with photosynthetically active radiation (PAR), dNEE/dPAR, is proportional to the simple ratio vegetation index (SR) and (2) that the relationship between dNEE/dPAR and SR can be used to estimate hourly NEE, thus showing the potential for estimating regional NEE.  An approach previously published relating the partial derivative of unstressed canopy photosynthesis (Pc*) to PAR (dPc*/dPAR) to the SR was adapted to address NEE of CO2. Hourly NEE, hourly PAR, and midday SR values for irrigated maize and soybean under best management practices for the 2002 growing season were used to create a statistically-based linear model to estimate NEE.  The resulting model was used to estimate hourly NEE for the 2003 and 2004 growing seasons using crop-specific midday SR and hourly PAR values as input.  The slope, dNEE/dPAR, and SR were near-linearly related. A major exception to the near-linear relationship occurred at low SR values. Hourly NEE was estimated using the linear model (RMSE values of approximately 0.25).  The model underestimated NEE at the beginning and end of the growing season for maize and soybean, due to the nonlinear relationship between dNEE/dPAR and SR at low SR values.  Future research should include investigating a nonlinear relationship to relate dNEE/dPAR and SR as the slope dNEE/dPAR increases greatly from the point where SR represents the bare soil condition to the point where the linear relation feature dominates.