Commercial Application Of Remote Sensing For Columbia Basin Potato Crop Estimation

Abstract

Potatoes are a high value specialty crop that require close management attention because market supply strongly influences sales price. Accurate, timely information on crop acreage and condiGon from remotely sensed data is potentially of great value to managers making marketing decisions. While remote sensing has been widely used in agricultural applications its potential for commercial applications has not been fully realized. This paper describes a procedure, developed for Cropix, Inc. (a small business), for efficiently processing remotely sensed data to provide accurate, timely information regarding regional potato acreage. Landsat MultiSpectral Scanner (MSS) and Thematic Mapper (TM) data were used in conjunction with masked, multi-temporal data sets in an unsupervised clustering approach. Overall map accuracies for the MSS data sets averaged 85%. A comparison of the methods developed using MSS and TM were evaluated; classification accuracies for the TM data set averaged lower in the northem portion of the study area. Factors effecting the efficiency of the procedure include data availability and, especially, data turnaround, i.e. the time it takes Cropix to receive data. Further research is quid to improve accuracies, to develop alternative methods for using TM data, and to investigate contingencies for dealing with limited data availability and data turnaround delays.