A major Federal initiative, begun in 1994, provides targeted
funds to dramatically accelerate the automation of the Soil Survey
Geographic Database (SSURGO). The United States Department of
Agriculture - Natural Resources Conservation Service (USDA -NRCS)
is charged with the enormous task of automating soil survey data
nationwide. All data produced under this initiative must conform
to SSURGO Standards. The goal of The National Cooperative Soil
Survey is to produce soil maps that are consistent for the entire
country. Despite the tremendous need for accurate digital soils
data, the availability of this data layer has been so far somewhat
limited.
However, there is a conflict between what Vermont GIS customers
demand for soils data and what Federal standards mandate. Vermont
GIS users want a soils data layer that is compatible with and
integrated with other State GIS layers. Producing soils data that
meet State Standards in regards to scale, photobase, surface water
features, and distribution format is very challenging since national
SSURGO Database Standards required by the NRCS are substantially
different. In Vermont we are committed to filling both needs.
Emphasis will be placed on strategies to minimize introduction
of errors in all phases of production of digital soil maps.
Efforts began in 1983 in Vermont to digitize soils data many years
before the current version of the SSURGO standards were finalized.
We face a potential dilemma: to produce soils data that fit in-state
needs and conform to Vermont GIS standards or to produce data
that comply with SSURGO standards.
In-state GIS users want soils data on the Vermont Mapping Program
Orthophotobase in State Plane Meters. They want the soils layer
at 1:20,000 scale and soils data compatible with and integrated
with other Vermont GIS layers, such as surface water, roads, and
civil boundaries. Yet we can not ignore the Federal mandate, which
requires mapping done on USGS full quadrangles at 1:24,000 or
quarter quadrangles at 1:12,000 scale in UTM coordinates. Must
we abandon our obligation to provide data needed at the local
level to fulfill the quest for a National Soils Database? Our
approach has been to avoid the "all or nothing" mindset.
It is vitally important to have compatible soils data for the
country as a whole. If each state had its own independent standards
use of the digital soils layer across state lines would be difficult
if not impossible.
Development of digital soils data has had a 14 year history in
Vermont. As standards have evolved, methodologies to automate
soil maps have also evolved, resulting in a menu-driven system
that runs through extensive quality control checks using Host
ArcInfo. Two data sets are produced: one that fits Vermont GIS
standards, then this data is transformed to generate the SSURGO
soils data set for national distribution.
What this paper will emphasize is that by using GIS tools, data
can be output to suit many needs. Because it is important to anticipate
the various end products required, certain methodologies must
be followed to allow for flexibility in data output. We have successfully
transformed data produced according to the Vermont GIS standards
to fit the SSURGO Certification requirements for three counties.
The system we have developed is neither inordinately time-consuming
nor cost prohibitive. In some cases it is possible to have your
cake and even get to it eat too!
The choice of photobase is one of the most crucial decisions in
database development. The photobase that underlies data layers
dictates the projection and coordinate system. To some extent
it sets the stage for the range of scales which can be used when
data layers are drafted and put into digital format. We all know
from GIS 101 that the whole point of digital maps is that data
layers can be transformed from different scales and projections.
Despite this flexibility, in practical terms it is more efficient
to use one photobase for data development. This is especially
true when two data layers need to be developed in concert with
each other.
The soils data layer is interdependent with the surface water
layer. Thus, soil lines should be built around water features.
Using only one photobase greatly facilitates this process. The
Vermont developmental approach is to select all polygon water
features (appropriate for 1:20,000 scale soil maps) from the existing
surface water layer and put them into the soils layer, as the
initial step in creating a soils layer. All soil lines are snapped
to the polygonal water. In addition, rivers and streams represented
by a single arc also influence considerably the placement of soil
lines. This is particularly important in cases where certain floodplain
soils require a river or stream to be located in them. SSURGO
standards do not mandate integration of the soil layer with a
specific surface water layer.
Coordinating the development of the soil and water layer is a
standard methodology for Vermont's GIS but this has no consequence
when it comes to national standards. SSURGO must take a national
focus and the only two options for allowable photobases are USGS
1:24,000 full quadrangles or 1:12,000 quarter quadrangles. Efforts
are underway to produce Digital Ortho Quads (DOQs) for the entire
nation. This will enable NRCS to have a consistent photobackground
for all soil maps in digital format across the country.
Instead of abandoning the use of the Vermont orthophoto base for
the development of Vermont's soil layer, we transform the data
to the seven and one half minute USGS quadrangle base. State Plane
Meter Vermont Ortho based coverages are appended, reprojected,
and clipped which is a perfect application of GIS functionality.
One key issue to be aware of in using this approach, is to anticipate
where the new coverage boundaries will be located. If this is
not done two potential problems occur. First, soil lines will
often snap onto the new neatline or will be coincident with the
newly clipped coverage boundary. This creates a situation strictly
prohibited by SSURGO standards called "excess vertices on
the neatline". The other taboo for SSURGO standards are polygons
that come to a point in such a way that the that point is located
on a neatline. This situation is referred to as an "unmatched
node".
One does not have to let the luck of the draw control where soil
lines may interact with the clipped coverage boundaries. We use
a technique similar to our way of integrating the soil lines with
the water layer. Before any soil lines are drafted we create mylar
templates that show the surface water and the quadrangle neatlines.
Also plotted on the templates are roads and civil boundaries to
avoid line collisions for cartographic purposes. Soil lines are
pulled back from all these features during the drafting process.
This ensures that the soil arcs are placed cleanly around the
quadrangle neatlines. Extensive editing is unnecessary once the
data has been clipped and reprojected, even though all data development
work is done utilizing the Vermont orthophoto base.
The final consideration when dealing with the photobase issue
is that of scale. Soil mapping is done in Vermont with the knowledge
that the final scale will be 1:20,000. Now that SSURGO requires
1:24,000 scale maps, our solution has been to simply reduce the
scale of the maps. This results in soil maps that have polygons
that are below the stated minimum size for delineations but this
has not proved problematic to date.
Eliminating all polygons below the 5.7 acre minimum delineation
size (for 1:24,00 scale soil maps) would require extensive editing
and massive re-correlation to absorb the undersized polygons.
This would significantly alter the content and usefulness of the
maps to our customers. The other alternative would be to enlarge
the maps to 1:12,000 scale but this, of course, would imply higher
accuracy than the true source scale at which the maps were produced.
In the days of hand produced maps, many errors went undetected
all the way to map publication. Human concentration was the only
tool available to find and correct errors. By modifying traditional
procedures in the creation of soil maps, GIS tools can be utilized
to avoid error-prone methodologies.
One of the most frequently encountered errors in soil maps stems
from labels that are not listed in the final legend but are found
in the maps. These can be introduced at any stage of the map production
process. The soil scientist out in the field perhaps does not
conform to the official field legend; the map compiler may misread
hand drawn field maps which he or she is drafting onto orthophotos,
or may incorrectly correlate symbols; and finally the digitizing
technician can misread or mistype soil labels when soil maps are
being automated.
To locate such errors, a look up table that contains the final
official legend is compared to the soil labels found in the soil
coverage Polygon Attribute Table, by relating the two tables.
Any offending labels can be plotted on check plots. Tracking down
the source of the error may require some detective work to discover
where the error was introduced. This often necessitates going
back to the original field mapping to research the correct label.
If the soil maps have not gone all the way through to publication,
but are at the stage where map compilation is occurring we recommend
doing any needed label transformations by computer with a look
up table. Traditionally, map compilers transferring soil lines
from field maps to orthophotos also concurrently convert field
labels to publication labels manually. At the beginning of a soil
survey, a wider range of soil labels are allowed than once a legend
is finalized, that is when certain soil labels are correlated
(combined with) other labels to streamline the legend. This often
requires erasing soil lines when two units are combined into one.
To sum up, computer label correlation from field symbols to publication
symbols allows for consistent label conversion throughout an entire
soil survey. Prior to running the label conversion a one to one
check of the field map labels to the digital data labels is done,
as well as a screen for any illegal field symbols. After the conversion
to publication symbols, a dissolve is run to eliminate any extra
arcs.
Conversion of existing published soils surveys to computer format
meeting SSURGO standards presents one set of problems. The conversion
of on-going soil surveys to digital format is another issue. There
are many advantages to incorporating GIS methodologies into the
soil map production process early on. Soil scientists could create
topologically correct data with no invalid labels from the outset,
if on-screen digitizing to digital orthos could be done in the
field. Three dimensional representations of soils data and more
integrated use of existing digital data would facilitate mapping.
Many checks for consistency in mapping that are currently never
done could be easily accomplished.
The Herculean task of automating all soil surveys for the nation could overshadow any ideas of innovative new mapping techniques or situations where individual states have methodologies that differ from the National Plan. Webster's Dictionary defines a standard as "something set up as a model to be followed". Standards are necessary to ensure a uniform end product but standards do not need to enforce the methodologies used to create that end product.
Our system allows Vermonters to get the data in the format they
need and it also produces soils data that conform to a uniform
set of national standards and specifications. The goal of SSURGO
standards is to build soils data of the highest quality. Increasingly,
natural resource planners are using a watershed approach for resource
management. Without a uniform national soils GIS layer, the task
of converting independently created data that had not undergone
centralized review would be overwhelming and a major hindrance
in utilizing the data.
"Partnering" seems to be the governmental buzzword of
the 90s. In Vermont it has been our standard operating mode for
the past decade in development of a State GIS database, long before
it was fashionable. Federal, State, University, and Local entities
have pitched in to create integrated digital data for Vermont.
Through coordinated efforts we have succeeded in reconciling the
dilemma of producing digital soils data for local needs versus
Federal mandates by meeting both needs.
Alves, Caroline, Automating Soils Data - An Integrated Approach.
Proceedings: GIS96 Vancouver, Canada. 1996.
Powers, E., and Alves, C. Issues Involved in the Incorporation
of Published Detailed Soil Surveys into a GIS. Proceedings: 11th
Annual Esri User Conference. Volume 1: 397-402. 1991.
USDA, Soil Conservation Service. Soil Survey Manual. Handbook
No. 18. GPO. 1993.
Caroline Alves
GIS Specialist
USDA - Natural Resources Conservation Service
18 Blair Park
Suite 207
Williston, Vermont 05495
Telephone: (802) 878-7402
Fax: (802) 879-3920