Content area
Abstract
This dissertation seeks to explore alternative methodologies for estimating green leaf area index (LAI) and crop developmental stages. Specifically this research [1] developed an approach for creating a Moderate Resolution Imaging Spectroradiometer (MODIS) high spatial resolution product for estimating green LAI on the base of data collected using two different close-range sensors. It was determined that the vegetation indices (VIs) Wide Dynamic Range Vegetation Index (WDRVI) and Enhanced Vegetation Index 2 (EVI2) were capable of accurate estimation of green LAI from MODIS 250 m data using models developed from hyperspectral (RMSE < 0.69 m2 m-2</super>; CV < 33%) or multispectral sensors (RMSE < 0.69 m2 m-2</super>; CV < 34%). [2] Explored a new approach for maximizing the sensitivity of VIs to green LAI. Rather than use one VI, we suggested using multiple VIs in different LAI dynamic ranges. Thus, the sensitivity of the VI to the green LAI was preserved and simpler linear models could be used instead of complex non-linear ones. Two combined vegetation indices (CVI) were presented using near infrared and either the red or red edge bands and were accurate in estimating green LAI. While the red band is more common in satellite sensors, the indices use red edge band were found to be species independent for maize and soybean. The two species-independent VIs used in the CVI were Red Edge Normalized Difference Index (Red Edge NDVI) and Red Edge Chlorophyll Index (CIred edge). [3] Algorithms were developed for estimating green LAI in four vastly different crops (maize, potato, soybean, and wheat) that do not require re-parameterization. The most promising VIs for developing a unified algorithm utilized either a green or red edge bands. [4] It was found that, in addition to traditionally used (VIs), the 2-dimensional spectral spaces (e.g. red vs. green reflectance) were capable of identifying four distinct stages of crop development (e.g. soil/residue, green-up, vegetative, and senescence).
