Date of this Version
Peng, Y. (2012). Chlorophyll-based approach for remote estimation of crop gross primary production: from in situ measurements to satellite imagery. Ph.D. dissertation. University of Nebraska-Lincoln, Lincoln, NE, USA.
The synoptic and accurate quantification of crop gross primary production (GPP) is essential for studying carbon budgets in croplands and monitoring crop status. The objective of this dissertation is to develop a quantitative technique to estimate crop GPP using remotely sensed data collected from close range to satellite altitudes. In this study, a model based on a recently developed paradigm, which relates crop GPP to a product of total crop chlorophyll content and incident radiation affecting vegetation photosynthesis, was justified for the remote estimation of GPP in crops. The model was tested with ground-observed incoming photosynthetically active radiation (PARin) and vegetation indices (VI), a proxy of total chlorophyll content, retrieved from in situ spectral reflectance collected at close range, for GPP estimates in maize-soybean croplands. The results showed that the VI-PARin-based model was able to provide accurate GPP estimates in maize and soybean under different crop managements, field history and climatic conditions. The algorithms using VIs with red edge spectral band were non-species-specific and yielded an accurate estimation of GPP in both crop types with contrasting canopy architectures and leaf structures (root mean square errors, RMSE, below 2.9gC/m2/d and coefficients of variation, CV, below 21%). To estimate crop GPP based solely on remotely sensed data, potential photosynthetically active radiation (PARpotential), which is PARin under a condition of minimal atmospheric aerosol loading, was used. The model, relating crop GPP to a product of chlorophyll-related VI and PARpotential, was applied to Landsat and MODIS 250 m data. The algorithms, based on this model, were calibrated over three Nebraska study sites and validated at AmeriFlux sites in Minnesota, Iowa and Illinois showing acceptable accuracy. This VI-PARpotential-based model was capable of estimating GPP in both maize and soybean with CV below 16% for maize and 21% for soybean, and yielded higher accuracy than the VI-PARin-based model when concentrations of atmospheric gases and aerosols were low-to-moderate.
Advisor: Anatoly A. Gitelson