SUGARBEET QUALITY CORRELATION TO LANDSAT CANOPY DATA FROM A LARGE GIS DATABASE

摘要

Models were developed to test correlation of late season, whole-field canopy reflectance characteristics in sugarbeet to root sucrose concentration at harvest. A GIS database of 1089 fields from the 2003 season was used to locate fields in Landsat imagery and to obtain field-average sucrose concentration. Pixel data were extracted for each field from Landsat5 and Landsat7 canopy images acquired July 18, August 19, and September 12, 2003, as well as from bare soil image acquired May 15, 2003. Linear models relating canopy indices on individual image dates were weakly correlated to sucrose concentration. A Green NDVJ consistently provided the highest correlations between canopy characteristics and sucrose among the indices tested. Linear models relating changes in canopy indices between image dates were correlated to harvest sucrose level, and were variety dependent. A change in the Green NDVI between July 18 and September 12 produced correlation coefficients (r) of 0.64 for the varieties B4811 and B4930. Multiple linear regression models using Landsat band values also were correlated to harvest sucrose level and were variety dependent. Correlations using multiple wavelength models for sugarbeet varieties B3945, B4811, and B4930 produced R{sup}2 values of 0.53, 0.56, and 0.41, respectively, from a single September 12 image. The developed models may be useful in strategic harvest planning to maximize the extractable sugar for a sugarbeet crop.