6.4   Remote Sensing of Pasture Mass and Quality

Principal Investigators

Michael Collins, Professor, Agronomy
Tom Mueller, Assistant Professor, Agronomy
David Ditsch, Associate Extension Professor, Agronomy
C.T. Dougherty, Professor, Agronomy

Cooperators

J.W. Wyles, Management Operations, Eden Shale Farm
Charles May, County Agent for Agriculture, Perry County
Larry Clay, Herdsman, D&D Ranch, Chavies, Kentucky

Introduction

            Forage crops occupy approximately 60% of the 13 million acres of land used in agriculture in Kentucky. Optimum use of these areas for forage production is dictated by slope, soil drainage characteristics, soil pH, fertility, and numerous other factors. Cash receipts to Kentucky producers from livestock and livestock products in 1997 totaled $2 billion. In 1997 Kentucky ranked 8th nationally in total cattle numbers. Dairy production is declining rapidly in Kentucky, in large part due to chronically low milk prices. Increased efficiency of pasture use in dairy systems would reduce the cost of milk production and help to reverse this disturbing trend.

Pressures of input costs and variable livestock and milk prices require producers to constantly search for improvements in productivity and/or efficiency of production to remain competitive. Forages, especially grazed forages, provide the lowest-cost source of nutrients for beef and dairy cattle in Kentucky. Much of this potential gain in efficiency of production is lost if too much of the forage produced on a pasture goes unutilized. It has been estimated that only about one-third of the forage produced on Kentucky pastures is actually utilized, the remainder becomes senescent and low in quality. A major contributor to poor utilization of pastures is a lack of accurate information on the amount and quality of forage available at any given time.

Remote sensing and terrain attribute information could provide us with new opportunities to better understand the factors that contribute to variable forage production and utilization and could form the basis for development of new management strategies that reduce waste and convert more of the forage produced into livestock products.

The objectives of this project are:

1)                  To use soil:plant spatial relationships to develop more efficient forage management systems for Kentucky grasslands; and

2)                  To improve forage utilization by using remotely sensed spectral data to provide livestock producers with accurate, timely information on forage availability and quality.

Background

            The basic concepts of precision agriculture for row crops include grid mapping of soil characteristics in the field using Differential Global Positioning System (DGPS) combined with grain yield monitoring during harvest (Shearer et al., 1999). By overlaying soil and grain yield maps, producers can identify areas of reduced yield that may indicate nutrient deficiencies for land units much smaller than the whole field, thereby more precisely meeting the unique management needs of each area.

            These site specific management concepts of precision agriculture might apply to high value forages, such as alfalfa for cash hay but, due to the soil limitations and lower productivity per acre inherent to most pastures, intensive and frequent soil or pest sampling may never be economically feasible in most situations. However, the great economic importance of livestock products produced from Kentucky grasslands warrants full consideration of the possible roles these technological advances could play in pasture and hayland management. Even in the event that detailed remote sensing or other site specific information is not economical for forage managers due to cost, the information that could be gained by improving our understanding of relationships between forage yield, botanical composition, forage quality and soil factors could form the basic for improved extension recommendations to Kentucky producers.

For example, DGPS and spatial analysis applied to typical Kentucky forage situations could provide important new information for development of superior extension recommendations on pasture management. Past research has failed to consider the high degree of variation in slope, aspect, drainage characteristics, as well as pH and nutrient levels in developing recommendations for pasture management. Newly available technologies for viewing and analyzing spatial variation in terrain characteristics and soil properties should allow great advances in our knowledge base for optimum grassland management.

Technological advances have made obtaining high quality, multi-spectral remote sensed imagery practical, straight forward, and affordable for agricultural producers. Satellite images (e.g. Agri ImaGIS, Spot Image Corp) and aerial images (e.g. Mercator, Emerge) are available at a variety of resolutions in digital formats. Figure 1 gives an example of remote sensed imagery of the type that could be obtained for application in grassland management.

Among the parameters successfully determined by remote sensing is plant nitrogen status. Blackmer et al. (1996a) found at the R5 growth stage that crop reflectance at 536 nm (with filtered black and white photography) predicted yield response to N very well (r2= 0.93).  They also found that the ratio of light reflectance between 550 and 600 nm to light reflectance between 800 and 900 nm provided a sensitive indicator of N stress (Blackmer et al., 1996b).  Schepers (1998) found that red and near infrared reflectance data could be used to provide reasonable estimates of crop stress. Accurate information on plant N status would be doubly useful in forage management since, in addition to being a frequently limiting plant nutrient, low N concentrations can also limit livestock productivity.

 The spectral range of information available by remote sensing includes visible as well as near infrared (NIR) and shortwave infrared (Figure 2). Routine forage analysis for N, fiber, digestible energy and other important forage constituents, utilize NIR spectroscopy of finely ground forage tissue (Collins and McCoy, 1997). It may be possible to gain useful estimates of several forage quality parameters using this remotely sensed spectral information.

Research Approach

Two sites will be utilized in this project, a site in Perry County, in southeastern Kentucky located on reclaimed surface mined pastureland and a site on the Eden Shale Farm in Owen County in north central Kentucky.

Site Mapping - Reclaimed Mineland Pastures:  Initial mapping, completed on the Perry County site as part of an earlier project, has revealed large variation in most soil parameters, including pH, available P, exchangeable K, and others (eg. pH map, Figure 3). The main objective of the earlier project was to determine optimum beef cow/calf stocking density for mined land pastures in the Appalachian region of Kentucky (Teutsch et al., 1998). Approximately 320 points were permanently placed at a density of 1 point per acre over pasture treatments that include pasture allowances of 3, 6 and 9 acres per cow/calf unit for mature Angus brood cows and their calves. Trimble GPS equipment was used to locate each permanent sample point and an additional 11,000 points were collected to accurately locate slopes, benches, drains and other features that make up the site. (Trimble, 1996; ArcView, 1996). Each pasture treatment is replicated twice and adjacent ungrazed areas are included as controls.
Funding for the stocking rate study, supported by the Robinson Trust, is expected to cease at the end of the current fiscal year. However, the pastures remain available for this proposed work and as do the 80 bred cows that graze the site. These pastures, with the wide range in biomass density created by the three-fold range in stocking rates, should provide an ideal situation for assessing the potential for remote sensing information.

 Soil samples will be collected annually in spring, by compositing several cores taken within a 5-meter radius of each permanent sample point. Each sample will be analyzed for pH, potential acidity, P, K, Ca, and Mg. These maps will also be used to describe and present information on soil types, soil fertility, slope, herbage mass, cover, botanical composition, waterways, erodibility, etc., as described recently by Teutsch et al. (1998). Biomass from two randomly selected 1 ft2 areas within a 5-m radius of each sample point will be collected three times each season; spring, summer and fall (May 15, July 15, September 15) in order to represent the full range in forage yield, species contribution to the sward, plant maturity, plant moisture status and other important variables.

Site Mapping - Permanent Grassland Pasture Site: Eight 1-ha tall fescue pastures at the Eden Shale farm in Owen County, which were established 30 years ago, severely infected with Acremonium coenophialum, will be used as the second site. We also expect to encounter large variation in soil and plant characteristics on the permanent pasture site on undisturbed soil because grasslands are generally relegated to the most variable fields on a farm that are not suited to row crop production. Because of the smaller total area in the second site, two hundred fixed sample sites will be located with GPS at 25 m intervals (25 per hectare) (Goovaerts, 1999) (excluding fence lines). Comprehensive management, climate, and production records are available for the Eden Shale farm over the past 30 years. This information should prove useful in interpreting spatial variation in soil parameters and in pasture responses.

Calibrated non-destructive procedures (sward surface height, rising plate, and electronic pasture probe) will be used to estimate herbage mass, cover, and botanical composition (Dougherty, 1999) at each sample point three times each season as previously described. Vegetation will be sampled in spring, summer and fall (May 15; July 15; and September 15). As for site one, remote sensing spectral data will be obtained in conjunction with each field sampling.

Remote Sensing Data

Ortho corrected 1-m resolution remote sensed imagery will be purchased from Mercator GIS and Environmental Corporation in Houston, TX. Each image has four bands. The wavelengths of the bands can be user-specified. In addition, aerial photos will be collected near the time of each sampling event. These photos will be scanned, digitized and geo-referenced. At each location where we have ground information, remote sensing software will spear through each of the images and determine the number of counts associated with each band. The SPOT (Satellite 20-m) system provides visible (green, red) and NIR spectral ranges, and the spot 4 system provides mid IR data. The Mercator (3-m aerial) system provides visible (red, green, blue) and NIR data.

Forage Quality Analysis

Forage samples will be dried at 65º C and ground to 1 mm size prior to composition analysis. Laboratory analysis of forage samples will include moisture (100º C), total nitrogen by the Kjeldahl procedure (Bradstreet, 1965), neutral detergent fiber, acid detergent fiber, acid detergent lignin, cellulose and hemicellulose using the detergent system (Goering and Van Soest, 1970; Robertson and Van Soest, 1980. In vitro dry matter disappearance will be determined as a estimate of in vivo digestibility of forages (Tilley and Terry, 1963; Marten and Barnes, 1980). Typical coefficients of variation for repeated NDF and ADF analyses on forage samples are 2 to 4. In vitro determinations are inherently more variable due to the use of rumen inoculum. Near infrared reflectance spectra will be collected for each ground forage sample (1100 to 2500 nm). Calibration software will be used to select a portion of the total for calibration analysis using the methods described above. Calibration equations for prediction of forage quality constituents will be conducted using equations developed from laboratory analysis for N, NDF, ADF, ADL, Cellulose, Hemicellulose, and IVDMD.

Expected Benefits

            Information gained in this research will be summarized and disseminated to producers through field days, winter meetings, and extension publications. Grazing schools are held annually at several locations in the state where producers and county agents gather to learn up-to-date information on pasture management. We will incorporate information learned through the proposed research into these training sessions.

            The potential impact of this work on livestock production is large. If even a 10% improvement over current estimates of forage utilization in the state resulted from adoption of recommendations resulting from this work, the impact on farm receipts would exceed $200 million annually. Significant findings such as those that could result from the proposed work are also very valuable when competing for future federal grant funds.


Deliverables

            Deliverables would occur primarily in the second year of the project, since data can be unreliable. These would include anticipated use of the information at 3-5 field days; 3 grazing schools, and 3 winter meetings. Publications resulting from the work will include extension publications written to provide information directly usable by Kentucky livestock producers.