Agricultural Crop Yield Prediction Utilizing Narrowband Multispectral, Temporally-Registered Imagery (A Feasibility Study's Technical Report and Recommendations).

摘要

Selected narrowband multispectral photography, dedicated to detecting levels of stress within corn, was collected regularly at low and high altitudes over eight large agricultural test sites in Central Pennsylvania during the period of April thru October of 1974. A spectroradiometry survey was conducted at the test sites during three intervals of the 1974 corn growth cycle to verify the selection of the multispectral camera system's filters. Superlative ground and image truth were collected regularly to supplement the multispectral photography of the test sites. After analyzing the multispectral photography with additive color analysis techniques, selected multispectral photography of three test sites, spanning the entire growth cycle, was digitized and registered, both spectrally and temporally, using the image processing resources of the Rome Air Development Center's Image Processing Facility. By analyzing the digital, temporally-registered, multispectral image data with weather and ground truth data, utilizing the techniques of multivariate and regression analysis, the feasibility of agricultural crop yield prediction was investigated.