Relationship of Red and Red-Edge Reflectance-Based Vegetation Indices with Stalk and Fiber Yield of Energy Cane Harvested at Different Dates.

Real-time optical sensor-based prediction of millable stalk and fiber yield has profound role on effective handling of feedstock during energy cane harvesting. This study was conducted at the LSU AgCenter Sugar Research Station in St. Gabriel, LA from 2013-2015 to evaluate the potential of optical remote sensing technology in predicting fiber yield in energy cane harvested at different dates. Reflectance readings were taken from plots planted to different energy cane variety (Ho 02-113, US 72-114) treated with N rates at 0, 56, 112, and 224 kg N ha^-1. Stalks were collected at three dates: one-, two- months earlier than the harvest date, and at harvest date. The experiment was arranged in split-split plot in a randomized complete block design with four replications. Based on correlation coefficient (r), simple ratio (SR) and normalized difference vegetation index (NDVI) computed from reflectance readings at wavebands 670 nm (red) and 705 nm (red-edge) had good relationship with stalk yield (r = 0.59–0.78) and fiber yield (r = 0.53–0.80) across cane age. Both SR and NDVI had stronger relationship with stalk and fiber yield for cane stalk collected at harvest date. Perhaps this indicates that the applicability and accuracy of fiber yield estimation using vegetation index (VIs) largely relies on cane maturity. The high r values suggest that VIs collected between 6 and 9 weeks after N (WAN) application (approximately June to mid-July) can estimate stalk yield. For fiber yield estimation, the possibility seems limited to ratoon crops only but can be done using VIs collected as early as 6 WAN application. The outcome of this study showed the potential use of optical remote sensing in energy cane stalk and fiber yield prediction. Future research should focus on building a large database combined with data transformation procedures (e.g. incorporation of growing degree days, adjusting VIs based on crop age) for refinement of prediction models.