![Sample our Environment & Agriculture journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5934/TOC-Subject-Sample-2021-Environment-Agriculture.jpg"})
Abstract
On-farm crop productivity and yield prediction are valuable for, among others, designing food policies, on-farm and after-farm planning and pricing, and marketing. Whereas existing prediction approaches are generally reliable, these require extensive and tedious field surveys and adequate ancillary agrometeorological data. Remote sensing offers great potential for quick and reliable means of predicting crop yield, but it necessitates establishment of optimal phenological and spectral bands. Given the socio-economic value of common dry bean, we sought to determine the potential of multi-temporal ground-based hyperspectral data to predict yield of three cultivars grown under irrigated and rain-fed regimes. Canopy-level hyperspectral data were collected from the Gadra, Ukulinga and Caledon cultivars at distinct phenological stages and sparse partial least squares regression was used for spectral analysis. Results indicated that variation in yield could be explained at specific growth stages under rain-fed and irrigated conditions. With the exception of the Gadra cultivar under irrigation, models developed with data at the flowering and pod development stage were more accurate. These findings highlight the potential of temporal phenological ground-based hyperspectral data in predicting common dry bean yield under different watering regimes. The findings provide an opportunity for large-scale yield prediction using airborne or space-borne hyperspectral sensors.