Date of Award
Program or Major
Natural Resources and Environmental Studies
Doctor of Philosophy
The MOderate Resolution Imaging Spectroradiometer (MODIS) provides an unprecedented opportunity to monitor and quantify seasonal changes of vegetation and phenology. MODIS has the potential to improve the estimation, which is based on the algorithms for the NOAA Advanced Very High Resolution Radiometer (AVHRR), of biophysical/biochemical variables of vegetation. My doctoral study improves estimation of gross primary productivity (GPP) through two aspects: first, my study improved the detection of vegetation phenology by distinguishing MODIS contaminated observations and contamination-free observations, and secondly, I inverted the fraction of absorbed photosynthetically active radiation (PAR) by chlorophyll using radiative transfer models and daily MODIS data.
My dissertation has five aspects: (1) to develop a procedure to distinguish atmospherically contaminated observations, snow contaminated observations and contamination-free observations; (2) to monitor vegetation phenology using reflectance of the seven MODIS spectral bands for land and relative vegetation indices; (3) to clarify the concepts of fractions of PAR absorbed by canopy, leaf and chlorophyll; (4) to explore the potential of estimating the fractions of PAR absorbed at different scales; and (5) to check if vegetation seasonal MODIS spectral variations during plant growing season are only due to vegetation's anisotropic nature.
A procedure to extract contamination-free daily MODIS observations is proposed and developed. It has been employed for the Harvard Forest site, the Howland Forest site, the Walker Branch Watershed Forest site, the km67 Forest site in tropic, a soybean site in Nebraska, the Xilingol grassland site in China, the Bartlett Experimental Forest site, and two broadleaf deciduous forest sites in Missouri. The extracted MODIS signals (reflectance and vegetation indices) provide rich information for interpretation. The richness of information from the results goes beyond the widely used normalized difference vegetation index (NDVI) and leaf area index (LAI). The more precise phenology information can be used for seasonal GPP estimation.
The concepts of fractions of PAR absorbed by canopy, leaf and chlorophyll are described. I extracted fraction of PAR absorbed by chlorophyll for the Harvard Forest site, the Bartlett Experimental Forest site and the two deciduous broadleaf forest sites in Missouri using a coupled canopy-leaf radiative transfer model and daily MODIS data. Metropolis algorithm is used to invert the variables in the radiative transfer model. It provides posterior distributions for individual variables. Some of the inverted variables have been partly evaluated though validation for all variables is extremely expensive. Using the values of inverted variables of the two forest sites in Missouri, I calculated reflectance for the seven MODIS spectral ranges with real MODIS viewing geometries through whole growing season. I found that there should be other factors, except vegetation's anisotropic nature, due to seasonal MODIS spectral variations of the forests during the plant growing season.
My study suggests that in addition to measurements of canopy-level variables (e.g., LAI), field measurements of leaf-level variables (e.g., chlorophyll, other pigments, leaf dry matter, and leaf water content) will be useful for both remote sensing and ecological research.
Zhang, Qingyuan, "Improving estimation of gross primary productivity of terrestrial ecosystems" (2006). Doctoral Dissertations. 332.