Abstract for presentation at The 13th Australasian Remote Sensing and Photogrammetry Conference

The remote sensing of vegetation in farming systems (e.g. pasture biomass and growth rate used for feed budgeting) differs from systems like forestry because of management practices including grazing, harvesting, and nutrient inputs. In non-farming systems biomass can be inferred from vegetation growth rates calculated from established methods such light-use efficiency models. However, management practices such as the length and number of grazings result in very different patterns of pasture biomass over time compared to a typical phenological cycle. Also, grazing a pasture is not the same as clear-felling in a forest because the new growth occurs on top of a pasture residue which changes with repeat grazings.
Traditional methods for measuring pasture biomass (e.g. pasture cuts, visual assessments and plate meters) are time-consuming and error-prone, resulting in increased interest in methods which are spatially-extensive, labour-saving, and reliable. The calibration of pasture measurements is complicated by the image pixel resolution, paddock size, edge effects at the fence-line, and problems with accurately geo-locating measurements. Other issues include the growth of pasture between the time of the image and the time of sampling, and recently grazed pastures which need several days re-growth due to factors such as leaf coverage and a visible soil background which affects the accuracy of results.
In this paper we discuss these ground-sampling issues, and suggest optimum sampling timings post-grazing. We examine high-resolution (SPOT-5 and Ikonos) images of a rotationally-grazed perennial dairy farm in Victoria (Australia). At two dates in the growing season we compare remotely-sensed observations of pastures to field measurements of biomass (from pasture cuts), and characteristics such as the number and length of grazings. Our results show that the timing of pasture measurements affects the use of such data for the validation of remotely-sensed estimates of pasture biomass.