A massive nation-wide effort is underway to automate existing USDA Natural Resources Conservation Service (NRCS) soil surveys to provide digital soil maps to users of soils information. Mapping has not yet been completed for the entire country. What about the areas which have no existing soils data? If traditional methodologies are used it will be a long time before digital data can be made available for areas without completed soil surveys.
NRCS is supporting efforts to use GIS in the field to accelerate production of digital soils data where mapping is on-going. This paper will examine the benefits, technical hurdles and potential methodologies of using GIS at the outset of soil mapping. In Vermont, mapping is not yet complete for three counties. We are currently exploring whether it will be possible to use "portable" GIS in the field to map soils, using available data layers and current technology.
It is possible to speed time of delivery of digital soils data to the public by incorporating GIS methodologies into the soil map production process early on. However, the mapping process itself is still a labor-intensive undertaking. Completing the field work in a timely fashion will require adequate staffing, which is difficult to maintain in an era of government down-sizing.
Until recently, field mapping techniques to produce a soil survey have remained unchanged for the past 100 years. The 3 standard pieces of equipment for a soil scientist include: some type of augering device, paper based maps (aerial photography or topographic maps) to record information on, and a device to measure slope. Ancillary equipment such as pH kits, Munsell Color books and other diagnostic tools are often used. After field mapping is completed, soil lines are transferred from field sheets onto orthophotography. The next stage is digitizing and/or scanning of soil lines and labels. Every step involved in both processes, map compilation and automation opens the door for introduction of errors.
By consolidating the entire process and using GIS functionality to help create maps in the field vast improvements in consistency and accuracy will be realized. More specifically, if a soil scientist can produce "instant digital data" that is topologically correct and passes a series of automatic error checks, the long wait for digital data will be a thing of the past.
We envision soil mapping using ArcView on a ruggedized PC, with real time GPS to provide a live trace of one's location. Soil polygons and special features can be drawn directly on digital orthophotos. Built-in error checking routines can screen for polygons below the minimum size delineation, illegal symbols, and other inconsistencies. Some error checks such as those based on acreage or queries to other data layers mandate GIS functionality, as opposed to a more simplified drafting software package.
ArcView is easy to use, affordable and customizable. It allows a variety of layers to be displayed at once in all possible combinations. Digitizing and editing functionality make it possible to create and modify digital data. Conversion of shape files to ArcInfo coverages is relatively problem-free. The ability to convert to coverages is important since the review process for soils data is built around ArcInfo AML routines.
Some significant technical barriers exist to our proposed methodology. Ruggedized PCs are limited in the greyshades that can be displayed, perhaps rendering the digital ortho with 256 greyshades unusable. The quality of the screen display when used outdoors in direct sunlight may present problems. Battery life may restrict how much can be mapped in a day. In addition, the soil scientist already is carrying a number of pieces of equipment, using a ruggedized PC instead of a light-weight map board will burden the mapper.
ArcView editing tools fall short of ArcEdit in ArcInfo. One outstanding example is the inability in ArcView to easily modify the shape of and attribute "island polygons", of which there are many in soil maps. Another problem is the lack of the ability to PUT polygons from one coverage (or shape file) into another, as one can in ArcInfo. This is useful when integrating an existing digital layer of water polygons with soil lines, allowing them to be snapped to the water layer.
Before a soil scientist ventures out into the field with soil auger in hand, it is necessary to do "pre-mapping". This entails checking out as many maps as possible of the area in question. Time is spent studying stereo-photography and contour maps to interpret landforms. Geology maps of bedrock and surficial deposits are consulted to better understand the geologic influences on soil genesis. Color-infrared photography is another handy reference to pinpoint water features, wetland areas, and vegetation patterns. Road maps are used to strategize how best to access the area.
This work involves juggling many maps, almost always at different scales. Staying located from map to map requires highly refined skills in photo-interpretation and feature recognition. The advantages of having all needed data layers registered to a geo-referenced orthophoto base are obvious.
In Vermont we are fortunate to have a wealth of digital data. The most crucial layer is the orthophoto base. Digital orthos are available for a large portion of the state. Each ortho measures 4000 meters by 4000 meters and has 0.5 meter resolution, making this an excellent base map. The size of the files, 64 megabytes, may prove too unwieldy for ruggedized PCs with limited memory. One incredibly useful by-product of the digital ortho production process is point files which contain Z-values (elevations).
From the elevation points contours can be generated and also TINs (triangulated irregular network). These layers can be shown as additional themes in ArcView to assist the soil scientist in visualizing the landscape from a three dimensional standpoint. We have yet to explore the use of the 3-D Analyst Extension for ArcView, which has great potential for our application. The use of stereo-photography may continue to be an irreplaceable reference, however, exciting new tools will revolutionize how maps can be displayed on the computer screen.
One of the most time-consuming activities in creating soil maps is the delineation of slope breaks. In pre-mapping a rough, initial version is produced to be further refined later. We are hoping to use a TIN which is broken down into the specific percent slope breaks used in soil mapping, as a backdrop and create an initial slope map as a new theme. Cross-referencing between the TIN, contour lines, and the photobackground will be necessary. This would be done in the office on a more powerful PC, to speed drawing time as TINs are very slow to draw.
Once outside the soil scientist can further refine the slope map and start assigning map unit names to the various soil types, modifying the original slope polygon theme. The primary background theme to be used in developing the soils layer would be the digital ortho. Displaying contour lines on top of the ortho would help the soil scientist better understand the landscape. Initially, the only attribute we wish to enter is the map unit label. Other activities related to mapping such as pedon descriptions, logging transects, obtaining filed notes could all be incorporated into a GIS application. This will be phase two of our mapping efforts.
There is a variety of other data layers which would be of use while in the field. Queries could be made to a bedrock geology layer to highlight, for example, all areas of limestone bedrock. Parcel maps would be useful in determining which landowners to contact to obtain permission to map. Road, trail, and surface water layers are all invaluable locational features. Wetland and floodzone maps would be important cross references, when delineating hydric and frequently flooded soils, respectively. Civil boundaries (town lines and county boundaries) should not influence soil delineations but often define the logistical boundaries of where mapping is done. New England has many conflicting versions of civil boundary locations. Digital layers would help point out overlaps and undershoots in boundaries that will effect mapping.
Another tool that has never been at the soil scientist's disposal is a live trace GPS. Having a "bread crumb" trail appear on a digital ortho as one moves across the landscape will allow the mapper to stay located at all times. As soil samples are taken or special features are encountered, the location can be recorded in new data layers (themes). Eventually, using Spatial Analyst one can determine if sampling is occurring in all soil types with the same frequency.
Staying true to the map unit concept will be more easily accomplished with GIS functionality at the soil scientist's fingertips. Currently, many units are drawn which fall below the minimum size delineation. In these cases, the units should be combined into other units and regarded as an "inclusion". It is often very difficult to estimate the exact acreage of a polygon when drawing manually on a paper map. In the digital world a simple query can alert the mapper to enlarge the unit or correlate it into another. As mapping in a county progresses, decisions are made to drop certain map units because of their limited distribution. Much time is spent using a hand planimeter on paper maps to get a handle on the extent and amount of each soil type. Here again GIS comes to the rescue making this arduous task a simple query.
The soil mapping process goes through a "progressive correlation" as a county is being mapped, meaning there is constant evaluation of the fieldwork that is done. The correlation process ensures soils are accurately mapped, named, and classified. Soil map units may be introduced into a survey only to be dropped out later. A "final correlation" process produces the finalized version of the legend. Revisions to the legend result in many label changes and alterations to the linework. It is important to caution users of digital soils data, from a survey in progress, that the preliminary maps are subject to change.
Using GIS in the field as part of the mapping process can speed delivery of our data to the public. The multi-phased traditional mapping process can be condensed into a more integrated process that creates instant digital soils data. GIS tools will improve the quality of the maps we produce. This goal is no longer a far off vision of the future but is not yet a current reality. There are no established methodologies and much experimentation will be needed to define a workable process.
In time, the technology will become more and more transparent, as what was once innovative becomes accepted. GIS methodologies will be applied in the field throughout the mapping process, as opposed to in the office environment in the final phases of map production. It is an exciting time to be at the beginning of a major transition. Soil mapping is a perfect application for GIS. The tools available in the GIS arena have continued to evolve at a rapid rate to allow us to make them an integral part of our mapping process.
Alves, Caroline, Creating Soils Data that meet National Standards and Local Needs. Proceedings: 17th Annual Esri User Conference. 1997.
Soil Survey Staff, Natural Resources Conservation Service, National Soil Survey Handbook, title 430-VI (Washington D.C., U.S. Government Printing Office, November 1995).
Many thanks to Gary Smith and Eileen Powers for all the help over the years. Additional thanks to Bob Gagliuso.