Comparison of three remote sensing techniques to measure biomass on CRP pastureland

Abstract

Biomass from land enrolled into CRP is being considered as a biofuel feedstock source. For sustainable production, harvesting, and soil protection, technology is needed that can quickly, accurately and non-destructively measure biomass. Remote sensing of vegetation spectral responses, which tend to be highly responsive to changes in biomass, may provide a means for inexpensive, frequent, and non-destructive measurements of biomass at management relevant scales. A valuable resource for land managers would be a biomass measurement model that could non-destructively measure biomass at different phenological growth stages across multiple growing seasons. The objective of this study was to compare remote sensing-based biomass measurement models using the normalized difference vegetation index (NDVI) and bandwise regression remote sensing techniques to determine which model best measures biomass at different phenological growth stages over multiple growing seasons on CRP pastureland in central Montana. Biomass and plant spectral response measurements were collected over the 2011 (n = 108) and 2012 (n = 108) growing seasons on an 8.1 ha CRP pasture. Measurements were stratified by phenological growth stage and growing season. Half of the data was used to build each measurement model and the other half was used to test the power of each model to measure biomass. Remote sensing-based biomass measurement models were constructed using NDVI measurements from an active ground-based sensor, NDVI measurements from Landsat images, and band combination measurements from Landsat images. All biomass measurement models showed no difference between actual and estimated biomass values (p-value > 0.05). The biomass measurement model using NDVI measurements from Landsat images had the smallest margin of difference between estimated biomass and actual biomass (22 kg/ha + or - 96 kg/ha), followed by the combination of individual spectral bands from Landsat images (128 kg/ha + or - 71 kg/ha), and NDVI measurements from a ground based sensor (182 kg/ha + or - 94 kg/ha). Results indicate remote sensing-based biomass measurement models are accurate at measuring biomass at different phenological growth stages across multiple growing seasons. Land managers can implement remote sensing-based biomass measurement models into their land management strategies to quickly, accurately, and non-destructively measure biomass across a landscape.