Validation of an existing potato yield model using airborne multispectral remote sensing data

This research investigated the validation of an existing potato yield model in different environments. The model was tested for the spot-yield samples collected from the potato fields for two years and also for the whole field average yield. Each location of spot yield samples was 3.63 m by 3.04 m in size. The single-date SAVI-Yield Model and the Three-date Integrated SAVI (ISAVI)-Yield Model were applied to estimate the yields. Actual yield and estimated yield were compared. The three-date Integrated SAVI value obtained was applied to the ISAVI-Yield Model and estimated yield was then compared with the actual yield.; The Single-date Potato Yield Model was developed using the image acquired seven to 10 days prior to the full cover. The single date model underestimated the yield for 2003 and overestimated for 2004 due to the difference in timing of image acquisition. Three-date ISAVI-Yield Model overestimated the spot yields as well as whole field average yield. The potential reasons for an overestimation were (1) problems with calibration of the imagery which resulted in SAVI values approximately 0.058 higher at full cover, and (2) the different duration of the crop growth cycle. The model was developed for the growth period of 100 days while the field used for the spot yield and whole field average yield validation had crop growth duration ranging from 111 to 138 days between emergence and vine kill.