12 Soil Erosion
279
photographs, Spomer and Mahurin (1984) developed topographic maps of an
expanding gully area and estimated soil loss of 14 tonne yr-l for a nine-year period
from the area. Dymond and Hicks (1986) used the same photogrammetric principles
on a series of historical aerial photographs for a period between the early 1940's and
1980 and estimated total catchment soil erosion for areas in New Zealand. Several
studies (Sneddon and Lutze 1989; Kirby 1991) have used close range photogrannnetry techniques to quantify soil erosion rates.
These selected examples show the application and benefits of photointerpretation
and photogrannnetric techniques in mapping and quantifying patterns and changes in
patterns of soil erosion on the landscape. Such applications have provided information
that has helped to develop a better understanding oft-h.e soil erosion process.
12.4.2 Model/GIS Inputs
Remotely sensed data have temporal, spatial, and spectral characteristics that can
provide information on landscape variables needed for input to soil erosion and other
natural resource models (Foster 1991; Lane et al. 1992; Agassi 1995). Pelletier (1985)
reviewed the principles of the application of remotely sensed data to USLE. Traditionally, most inputs for soil erosion models (and other natural resource models) have
come from historical records, soil surveys, and field surveys ofland use and conservation practices. Remotely sensed data, digitized soils data, DEM, and ancillary data
used for input to erosion models in a GIS framework can provide valuable insights
for understanding and managing soil erosion, water quality, and natural resources.
Studies in recent years have shown that many inputs for these models can be derived
from aerial photographs or satellite data (Fig. 12.2). As spatially distributed models
of the erosion process become available, remote sensing techniques may be the only
way to collect the spatially distributed data necessary for input. Many recent studies
have shown the potential for using remote sensing data for input to soil erosion
models. A few examples are be given.
Stephens et al. (1985) used aerial CIR to delineate homogeneous soil erosion areas
and to estimate the cropping factor (C), management practice factor (P), slope factors
(L, S), and soil factors (K) for input in the USLE. Using CIR they were able to
estimate soil loss faster and with the accuracy (88 ± 1.2%) needed for farm planning.
Cihlar (1987) used Landsat TM and SPOT data to determine cropping practice factors
(C) for input to the USLE.
Jakubauskas et al. (1992) discussed the application of remotely sensed data for soil
erosion model and used remotely sensed data for input to the Agricultural Nonpoint
Source (AGNPS) model. They concluded that SPOT multispectral data with supervised classification was the most cost effective source of data for catchments greater
141 lan 2 • Photointerpretation of aerial photographs was best for catchment less than
141 km 2 • They also concluded that only on larger catchments does the use of digital
satellite data become economically effective. Fraser et al (1995) concluded that
classification of Landsat TM data provide a rapid, objective technique and an
attractive alternative to photointerpretation for developing data bases for large
catchments. Other researchers have used either photographs or satellite data to
279
photographs, Spomer and Mahurin (1984) developed topographic maps of an
expanding gully area and estimated soil loss of 14 tonne yr-l for a nine-year period
from the area. Dymond and Hicks (1986) used the same photogrammetric principles
on a series of historical aerial photographs for a period between the early 1940's and
1980 and estimated total catchment soil erosion for areas in New Zealand. Several
studies (Sneddon and Lutze 1989; Kirby 1991) have used close range photogrannnetry techniques to quantify soil erosion rates.
These selected examples show the application and benefits of photointerpretation
and photogrannnetric techniques in mapping and quantifying patterns and changes in
patterns of soil erosion on the landscape. Such applications have provided information
that has helped to develop a better understanding oft-h.e soil erosion process.
12.4.2 Model/GIS Inputs
Remotely sensed data have temporal, spatial, and spectral characteristics that can
provide information on landscape variables needed for input to soil erosion and other
natural resource models (Foster 1991; Lane et al. 1992; Agassi 1995). Pelletier (1985)
reviewed the principles of the application of remotely sensed data to USLE. Traditionally, most inputs for soil erosion models (and other natural resource models) have
come from historical records, soil surveys, and field surveys ofland use and conservation practices. Remotely sensed data, digitized soils data, DEM, and ancillary data
used for input to erosion models in a GIS framework can provide valuable insights
for understanding and managing soil erosion, water quality, and natural resources.
Studies in recent years have shown that many inputs for these models can be derived
from aerial photographs or satellite data (Fig. 12.2). As spatially distributed models
of the erosion process become available, remote sensing techniques may be the only
way to collect the spatially distributed data necessary for input. Many recent studies
have shown the potential for using remote sensing data for input to soil erosion
models. A few examples are be given.
Stephens et al. (1985) used aerial CIR to delineate homogeneous soil erosion areas
and to estimate the cropping factor (C), management practice factor (P), slope factors
(L, S), and soil factors (K) for input in the USLE. Using CIR they were able to
estimate soil loss faster and with the accuracy (88 ± 1.2%) needed for farm planning.
Cihlar (1987) used Landsat TM and SPOT data to determine cropping practice factors
(C) for input to the USLE.
Jakubauskas et al. (1992) discussed the application of remotely sensed data for soil
erosion model and used remotely sensed data for input to the Agricultural Nonpoint
Source (AGNPS) model. They concluded that SPOT multispectral data with supervised classification was the most cost effective source of data for catchments greater
141 lan 2 • Photointerpretation of aerial photographs was best for catchment less than
141 km 2 • They also concluded that only on larger catchments does the use of digital
satellite data become economically effective. Fraser et al (1995) concluded that
classification of Landsat TM data provide a rapid, objective technique and an
attractive alternative to photointerpretation for developing data bases for large
catchments. Other researchers have used either photographs or satellite data to