Submitted to: International Geoscience and Remote Sensing Symposium Proceedings
Publication Type: Proceedings
Publication Acceptance Date: April 1, 2000
Publication Date: N/A
Interpretive Summary: Light reflected from crop canopies over visible (VIS) and near-infrared (NIR) wavelengths provides important clues about crop development and future yield. In practice visible and near-IR reflectance values are used to form vegetation indices that express many reflectance values as one number proportional to some measure of crop development. One such measure is Leaf Area In ex (LAI),which is total leaf area over a fixed surface are divided by that surface area. Another less widely used measure is fractional vegetated area (FVA),which is simply the fraction of a scene occupied by active vegetation . Many index designs have been proposed since the 1970s, and the purpose of our research was to field evaluate a number of recent designs. A field experiment was conducted to collect spectral reflectance, LAI, and FVA data over 4 irrigated and 4 dryland cotton plots. Using the reflectance data, 5 vegetation index designs were evaluated in terms of how they tracked changes in LAI and FVA during the growing season Index response to variation in soil reflectance was also compared. Overall, the Enhanced Vegetation Index (EVI) stood out in terms of being able closely track FVA variation and to remain constant while soil reflectance varied. An analysis based on our LAI our LAI and FVA field measurements was also conducted to estimate the contributions of sunlit and shaded LAI in net canopy photosynthesis. This analysis showed the dominant influence of sunlit leaf area. As sunlit LAI and FVA are roughly equal measures, we propose FVA as a key property determining a canopy's photosynthetic ability. As most of the indices tested here tracked FVA more closely than LAI, we propose soil resistant and FVA-linear indices as the most suited to monitoring crop growth.
Technical Abstract: During the summer of 1998 spectral reflectance, leaf area index (LAI) and fractional vegetated area (FVA) measurements were collected over irrigated and dryland cotton plots. The response of five vegetation index designs were derived from the reflectance data and then evaluated in terms of linearity of response to LAI and FVA variation over the course of the growing season. Index response to variation in soil reflectance was also compared. The indices considered here were the simple near-IR/red ratio (SR), the Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Global Environmental Monitoring Index (GEMI), and the Enhanced Vegetation Index (EVI). All five indices varied more linearly with FVA than LAI over the irrigated plots, with the GEMI showing the most linear response to FVA. Over the dryland plots none of the indices appeared to project onto either FVA or LAI, suggesting a sensitivity threshold somewhere between the two water stress levels. Overall, the EVI stood out in terms of possessing both a high degree of linearity in a biophysical variable (FVA) and high resistance to varying soil reflectance. Using the irrigated LAI and FVA levels observed here, a hypothetical canopy model was proposed to estimate the relative contributions of sunlit and shaded canopy in the absorption of photo synthetically active radiation (PAR). A comparison of shaded and sunlit PAR absorption within this hypothetical canopy shows the dominant influence of sunlit leaf area. As sunlit PAR absorption is a function of FVA measured from a solar vantage point, we go on to propose soil resistant and FVA- linear indices as the most suited to monitoring crop growth.