Developing New Hyperspectral Vegetation Indexes Sensitive to Yield and Evapotranspiration of Dry Beans

Abstract

Dry bean (phaseolus vulgaris L.) field experiment was carried out under the subhumid climatic conditions of Bafra, Turkey, to evaluate the possibilities of using hyperspectral reflectance data for estimating yield and water consumption. Three irrigation management treatments that depend on monitoring soil water content and one rain-fed treatment were subjected to dry beans arranged in a randomized complete block design with three replications. In addition to soil water measurements, hyperspectral reflectance observations were made throughout the dry beans' growing season. Actual crop evapotranspiration values were calculated by using a detailed soil water balance approach. After smoothing the hyperspectral reflectance data, the first and second derivatives of spectra were calculated. Statistical analysis methods were applied to determine the most sensitive wavelengths to dry bean yield and evapotranspiration (ETa) for developing new spectral indexes based on measured spectral data. Furthermore, some of the well know spectral vegetation indexes were calculated. Statistical comparison between spectral indexes and yield showed that the difference of the second derivative of spectra in 749 and 697 nm could be a good indicator (r = 0.998 and RMSE = 0.027) for estimating dry bean grain yield. Similarly, the difference of the second derivative of spectra in 1003 and 717 nm gave the most significant statistical comparison result (r = 1.0 and RMSE = 3.0) for ETa.