Spectrally Derived Inputs to Crop Yield Models

Solar radiation as an energy source for plants is available only when it interacts with leaves. The ratio of total solar radiation intercepted (SRI) by a corn canopy has been described as a function of leaf area index (LAI) using Bouguer\u27s Law, and is included in models which predict final corn grain yield. LAI for corn may vary greatly over large areas due to differences in planting dates, hybrids, stresses, and plant populations. Measurement of LAI is tedious and time consuming, which limits the use of models requiring LAI to relatively small areas. Spectrally-derived estimates of SRI may more accurately depict conditions in large crop production areas, and permit the application of crop models to large areas. Agronomic and spectral data were collected in 1979 and 1980 at the Purdue University Agronomy Farm, West Lafayette, Indiana. Agronomic data acquired from several planting dates of corn (Zea mays L.) included LAI and final grain yield. Reflectance measurements were made in four wavelength bands using a Landsat band radiometer (Exotech 1OOA). In addition to the data collected at the Agronomy Farm, in 1978 Landsat MSS data were acquired for nine 5 x 6 n. mile segments with commercial corn fields in five states. Ten fields per segment were identified, and their spectral data analyzed. Crop development stages were monitored and final yield estimated for each of these fields. Agronomic variables studied at the Purdue Agronomy Farm including LAI and calculated SRI were regressed on spectral data. In both years studied, SRI was better predicted than LAI, by the Greenness transformation. Spectrally estimated SRI, linearly interpolated and summed from planting to maturity was found to be positively correlated with final grain yield. Measured LAI, however, performed better in two crop yield models than spectrally predicted SRI. The relationships developed from the data acquired at the Purdue Agronomy Farm were applied to the commercial corn fields using Landsat MSS data. In six of the nine cases examined, spectrally estimated SRI, summed from six weeks prior to six weeks post silking, was found to be positively correlated with final grain yield. The use of spectrally estimated SRI alone to predict crop yields is not recommended. Spectrally estimated SRI used in conjunction with ancillary data, including meteorological data, is recommended in applications over large areas where it is not feasible to directly measure LAI