Kansas Agricultural Experiment Station

A 16-band mult ispectral radiometer (MSR) was used to measure the amount of forage biomass present on several dates in native tallgrass prairie pastures during the 1992 to 1994 growing seasons. Reflectance data collected with the MSR were used a s inputs for a neural network computer program. The neural network used the reflectance data to predict forage biomass. Biomass estimates made with the MSR w ere found to predict actual biomass, as measured by hand-clipping, across all plant growth stages with an error of approximately 6%. Radiometri c determination of biomass is a reliable alternative to hand-clipping and can be accomplished in much less time.


Introduction
Measurement of pasture forage production (i.e., biomass) is essential for determining proper stocking rates and range condition.Curren t methods involve hand-harvesting of forage in some defined area (e.g., 1 m).This 2 procedure must be repeated many times to adequately char acterize the amount of forage in an entire pasture and is extremely slow and laborious.Mul t ispectral radiometry (MSR) has the potential to predict forage biomass much more rapidly.It is based on the principle that every substance absorbs and reflects various wavelength s of electromagnetic radiation (i.e., sunlight) in a manner characteristic of its physica l and chemical structures.The amount of sunlight reflected by a substance is directly proportiona l to its mass .Further development of this technology may allow estimation of chemical characteristics of forages, such as nitrogen.

Experimental Methods
One ungrazed and three grazed tallgrass prairie pastures, located at the Kansas State University Range Research Unit, were used in this study.Three soil types were identified within each pasture before the study began: loamy upland, limestone breaks, and thin claypan.Sampling times were late May, mid June, late July, late August, mid September, and mid October.At each sampling date, biomass on 30 to 120 plots was measured with the multispectra l radiometer (MSR; Cropscan®, mode l MSR -87) .A total of 334 plots was measured.An equal number of samples was collected on each soil type.After a radiometric measuremen t was collected, a corresponding .25 m area was clippe d at ground level, and the 2 forage was dried and weighed to determine actual biomass production.Reflectance information col lected with the MSR was used as the input for a neural network.This is a computer program that simulates the inductive reasoning proces s in human beings.In this case, it was used to predict biomass from reflectance features of the forage.Biomass predicted via the MSR and neural network were compared with the actual weights of clipped samples.

Results and Discussion
Eleven categ ories of information relating to forage characteristics were used as inputs for the neural network (Figure 1).The categories that wer e most important in predicting biomass were reflectance at 510 nm, 610 nm, 660 nm, and 760 nm.The NDVI ( normalized difference vegetati on index; (810 nm -610 nm)/(810 nm + 610 nm) was also important.The soil type and sunlight intensity (IRR) appeared to be less important in the final prediction equation.
Biomass was predicted by the MSR/neural network combination across season and soil type with an overall estimation error of 6.04% (Figure 2).The MSR/neural network appeared to predict clipped biomass very accurately when lower amounts of standing forage dry matter were present.
However, the relationship appeared to become weaker with greater amounts of standing dry matter.This may have been due to the limited number of measurements available for large amounts of biomass and/or to forage growth characteristics.For example, more stem and leaf material is raised above ground level as biomass increases.As a result, leaves and stem material closer to the ground become shaded by the upper parts of the plant and may not reflect sunlight proportional to their surface area.
The MSR/neural network used in this study adequately predicted clipped forage biomass over a variety of forage growth stages and levels of biomass, although the accuracy of prediction appeared to be less with high amounts of forage biomass.Radiometers like the one used in our study should prove useful for rapid determination of forage biomass for stocking rate or rangeland monitoring purposes in the future.

Figure 1 .
Figure 1.Relative Importance of Different Inputs.IRR Is Sunlight Intensity.NDVI Is Normalized Difference Vegetation Index.Numbers Are Wavelengths of Light in Nanometers