Retrieval winter wheat ground cover by short-wave infrared spectral indices in field and sensitivity analysis

The ground cover of vegetation indicated light interceptor of plant, and plant productivity. The red and near-infrared vegetation indices had been employed to predict the ground cover of crop, however these vegetation indices were affected by uncertainty factors such as leaf color, crop cultivar and others. This research focused on the feasibility of predicting ground cover by near infrared and mid-infrared spectral indices which were not sensitive to cultivar, fertilization treatment, irrigation treatment and leaf color. In field the digital photographs of wheat canopy were taken vertically 1.5 m aboveground and then spectra were measured by an ASD Fieldspec FR2500 spectrometer. The ground covers were automatically extracted by a novel image processing procedure based wheat and soil background features. The 8 diagnostic spectral bands in infrared and short-infrared region were selected to calculate 56 ratio indices and 28 normalized difference indices. Also 8 popular red and near-infrared vegetation indices were calculated. The general linear model(GLM) was applied to assess the relationship between ground coverage and spectral indices, and to evaluate whether they were sensitive to cultivar, fertilization treatment, irrigation treatment and leaf color. Results show that red and near infrared vegetation indices to predict ground cover were affected by at least one of the four factors mentioned above, however it is promising that some infrared indices, such as R1690/R1450, R1450/R1690, (R1450-R1690)/(R1450+R1690), predict ground cover of wheat well and they are not sensitive to cultivar, fertilization treatment, irrigation treatment and leaf color.