| |
Bathmaster
An inexpensive system for
bathymetric mapping.
Roger Windhorn
Background
Reservoirs for public water supplies,
flood control and recreation are common across the country. In
Illinois there are 1,222 reservoirs, 162 of which have public
water supply as a primary purpose (U.S. Army Corps of Engineers,
1996). Sedimentation of these structures is a natural phenomenon,
the rate of which is dependent on landuse, landcover, soil, slope
and climatic factors. Maximizing the lifespan and economic investment
of these structures is dependent on controlling erosion and sediment
delivery within the watershed.
Watershed planning is a common activity
of the Natural Resources Conservation Service (NRCS). NRCS typically
works with local communities and planning committees to identify
concerns and develop solutions to address problems within the
watershed. When the watershed contains a reservoir, mapping water
depths and measuring sediment accumulation are common inventories
performed in support of the planning process. The goal is to estimate
the current normal water volume and rate of sedimentation.
How it was done
The traditional method for sampling
water depth required surveying and creating permanent concrete
monuments on the ground along several perpendicular lines (cross-sections)
across the reservoir. This could typically require several weeks
for a survey crew to finish. Cable or rope was then suspended
across the reservoir from adjacent stations. Water and sediment
depths were measured at predetermined intervals along the line
using a hand held probing device (Figure 1).
Due to the time and labor involved,
a minimum of cross-sections were usually surveyed and sampled.
For example, a 100 acre reservoir might have had 8 cross-sections,
sampled every 2.5 feet along the section. The maximum depth one
could sample was limited to 30 feet, the maximum length of a probe
one could manipulate. A manual procedure was then employed to
estimate volume based on the cross-sections (United States Department
of Agriculture, 1975). These surveys were limited to small and
moderate sized lakes, due to the physical limitations of stretching
cable or rope an extended distance. These surveys were also labor
and time intensive and precluded water skiing, fishing and other
recreational opportunities of others for the duration of the sampling
period.
New needs, new tools
In Illinois, there is now an increased
demand for sediment information on all sizes and types of lakes.
With the increased boat traffic on many lakes, a method was necessary
that was easier, quicker, safer, less personnel demanding and
less costly than traditional methods. The Global Positioning System
(GPS) provided the tool to streamline much of the effort. Several
surveys were performed during the summer of 1995 by georeferencing
depth and sediment sample sites using GPS receivers obtaining
Precise Positioning System (PPS) signals. This data was then input
into a GIS for generation of surfaces using various interpolation
methods, and eventual volumetric determinations. This method was
an improvement over traditional methods, but still required manual
determination of depths, and interactive use of the GPS receiver
by an operator. These limitations also restricted the number
of samples that could be obtained and the maximum depths that
could be observed. A more streamlined method was needed.
Bathmaster
As our GPS assisted sampling was proceeding,
we began investigating the possibility of automatically capturing
sensor data along with position, eliminating manual observations
and interactions. We use the real-time mapping package Geo-Link1
for building and updating GIS data and have realized the benefits
of collecting data this way. The Geo-Link package has
an additional version, XDS, specifically designed for capturing
data from external sensors, such as sonar, data loggers, and laser
range finders. Many of those involved in bathymetric mapping are
engaged in underwater exploration, drilling or mapping and maintaining
navigational waters. Sonar devices and software used by these groups
were beyond the budget and requirements of our needs.
XDS
can read the Naval Marine Electronics Association (NMEA) 0183
sentence (Naval Marine Electronics Association, 1997). We needed
to find a sonar within our budget that output the NMEA 0183 data
stream. The Lowrance LMS 350A fit the bill. Our finished
system included a laptop with a serial port PCMCIA card, Geo-Link
XDS, Lowrance sonar, and a Rockwell Precision Light Weight Global
Positioning Receiver GPS receiver capable of receiving PPS. All
components of the system were previously in place except the sonar
and XDS module, so our monetary investment was modest.
Using Bathmaster
A number of trial runs at a local lake
were required to become familiar with the system and develop operational
procedures. During these tests, a case was built to easily transport
and contain the unit during operation
(Figure
2). This was important, as all of our sampling is dependent
on boats loaned or operated by other agencies or individuals.
Since Geo-Link XDS displays background
maps, our first step includes digitization of the project area.
United States Geological Survey 7.5' topographic quadrangle maps,
(Figure 3)
augmented with current aerial photography,
(Figure 4) serve as the base for
developing this layer. These two sources usually are adequate
for providing current and historical surface area estimates. The
most current boundary is used as the background map. The definition
of the sensor type and sample interval is required before mapping.
Samples can be collected at any fixed rate between 1 and 255 seconds.
We select an interval that will provide a posting every 30-40
feet along our sample route, typically 2-4 seconds for the rates
commonly traveled (Figure 5).
The sonar must be set to output depth in meters
to conform to software requirements. Once XDS and the sonar are
set and the GPS receiver has established a lock on satellites,
the current position is displayed in relation to the base map.
At this point, sampling is a matter
of driving the boat along random cross-sections of all navigable
portions of the project area. The screen display is a good reference
for insuring adequate sample distribution. Two problems we have
encountered include the GPS receiver losing lock on satellites
and operational length of batteries. The only recourse when satellites
are the problem is to stop and wait until a lock is reestablished.
It is important to verify the satellite constellation for the
intended sample period before sampling and plan fieldwork accordingly.
In our region, periods lacking visible satellites are usually
less than 30 minutes during a working day. Prior knowledge of
this helps insure efficient use of time. Battery problems are
a larger issue as all components are dependent on them. We use
several rechargeable 12-volt batteries to power everything and
often drain the power after 3 to 4 hours of continuous operation.
This is often enough time to map areas 300-400 acres in size.
For larger lakes, the only proven solution is more batteries.
The ideal solution would be solar power, but we have yet to pursue
this option.
Post Processing
When sampling is completed, data is
input into a GIS package back at the office for processing. The
depth data are supplemented with points of zero depth, derived
from the current lake boundary. (Figure 6)
At this point, any number of interpolation
techniques are employed to develop a surface including, triangulated
irregular networks, inverse distance weighting, spline and kriging.
Usually several interpolations with several methods are performed
and evaluated before selecting the surface that appears to be
the best fit.
When it comes to interpolation, there
is no one correct way. It is an iterative procedure with results
evaluated visually, and quantitatively with cross-validation (Berry,
1997). When local experts are available, it is beneficial to have
them review the map to insure the accuracy of the interpolation.
When an acceptable surface has been developed
(Figure 7),
determination of volume is possible with the following formula:
SUM(resolution)2 x depth x (number of pixels of given depth).
Closing thoughts
This method provides us with better
estimates of current water volumes than any previously used.
It is efficient, repeatable, digital, and by virtue of graphical
output, content rich for all levels of end user. We will use
these techniques as standard operating procedures, while striving
to improve the limitations. The data sets produced provide excellent
benchmarks of conditions at a point in time, and will serve as
a baseline for determining sedimentation rates when coupled with
periodic sampling of project areas.
One factor that continues to vex us
is determination of original volumes. Most available estimates
of this measure vary, sometimes significantly. When original volume
estimates are compared with current volumes, unreasonable rates
of sedimentation are required to get the figures to match. As-built
plans are often hard to find, lacking, or too costly to convert
to digital form. Under ideal conditions, a digital version of
the as-built, of comparable horizontal and vertical accuracy to
current version, would be used to determine original volume. A
simple subtraction of the current volume from the original would
provide volume loss due to sedimentation, and an estimate of sedimentation
rate.
Otherwise, determination of the original
bottom must be made using one of two methods: manual sampling
or acoustic subbottom profiling (LeBlanc, et. al., 1992). Manual
sampling has all of the previously mentioned limitations, but
is definitely affordable. Acoustic subbottom profiling penetrates
sediment and would likely be the most accurate way to determine
sediment volume and distribution. However, their cost can be beyond
the budget of many agencies. The two most viable options for
us in determining original volumes are to pursue conversion of
as-built plans through cost-share agreements with planning committees
or manual sampling. A third option, contracting the services of
those with acoustic subbottom profiling equipment is one we plan
on investigating in the future.
Citations
Berry, J.K. 1997. Justifiable Interpolation.
GIS World, Vol 10, No.2:34.
LeBlanc, L.R., L.Mayer, M.Rufino,S.G.Schock,
J.King. 1992. Marine sediment classification using Chirp sonar. Journal of Acoustical
Society of America. Vol 1, p. 107-115.
Naval Marine Electronics Association.
1997. NMEA 0183 Standard for Interfacing Marine Electronic Devices, Version 2.20. NMEA,
New Bern, N.C.
United States Army Corps of Engineers.
1996. National Inventory of Dams, Water Control Infrastructure. Federal Emergency Management
Agency. CD-ROM.
United States Department of Agriculture.
1975. Sedimentation. Natural Resources Conservation Service, National Engineering Handbook,
Section 3.
Acknowledgements
The authors thank the assistance of
Larry McGuire and Gene Barickman in this effort.
The mention of trade or product names
does not signify an endorsement by the NRCS.
| |
|