|Beeri, Ofer - UND, GRAND FORKS, ND|
|Frank, Albert - RETIRED, USDA-ARS|
Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 23, 2007
Publication Date: April 10, 2007
Citation: Beeri, O., Phillips, R.L., Frank, A.B., Hendrickson, J.R., Kronberg, S.L. 2007. Estimating Forage Quantity and Quality Using Aerial Hyperspectral Imagery for Northern Mixed Grass Prairie. Remote Sensing of Environment 110:216-225. Interpretive Summary: Many environmental questions require accurate quantitative analysis of range capacity, and research is limited by relying on knowledge acquired only at plot scales or for site-specific pastures. We developed spectral tools for estimating total rangeland live and senescent vegetation biomass in physical units (kg ha-1) and for estimating biomass quality (as plant carbon:nitrogen ratio) to calculate the potential amount of crude protein available on a pasture level for northern mixed-grass prairie landscapes. We developed and tested the accuracy of these tools on two ecoregions bordering the Missouri River in North Dakota to reduce site-specificity and to provide a foundation that would support spatially broad applications, such as rangelands covering hundreds of square miles. We quantified range capacity on a pasture-by-pasture basis, as measured by forage quality and quantity, and found results were accurate across ecoregion landscapes to within 8% and 18% respectively. Results suggest remote sensing-based observations can provide quantitative pasture-scale information for large landscapes; a necessary step towards advancing our knowledge of rangeland sustainability from field plots to ecoregions.
Technical Abstract: Sustainable rangeland stewardship calls for synoptic estimates of rangeland biomass quantity (kg dry matter ha-1) and quality [carbon:nitrogen (C:N) ratio] to calculate crude protein (CPc) as mass per unit area (CPm) available to grazing organisms. Rangelands are comprised of photosynthetically active (PA) and non-photosynthetically active (NPA) grasses, which often compromise spectral delineation of biomass at landscape scales. Spectral algorithms aimed to explicitly quantify PA and NPA biomass were developed using HyMap hyperspectral imagery, and forage quality (C:N ratio for PA and NPA biomass) estimated with a previously published algorithm. These independent algorithms for forage (NPA + PA biomass) quantity and quality were evaluated in two northern mixed-grass prairie ecoregions, one in the Northwestern Glaciated Plains (NGGP) and one in the Northwestern Great Plains (NGP). Forage quantity was mapped with 18% relative error and quality with 8% relative error. We combined spectral estimates for biomass and quality to calculate CPm for all aboveground vegetation. We tested for differences within and between landscapes for biomass, C:N ratio, and CPm . We found total biomass was significantly different between landscapes, averaging 5740 kg ha-1 in NGGP and 5030 kg ha-1 in the NGP, but landscapes were similar for CPm and C:N ratio. Average CPm and C:N ratio was 445 kg ha-1 and 32.4 in the NGGP, respectively, and 397 kg ha-1 and 32.0 in the NGP, respectively. Instead, CPm was significantly different among pastures (from 276 to 544 kg ha-1 per pasture), while C:N ratio varied significantly within pastures, from 25.3 to 38.3. Results demonstrate development, accuracy assessment, and application of remote sensing-based algorithms for quantifying pasture-scale variability within the larger, mixed-grass prairie landscape.