Agricultural fields in East-Central Illinois, as in
many other regions of the central plains, are predominantly
drained by subsurface tile drains. There are very few records
of the actual locations of many of these drainage systems,
especially those installed more that 75 years ago. The
unavailability of drainage maps makes it difficult to locate
nonfunctioning tile lines, or even to determine the position of
functional systems in cases where additional drains are to be
installed. This paper describes the use of color infrared (CIR)
aerial photographs and GIS analysis for mapping tile lines in
Vermilion County, Illinois. The method appears to be a promising
and cost effective tool as compared to conventional tile probe
methods.
The drain (tile) mapping procedure is based on the fact that the
soil over efficiently draining tile line dries faster than the
soil at other locations in the field and has higher reflectance
in the infrared region of the radiation spectrum. CIR aerial
photographs for the study area were taken in spring, a few days
after a heavy rain storm, converted to digital format, and
subjected to edge enhancement to heighten the sharpness of the
images. A GIS package (IDRISI) was used to overlay soil data,
hydrological parameters, topography, and vegetation cover. The
combination of these map layers made it possible to identify the
layout of functional tile drainage systems. The accuracy of the
method was evaluated by comparing the locations of some of the
systems thus identified with ground-truth data.
KEY TERMS: Color infrared aerial photographs; tile line mapping;
remote sensing; GIS; subsurface drainage.
Flooding during plant growth periods cause extensive damage to
crops. Reduced growth in corn and soybean, related to excess
soil moisture or high water table during the growth season, has
been a common problem faced by Midwestern farmers. High water
table conditions can be alleviated by using subsurface drainage.
Many existing tile lines were laid more than 75 years ago, and
over the years some of these have become nonfunctional. Most
farmers do not have records of the layout of the drainage
systems on their farms. It is difficult, therefore, for them to
optimize the benefits that accrue from installing additional
drain lines to improve subsurface drainage.
Improved subsurface drainage will not only improve crop
production and farm income, but also help to reduce surface
runoff. This reduction will result in reduced soil erosion and
sediment load to streams and water bodies. The reduced sediment
load will enhance wetland functions during non-crop periods,
improve wildlife habitat, and reduce pollution associated with
sediment-bound agricultural chemicals. Traditionally, farmers
have used tile probes to locate tile lines by guessing at their
locations, based on the location of dry and moist spots in the
field. There is indeed a strong correlation between dry regions
and the locations of tile lines if the drains are functioning
properly. However, this correlation if often masked by the
effects of variations soil type or topography. In addition, the
visual clues are difficult to pick up in some soils, and tile
probing is time consuming, labor extensive and uneconomical for
large fields. The remote sensing method under discussion
addresses these limitations of tile probe methods.
Remote sensing can be applied to drainage studies using proxies
or surrogates (Campbell, 1987). Proxies play a significant role
in drainage studies because features such as drainage tiles are
below the ground surface and cannot be seen on remotely-sensed
images. The identification of tile lines on color infrared
aerial photographs or high resolution satellite images must be
accomplished by means of proxies such as topography, vegetation
and soil features that serves as clues for mapping tile lines.
It has been well documented in numerous research and field
studies that remotely-sensed data can be used to monitor
spatio-temporal distribution of land use and vegetation cover
(Verma and Cooke, 1996; Tin-Seong, 1995; Betts et al., 1986;
Buchan and Barnett, 1986; Essery and Wilcock, 1986; Gautam and
Narayan, 1983; Kachhwaha, 1983; Diwvedi et al., 1981; Singh et
al., 1979). Satellite data have been successfully used to map
surface drainage patterns (Barrett and Curtis, 1992; Haralick et
al., 1985; Merritt, 1982). However, in applications, such as the
mapping of tile lines, where high resolution data are required,
satellite data have been found to be mostly inappropriate. In
this study, low-altitude CIR aerial photographs were used.
Reflectance in the infrared (IR) range of the radiation
spectrum is very sensitive to soil moisture content (Hoffer,
1972). Variations in soil moisture and plant vigor show up as
variations in near IR (0.7 - 1.3 mm) and mid IR (1.3 - 3 mm)
reflectance (Lillesand and Kiefer; 1987). Some of the factors
affecting soil surface reflectance include soil moisture
content, soil texture (proportion of sand, silt and clay),
surface roughness, the presence of iron oxide, and organic
matter content. Soil moisture content is strongly related to
soil texture. Coarse, sandy soils are well drained, resulting in
low moisture content and relatively high reflectance, while
poorly drained fine textured soils will generally have lower
reflectance
(table 1). Visual similarity of
reflectance can be
encountered in soils with different moisture content depending
on the combination of the other factors that affect reflectance.
However, this problem can be overcome by computer-assisted
digital image analysis, particularly by separating out the
effects of variations in soil type and ground elevation.
| Soil texture
| Soil moisture
| Spectral reflectance
|
| Sand
| low
| moderate
|
| Silt
| moderate
| high
|
| Clay
| high
| low
|
Tile maps were produced for Champaign and Ford counties, and
for sections of Vermilion county, all in East-Central Illinois.
The images and maps that are presented in this report are for a
259-hectare (640-acre) field in Vermilion County, Illinois. The
main crops is this area are corn and soybean. The soil found on
the field are described in
table 2. These soils are
very
common
in Central Illinois. The poorly-drained Drummer soil is lower on
the landscape than the Flanagan soil, but is higher on the
landscape than the Peotone soil (USDA-NRCS, 1996).
The subsurface drainage (tile) mapping procedure necessitated
the acquisition of color infrared, and black & white aerial
photographs; soil maps; and contour maps. The black & white
aerial photograph were used as a base for defining land use and
vegetation cover.
A Hasselblad 500 EL/M camera and Kodak Aerochrome infrared type
2443 film was used for this study. The flight altitudes were
about 2600 meter (8500 ft). The CIR aerial photographs were
produced during spring (March to April) of 1984. During the
period immediately following the spring thaw, tiles start to
flow and moisture differences on soil surfaces can be detected
in the infrared spectral range. The best photographs can be
acquired on a cloud-free day, two or three days after a rainfall
event that exceeds 5 cm (2 inches). However, it is rarely
cloud-free in the first few days after a suitable rainfall
event, as solar heating of the moist soil surfaces produces
overcast skies. Typically, there are no more than two or three
times each year when the conditions are suitable for producing
the required CIR photographs.
Initially, several of the 70 mm CIR positive slides were
scanned at densities ranging from 300 to 1200 DPI to determine
the optimum density for further analysis. The scanning was done
on flat bed scanner and the scanned image was stored in TIF
(tagged image file) format on compact disks (CDs). Since no
significant qualitative differences were found between 300 DPI
and 1200 DPI images, the final images were stored at a density
of 300 DPI. The 300 DPI scanned CIR digital image for study area
is presented in
Figure 1 .
The soil map, elevation map, administrative boundaries and
surface drainage map for the study area were digitized and added
to the GIS database. These maps were overlaid as necessary to
facilitate easier delineation of tile lines. The elevation map
was derived from a digitized contour map. The sample field is
extremely flat
(Figure 2)
and consequently, there was but little
variation in spectral reflectance that could be attributed to
elevation effects. The soil map for the sample field is shown in
Figure 3 .
The boundaries between soil units were extracted in
digital format from the 1:15,840-scale soil survey map of
Vermilion county (USDA-NRCS, 1996). The Drummer and Flanagan
soils have similar properties and are almost always found in
association with each other. However, the Drummer soils have a
distinctly darker tone than the Flanagan soils. Drummer is lower
on the landscape and has a greater organic matter content.
CIR photographs can indicate drainage patterns either from
inferences from plant stress or from soil moisture (Baber,
1982). Remotely sensed data from tile drained area will have
fairly uniform texture representing good crop condition, or on
bare fields, CIR aerial photographs exhibit a uniformly light
gray tone (Robert and Rust, 1982). The degree of the gray tone
depends on the soil moisture level; dry soils are light gray,
moist soils are gray, and wet soils are dark gray
(figure 4).
Although the gray levels may be different for different soil
types, if a tile is functioning properly, the moisture content
of the soil in the immediate vicinity of a drainage tile will be
less than the soil away from the drainage tiles and landscape
will exhibit variations in reflectance in the infrared region of
the radiation spectrum. Dry soils have higher spectral
reflectance than wet or moist soils. This differential in
reflectance can be used to reveal the location of tile drains if
the effects of soil type and elevation are filtered out. Once
the location of a tile line is known, differences in reflectance
can also be used to determine if sections of the line are not
functioning properly.
The scanned digital CIR images were transferred into IDRISI
format using the BIPIDRISI command. The images were converted
into three separate bands (red, green and blue). The images were
then geometrically corrected based on ground control points from
1:24,000 topographic maps. A minimum of four control point were
required for geometric correction. A linear polynomial equation
with nearest neighbor interpolation was employed in this
process.
Images representing each band, and various combinations of the
three bands were evaluated to see how well differences in
moisture content would be displayed. The result of different
combinations bands is presented in the
table 3 and
figure 5.
The band 1, band 2 and band 3 correspond to the red, green and blue
bands, respectively. The product of Band 2 and Band 3 yielded
satisfactory results for tile line mapping. In this image, the
dry, moist and wet regions in the field can be easily
delineated. This delineation not only helps in identifying the
location of the tile drains, but it can also be used to identify
the regions of a field that would benefit the most from the
installation of additional drains. Such information would have
been difficult to obtain from ground truthing.
The product of the Band 2 and Band 3 images was subjected to
edge enhancement to heighten the sharpness of the image. For
this study, a 3x3 median filter was employed prior to
classification.
| Combination
| Status
|
| Band 1 x Band 2
| Poor
|
| Band 2 x Band 3
| Excellent
|
| Band 3 x Band 1
| Fair
|
| Band 1/Band 3
| Poor
|
| Band 2 / Band 3
| Good
|
| (Band 1-Band 2) / (Band 1+Band 2)
| Fair
|
When the Band 2 x Band 3 image in displayed in Bipolar 16 mode,
at least three distinct moisture zones are observed
(figure 5d).
The images can also be reclassified
to give "wet", "moist"
and "drained" regions, if a relationship between moisture
content and reflectance is developed for a particular soil type.
Once the wet and moist regions are delineated, drainage system
additions can be proposed, based on the locations of the
existing tile lines and of the wet regions. The locations of
partially clogged sections of the tile line can also be
identified. If the correlation between moisture content and
reflectance is known, the area of influence of random tile lines
can be demarcated. Such a demarcation will be an objective of
the next phase of this project. Currently, in the absence of the
required correlation data, no supervised classification of the
images were performed.
In the image shown in
figure 5 (d),
three distinct moisture
regions are observed. The wettest regions appear black and blue,
and the color graduates to reddish yellow in the driest regions.
For the most part, tile lines show up as reddish yellow linear
features. The on-screen digitization option in IDRISI was used
to digitize these features and the results stored in vector
format. The resulting vector files can be used to produce maps
that can be supplied to farmers.
When the vector files were laid over the corresponding soil
maps, it was observed that most of the linear features lay to
the regions of the fields that have Drummer soils. Such an
observation is consonant with the fact that Drummer soils tend
to be darker and appear to be wetter than other soil types.
The accuracy of the prepared tile map was verified by probing
for the drains at the locations indicated on the map. In almost
every instance, the main tile lines were encountered on the
first attempt. It took a little more effort, however, to locate
the laterals, mainly because of their smaller size. The
farmer/property owner also indicated that changes had been made
to the field since the CIR photographs were taken. In some
cases, tile lines have removed or repaired.
Remotely-sensed CIR aerial photographs can be successfully used
to reveal the locations of partially or fully draining tile
lines, and to delineate wet, moist and drained regions in a
field. The time of data acquisition has direct influence on the
efficacy of the process. The CIR aerial photographs should be
acquired two or three days after a significant rainfall event.
The integration of GIS data layers for soil and topography was
very useful in delineation the wet, moist and drained regions
and mapping the tile lines
(figure 6).
Once the location of the
existing tile lines are known, the cost of installation of new
tile lines will be reduced and any further damage to existing
line can be avoided during the installation of new tiles or pipe
lines. One such example of final tile line map is presented in
figure 7.
Without moisture condition maps, it is difficult to make
drainage improvements in an efficient manner. The mere existence
of comprehensive subsurface drainage maps tend to increase the
value of farmland in flat, tiled-drained regions.
The tile line maps can also be incorporated into precision
farming practices. They provide a covariate that can be
considered in developing relationships between chemical
application and yield. This consideration will result in reduced
loss of agrochemical through subsurface drainage water.
Baber, J.J, Jr., 1982. Detecting Crop Conditions with
Low-Altitude Aerial Photography. In J.J Johannsen and J.L.
Sanders (eds.): Remote Sensing for Resource Management. Soil
Conservation Society of America, Ankeny, Iowa, 665 pp.
Barrett, E.C. and L.F. Curtis, 1992. Introduction to
Environmental Remote Sensing. Chapman $ Hall, London, 426 pp.
Betts, N.L., M.M. Cruickshank and R.W. Tomlinson, 1986. An
Evaluation of SPOT-Simulation Imagery for Land Use Mapping and
Ecological Investigation in Upland Area in Northern Ireland,
International Journal of Remote Sensing 7(6):779-790.
Buchan, G.M. and N.K. Barnett, 1986. Remote Sensing in Land Use
Planning: An Application in West Central Scotland, International
Journal of Remote Sensing 7(6):767-778.
Campbell, J.B., 1987. Introduction to Remote Sensing. Guilford
Press, New York, NY, 551 pp.
Diwvedi, R.S., A.N. Singh and K.K Raju, 1981. Spectral
Reflectance of Some Typical Indian Soil as affected by Tillage
and Cover Type, International Journal of Remote Sensing
7(6):767-778.
Essery, C.I. and D.N. Wilcock, 1986. SPOT-Simulation Campaign: A
Preliminary Land Use Classification for 200 sq. km River
Catchment. International Journal of Remote Sensing 7(6):801-814.
Gautam, N.C. and L.R.A. Narayan, 1983. Landsat MSS Data for Land
Use/Land Cover Inventory and Mapping: A Case Study of Andhra
Pradesh, Photonirvachak: Journal of Indian Society of
Photo-Interpretation and Remote Sensing 11(3):15-26.
Haralick, R.M., S. Wang, L.G. Shapiro and J.B. Campbell, 1985.
Extraction of Drainage Networks by Using the Consistent Labeling
Technique. Remote Sensing of Environment 18:163-175.
Hoffer, R.M., 1978. Biological and Physical Considerations in
Applying Computer-Aided Analysis Techniques to Remote Sensing
Data. In P.H. Swain and S.M. Davis: Remote Sensing: The
Quantitative Approach, McGraw-Hill, New York, NY 396 pp.
Hoffer, R.M., 1972. Agricultural and Forest Resource Survey from
Space. LARS #011579, LARS/Purdue University, West Lafayette,
Indiana.
Kachhwaha, T.S., 1983. Spectral Signature obtained from Landsat
Digital Data for Forest Vegetation and Land Use Mapping in
India, Photogrammetric Engineering 49(5):685-689.
Lillesand T.M. and R.W. Kiefer. 1987. Remote Sensing and Image
Interpretation, John Wily and Sons, Inc. New York, 2nd ed, 721
pp.
Merritt, E.S., 1982. The Role of Resource Information Companies:
Bridges to Application. In J.J Johannsen and J.L. Sanders
(eds.): Remote Sensing for Resource Management. Soil
Conservation Society of America, Ankeny, Iowa, 665 pp.
Robert, P.C. and R.H. Rust, 1982. Remote Sensing for Real Time
Agricultural Management in the Corn and Soybean Region of
Minnesota. In J.J Johannsen and J.L. Sanders (eds.): Remote
Sensing for Resource Management. Soil Conservation Society of
America, Ankeny, Iowa, 665 pp.
Singh, A.N., S.J. Kristoff and M.F. Baumgardner. 1979.
Delineating Salt-Affected Soils in the Ganges Plane by Digital
Analysis of Landsat Data, Photonirvachak: Journal of Indian
Society of Photo- Interpretation and Remote Sensing 7(1):35-39.
Tin-Seong, K. 1995. Integrating GIS and Remote Sensing
Techniques for Urban Land-Cover and Land-Use Analysis, Geocarto
International 20(1):39-49.
USDA-NRCS, 1996. Soil Survey of Vermilion County, Illinois.
United States Department of Agriculture, Natural Resources
Conservation Service, Vermilion County, Illinois.
Verma, A. and R.A. Cooke. 1996. Visual and Digital Analysis of
SPOT HRV Data for Land Use and Vegetation Cover Classification
in the Central Rayong Basin, Thailand. Submitted to Geocarto
International.