Xem mẫu
CHAPTER 19
Remote Sensing and GIS for Site-Specific Farming
John G. Lyon, Andrew Ward, Bruce C. Atherton, Gabriel S. Senay, and Tom Krill
INTRODUCTION
There are a number of capabilities that have been postulated to be useful in agriculture. These capabilities have been realized through numerous efforts over the years (e.g., Moran et al., 1997). The focus now is to demonstrate the capabilities of available technologies for operational and practical applications.
The hope is that research can demonstrate that available remote sensor, GPS, and GIS tech-nologies can supply good information for management of crops, soils, and waters on a within-field basis, and do so at reasonable cost. The goals are to potentially improve yields while maintaining or improving soil tilth and water quality (Ward and Elliot, 1995; Blackmer and Schepers, 1996; Gowda et al., 1999).
Precision agriculture approaches can be implemented using a suite of technologies. To address within-field management concerns it is necessary to navigate over short distances and small eleva-tions. The advent of Differential Global Positioning Systems (Van Sickle, 1996) now provides for the required precision and accuracy. The need for mapping the spatial data collected through a va-riety of means is met by position information from GPS and mapping capabilities of Geographic Information Systems (GIS)(Lyon and McCarthy, 1995). GIS supplies the maps of soil grid sam-pling, management applications, and on-the-go yield, and can do so for each season and each year of evaluation with high accuracy (Bolstad and Smith, 1995). To supply detail as to crop, soil, and hydrological conditions over the growing season and from year to year, remote sensor data from aircraft or spacecraft are employed (Eidenshink and Haas, 1992; Thenkabail et al., 1992, 1995; Ward and Elliot, 1995; Yang et al., 1998). Use of different parts of the electromagnetic spectrum can help to separate types of crops and weeds, general soil and soil moisture characteristics, and the potential influences of hydrology on soils and crops (Huete and Tucker, 1991; Lyon et al., 1998).
Our work over the last 12 years and the work of others helps to illustrate the experimental and operational capabilities of remote sensor, GPS, and GIS technologies for agriculture. Here, the background to these methods, their results, and options for the future are described and discussed using our experience from three experiments on commercial farms in Ohio, and the experience of other researchers.
241
© 2003 Taylor & Francis
Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 ©
Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing
242 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT
BACKGROUND
In this chapter we focus on the variables which are most useful in site-specific farming and can be best related to remotely sensed data. These techniques and methods can be incorporated with GIS and watershed models to answer important water resource questions. In general, the following variables can be measured in fields and related to remotely sensed measures of light: wet and dry biomass, crop residue, leaf area index (LAI), plant population density, areal cover of crop canopy, concentration or quantity of chlorophyll, leaf tissue nutrient content, crop height, plant moisture content, yield after harvesting, on-the-go yield, soil moisture content, soil texture, soil fertility, soil nitrogen, soil phosphorus, soil potassium, soil water release characteristics, the presence and loca-tion of weeds, wet and droughty areas, and plant stresses associated with diseases, insects, or other factors.
Crop Characteristics
The wet, green biomass of healthy plants can be measured in a number of ways remotely using the visible, near, and middle infrared regions of the electromagnetic spectrum (Singh, 1989; Thenkabail et al., 1994; Senay et al., 2000a, 2000b). A number of people make use of the fact that green plants absorb most of the red light from sunshine for photosynthesis, and reflect very little red light to the sensor above the crop (Price, 1992). In addition, green plants absorb little near in-frared light, but reflect great quantities of it to the sensor above. This differential light reflectance can be used as a tool to identify green plants from the background materials including soils, crop residue, and water.
Conversely, plants that do not perform optimally or are “stressed” will often have less chloro-phyll and be chloretic or yellow (Blackmer and Schepers, 1996). This decrease in chlorophyll can be detected by a decrease in red light absorbance and infrared light reflectance. This differential light reflectance can be used as a tool to identify green plants from the background materials in-cluding soils, crop residue, and water (Figure 19.1).
Residue
An important consideration to many farm management activities is the maintenance of crop residue cover during the nongrowing season. The reasons for this practice include decreasing soil erosion from water and wind, reduction of plowing costs, decreasing water quality problems downstream, and others. The implementation of crop residue cover or conservation tillage prac-tices has been of great interest, as have methods for measuring the extent of this practice (Messer et al., 1991; Jakubauskas et al., 1992). We have conducted work on measuring crop residue to track the implementation of conservation tillage, and to better understand residue influences on spectral responses during the growing season. This work began in the early 1980s and continues today (Thenkabail et al., 1992, 1994; Van Deventer et al. 1997; Gowda et al., 1999).
Crop residue or senescent plants reflect different amounts of light as compared to green plants or soil. It is possible to identify crop residue because of these differences, that include the absence of chlorophyll, presence of nonchlorophyll plant pigments, difference in water holding capabilities as compared to live or green plants, different leaf or stalk structure as compared to live plants, and different water holding, pigment, and structural characteristics as compared to soils. These differ-ences manifest themselves as a distinct spectral signature or differential light reflectance that can be measured in the visible, near, and middle infrared regions of the spectrum (Ward and Elliot, 1995).
© 2003 Taylor & Francis
Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 ©
Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing
REMOTE SENSING AND GIS FOR SITE-SPECIFIC FARMING 243
Figure 19.1. Spectral responses of corn and soybeans during vegetative growth stages, soybeans at maturity and corn residue.
Soils
The reflectance characteristics of soils are related to the parent materials, the texture, the mois-ture content, organic matter, and to a certain extent slope and elevation. Soils are generally lighter or brighter toned than plants and water. The exceptions would include soils of high organic matter or low bulk density or high moisture content. Bright tones of soils can be distinguished from plant materials in the visible and infrared portions of the spectrum. Work on using remote sensing to study a wide range of soil properties has been conducted by many researchers (Hatfield and Pin-ter, 1993; Moran et al., 1997)
An important consideration is that a dense crop canopy will obscure the bright tone of soils. This supplies a ready method to make estimates of canopy closure or leaf area index due to the darkness of the plant biomass obscuring the soil background. Conversely, measures of the pres-ence of bare soil can help assess the absence of crop or crop canopy that has not developed (Lyon et al., 1986).
© 2003 Taylor & Francis
Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 ©
Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing
244 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT
Water in a Crop Environment
Water characteristics of plants and soils can be identified and differentiated from other materials due to the fact that water or wet soils or wet plants have a lower reflectance of light to the sensor above (Lyon and McCarthy, 1995; Ward and Elliot, 1995). In general, the presence of water in any concentration decreases the light reflectance of materials. This is well known for the case of soils, where the presence of either water at the surface, water saturation, or standing water will greatly re-duce the generally relatively bright reflectance of dry or relatively dry soils (Lyon, 1993).
A major use of remote sensor data is in the evaluation of crop moisture conditions. In particu-lar, the middle infrared (approximately 1.5 and 2.2 µm) and to a certain extent the near infrared (approximately 0.7 to 1.1 µm) supply good detail as to relative plant moisture conditions of the leaves and stalk.
METHODS
Several experiments were developed over time to test the capabilities of remote sensor data for precision agriculture. The results of those experiments are presented here and in other publica-tions. In general, the experiments included: evaluations of satellite and ground measurements of crop, soil, and hydrology variables for more than 50 commercial farms in Seneca County of north central Ohio (Thenkabail et al., 1992, 1994; Van Deventer et al., 1997); evaluations of crop, soil, hydrology, and weather characteristics using fine spatial resolution (1 m picture elements or pix-els) and moderate spectral resolution (12 spectral bandwidths) for a commercial farm in Pike County of south central Ohio used in the Midwest Systems Evaluation Area (MSEA) studies (Senay et al., 1998, 2000a, and 2000b); and evaluations of low-cost aerial photographs taken from low altitudes of four commercial farms and one experimental farm station in Van Wert County of west central Ohio, and the MSEA site. Details on the methods to collect and analyze field data for these experiments are reported in several publications (Thenkabail et al., 1992, 1994; Nokes et al., 1997; Senay et al., 1998).
Landsat Thematic Mapper Data
During a three-year period, ground and satellite data were collected and analyzed for commer-cial farms in Seneca County (Thenkabail et al., 1992, 1994; Van Deventer et al., 1997). The goal was to evaluate the utility of Landsat satellite Thematic Mapper (TM) data for crop, soil, residue, and hydrology characteristics. This large experiment involved many field data collections during the growing season as detailed elsewhere. For remote sensing purposes, the research team visited the county on a biweekly basis to collect data during the Landsat satellite overpasses. The result was a very rich data set to evaluate field and satellite measured characteristics of crops, soils, and management practices.
Landsat 5 TM data and now Landsat 7 ETM+ data provides 30 m by 30 m resolution data every 16 days or so, and some type of Landsat data has been available for 25 years. It provides informa-tion in the visible, infrared, and thermal wavelengths.
Airborne Multispectral Scanner Data Collection
Several overflights of the MSEAsite in Pike County were conducted in 1994. The platform was an Aero Commander twin engine aircraft with a camera bay holding the Daedalus Multispectral Scanner (MSS) instrument (model 1260) and an aerial camera by Wild (RC–8). The overflights were made on April 4, July 11, August 15, August 23, and September 15, and corresponded to
© 2003 Taylor & Francis
Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 ©
Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing
REMOTE SENSING AND GIS FOR SITE-SPECIFIC FARMING 245
Figure 19.2. Ohio MSEA soil map (Wu et al., 1996). Source: US Department of Agriculture, Natural Resources Conservation Service, Columbus, Ohio.
early spring and preplanting, early planting and germination, mature crop, and senescent soybean, and early senescent corn crop growth periods, respectively. Ground data were collected close in time to the overflights, by a crew that measured the MSEA on a regular basis in support of a num-ber of studies, and by personnel concerned specifically with the overflights.
Other remote sensor data were collected during 1994 and have been analyzed. These data in-cluded airborne radar data in X, C, and L-bands from a Lockheed P–3 Orion aircraft and sensor supplied by the Navy, and the hyperspectral sensor AVIRIS flown by a NASA U–2 aircraft.
The MSEAeffort is ongoing, and is significant because of the extensive data collection effort in the field. In addition, products such as a Digital Elevation Model (DEM) were developed from a Differential Global Positioning System (GPS) experiment, on-the-go yield measurements were made from a combine with GPS capability, and a very extensive soil type map was developed for the site (as shown in Figure 19.2, Wu et al., 1996), and much of these data were processed for analysis using Geographic Information System (GIS) technologies (Senay et al., 1998).
Low Altitude, Small Film Format Data
The Van Wert County area was flown approximately six times before and during the growing season of 1997. A single engine, six place aircraft was supplied by the local airport flight service for these flights of approximately 45 minutes duration. One or two photographers collected photo-
© 2003 Taylor & Francis
Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 ©
Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing
...
- tailieumienphi.vn
nguon tai.lieu . vn