Location: Oklahoma and Central Plains Agricultural Research Center
Project Number: 3070-12610-001-022-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Sep 1, 2024
End Date: Aug 31, 2026
Objective:
The goal of this pilot study is to develop a framework to quantify broad-extent forage mass and production using fine-scale data derived from drone imagery. Our objectives are (1) to quantify forage mass and forage production at very fine scale along a rainfall gradient and (2) assess the correspondence of very-fine scale forage production to existing broad scale models for forage production that provide the basis for Decision Support Tools developed and supported by the USDA Climate Hubs such as the Rangeland Analysis Platform and Grass-Cast.
Approach:
Our study will focus on three ecoregions where Grass-Cast and the Rangeland Analysis Platform have been developing models and have remote sensing data available. These include the High Plains, the Rolling Plans and the northern part of the Edwards Plateau.
We propose to collect drone-based imagery (RGB and multispectral where possible) at least once a year in three ranches within each ecoregion. The dates and specific locations for data collection will be coordinated with the Rangeland Analysis Platform (RAP) team, Grass-Cast team, and the USDA Southern Plains Climate hub. We plan to locate three ranches along the gradient and collect imagery biweekly to develop forage production curves. To standardize analyzes, we propose that the properties to be sampled have identical slopes, disregarding canyons, steep hills, and mountainous areas. We will use previously tested flight and data collection protocols.
Once data is collected, we will stitch the imagery using structure from motion and develop an orthoimage, a digital terrain model (DSM) and a digital surface model (DTM). We will combine the DSM and DTM to create a canopy height model (CHM). The CHM dataset will be used to estimate forage mass for each ranch. Forage mass collected throughout the growing season will be used to develop forage production curves. Field data will be collected for each flight to validate the CHM model and the forage mass estimation. We will use the orthomosaic to conduct a land cover classification. This will remove woody cover and other non-herbaceous cover from the model and will help estimate forage mass within the area where imagery was collected.
We will integrate the forage mass and forage production information with data derived from the RAP and Grass-Cast database. Drone surveys will be performed primarily during the growing season. We will work in coordination with both the RAP and Grass-Cast team to develop the scaling up and validation methodology to compare forage production curves. Orthoimagery at very fine resolution (2 cm pixel size) will be used to quantify and validate fractional cover values for RAP and Grass-Cast.
