Forest Stewardship Spatial Analysis Project
Missouri Methodology
September, 2004
Project Summary
One purpose of the Spatial Analysis Project (SAP) is to create a data layer for a state that
represents levels of potential benefit from, or suitability for inclusion in, the Forest
Stewardship Program as delivered by state forestry agencies and the U.S. Forest Service.
Private land program and GIS staff from the four states involved in the pilot SAP effort
(Connecticut, Maryland, Massachusetts and Missouri), along with Forest Service
program and GIS staff identified 12 factors which help identify the “Stewardship
potential” of a given piece of land. The factors were differentiated into two groups:
resource potential and resource threats.
The resource potential factors include:
ƒ Riparian Zones
ƒ Priority Watersheds
ƒ Forest Patch Size
ƒ Natural Heritage Data
ƒ Public Drinking Water Supply Sources
ƒ Private Forest Lands
ƒ Proximity to Public Lands
ƒ Wetlands
ƒ Topographic Slope
The resource threat factors include:
ƒ Forest Health
ƒ Development Level
ƒ Wildfire Assessment
Certain lands within any state are not eligible for inclusion in the Forest Stewardship
program. Land use / land cover factors which identify these areas are open water, urban
areas and publicly owned lands. A mask was created to exclude these areas from the
analysis.
Once the 12 factors were identified, each state determined the relative importance of each
of the criteria based on their state-specific conditions. Missouri Department of
Conservation Regional Forest Supervisors and Forestry Central Office staff (19 people
total) ranked the criteria and an average weight was calculated for each.
The 12 layers were then combined in a GIS overlay analysis which took into account the
weight for each factor. The final product was a single data layer which represents the
suitability of the land for inclusion in the Forest Stewardship Program. Possible values
from this analysis range from 0 to 1, with a value of 1 representing the highest level of
suitability. Actual values for Missouri ranged from 0 to 0.948. A natural breaks
classification algorithm was used to break the values into low, medium and high classes.
The result is shown below.
Summary statistics were calculated and a series of maps was then created to display the
data.
While the process outlined above was taking place a parallel effort was occurring. In
order to understand where the Forest Stewardship Program has been previously
implemented, the property boundaries for ownerships with a Stewardship plan were
digitized.
Stewardship plan polygons were then overlaid on the Stewardship potential layer to
assess Stewardship efforts to date.
Forest Stewardship Potential - Weighted Composite
Stewardship Potential
Masked Areas
Low
Medium
High
Data Development
Riparian Zones: Riparian zones
were created by buffering (300 feet
each side) the perennial streams from
the U.S. Geological Survey
1:100,000 source scale Digital Line
Graph files. The buffers are shown
in blue on the map at right.
Priority Watersheds: These
watersheds are considered priorities
for various reasons including: nonpoint source pollution focus areas
(SALT AgNIPS), presence of
threatened / endangered species,
Farm Bill program focus areas, etc.
Watershed polygons were created by
aggregating polygons from 11 and
14-digit U.S. Geological Survey
Hydrologic Unit shapefiles. Source
scale for this data is 1:24,000. The
priority watersheds are shown in
blue on the map at right.
Forest Patches: All forest cover
used in this project was extracted
from the early 1990s National Land
Cover Database (NLCD). This
landcover classification was based
on Landsat Thematic Mapper 30
meter satellite imagery acquired in
1992 & 1993. Five NLCD classes
were used to create a "forest" layer:
deciduous forest, coniferous forest,
mixed forest & woodland,
shrubland, and woody wetlands
(NLCD classes 41, 42, 43, 51, 91).
All forest polygons less than 10
acres in size were deleted to reduce
noise in the analysis, and because
the minimum size tract eligible for the Stewardship Program is 10 acres. Large roads
create discontinuities in forest cover and reduce forest patch size for some wildlife
species. Accordingly, the state-maintained roads were buffered by road type: 100 feet
(each side) for interstates, 55 feet (each side) for U.S. and MO highways, and 38 feet
(each side) for lettered roads. This buffered road layer was then erased through the
private forest lands layer. The erase process has the effect of cutting a hole in the forest
layer wherever the buffered roads occur. Smaller forest polygons were then eliminated
based upon their location in the state. Within the Ozark Highland Ecological Section the
minimum forest patch size was set at 100 acres. 50 acres was used for the rest of the state.
The result is the forest patch layer. This 30m dataset was then resampled to 45m to match
the other layers used in the analysis. The result is shown at right in green.
Natural Heritage Data: Individual
plant and animal records, and high
quality community records from the
Missouri Natural Heritage Database
were used to represent areas of
biological importance. Point and
polygon records were both buffered
by 2640 feet. Old, general historic
records were not used. Records are
created using 1:24,000 scale USGS
topographic maps as the base. These
areas are shown at right in brown.
Public Drinking Water Supply
Areas: Shapefiles from the Missouri
Department of Natural Resources
Public Drinking Water Program were
downloaded from the University of
Missouri Center for Agricultural,
Resource, and Environmental
Systems web site. Source scale for
this data is 1:24,000. Watersheds for
lake & river intakes and buffered (1
mile) surface and well intake points
were unioned into one file and
clipped with the Missouri boundary
polygon.
Private Forest Lands: All forest
cover used in this project was
extracted from the early 1990s
National Land Cover Database
(NLCD). This landcover
classification was based on
Landsat Thematic Mapper 30
meter satellite imagery acquired in
1992 & 1993. Five NLCD classes
were used to create a "forest"
layer: deciduous forest, coniferous
forest, mixed forest & woodland,
shrubland, and woody wetlands
(NLCD classes 41, 42, 43, 51, 91).
All forest polygons less than 10
acres in size were deleted to reduce
noise in the analysis, and because
the minimum size tract eligible for the Stewardship Program is 10 acres. Private forest
land is created by erasing a public lands shape file through the forest layer, creating
"holes" in place of the public lands. This 30m dataset was then resampled to 45m to
match the other layers used in the analysis.
Proximity to Public Lands: A
single shapefile of all public land
ownership in Missouri was created
by merging individual agency files
and then buffering the polygons by
2640 feet. The source scale for all
the individual shapefiles is
1:24,000. These areas are shown
at right in brown.
Wetlands: Digital National
Wetlands Inventory shapefiles
originally compiled at 1:24,000
scale by the U.S. Fish and Wildlife
Service were used as the source for
this layer. NWI classes FO
(forested) and SS (scrub/shrub) were
used. These areas are shown at right
in green.
Topographic Slope: A statewide 30
meter Digital Elevation Model
(1:24,000 scale source) from the
National Elevation Dataset (USGS)
was used to select areas where slope
is between 5% and 40%. The lower
threshold of 5% was selected
because land with a slope less than
5% in Missouri is nearly exclusively
in high value agricultural production
and thus not likely for inclusion in
the Stewardship Program. 40%
represents the general maximum
operability limit for a skidder
operating in Missouri. The grid was
then resampled to a cell size of 45
meters. These areas are shown at right in brown.
Forest Health: The only extensive,
silviculturally addressable forest
health concern currently occurring
in Missouri is oak decline
syndrome. This malady of oaks is
primarily a combination overmature red oaks (black & scarlet) on
poor sites that are “pushed over the
edge” by drought. Statewide data
concerning species composition and
age are not available. The best we
can do is assume that all forest south
of the Missouri River in the
"Ozarks" is a candidate for this
problem. Because we don't have
finer data, the private forest land in
the above state geographic area was used. The area described is shown at right in gold.
Developing Areas: Based on U.S.
Census Bureau 1990 and 2000 data.
Census block group polygons from
1990 and 2000 were unioned to deal
with differing census geography
boundaries and the number of
households per square kilometer
was calculated. The change in
number of households / sq. km.
from 1990 to 2000 was then
calculated. Block group polygons
where the change was >= 1 and <=
8 households / sq. km. and not in
urban areas were selected. Areas
that meet these criteria are
beginning to develop, but are not yet
at the point where development pressures are great and would more likely be a candidate
for the Forest Legacy program. These areas are shown at right in purple.
Wildfire Assessment: This layer is a
composite of: wildfire risk (county
based historic number of fires),
wildfire hazard (county based fuels
assessment) and wildfire sensitivity
(population - 1990 census tracts).
The medium and high categories
from the original assessment were
used to create the "1" or hit cells for
this layer. These areas are shown at
right in red.
Analysis Mask: The analysis mask
contains the areas not considered in
the analysis: urban/developed areas,
publicly owned lands, and open
water. Urban & developed areas
were compiled from the U.S. Census
TIGER files of incorporated places
in Missouri, urban lands from the
NLCD land cover classification and
census block-groups whose
household density was greater than
20 households / sq. km. or whose
growth in household density from
1990 to 2000 was greater than 15
households / sq. km. A composite
shape file of these areas, public lands
and open water polygons was created. The final layer was created by erasing the
composite shape file of unavailable areas through the Missouri state polygon, creating a
state with holes in it corresponding to the unavailable areas. The masked areas are shown
at right in white.
Stewardship Plan Ownership
Boundaries: Entire tract
boundaries for which a
Stewardship Plan has been written
since the beginning of Program
implementation in Missouri, 1992 2002. Tract boundaries were
compiled at a source scale of
1:24,000. The Stewardship Plan
tracts are shown at right in green.
Weighting
Of the twelve criteria identified as contributing to the potential Stewardship Program
benefit of a given piece of ground, some will likely be more important than others. To
account for differing levels of importance, Missouri Department of Conservation Forestry
Regional Supervisors and Central Office staff (19 people total) ranked the twelve criteria.
Staff were asked to rank each factor from 1 to 12, with 1 being the most important. A
mean response value was then calculated for each of the 12 factors. In order to give the
most important factor the largest numerical value, the mean response for each factor was
subtracted from the highest possible rank (12). These inverse response values were then
converted to relative weights by dividing each by the sum of the inverse response values
(66.11). The relative weights were the values used in the analysis. The following shows
the final weights for each factor.
Criterion
Riparian Corridors
Priority Watersheds
Forest Patches
Forest Health
Natural Heritage Sites
Drinking Water Supply
Private Forest
Public Land Adjacency
Wetlands
Slope
Development Risk
Wildfire Assessment
Weight
0.116
0.111
0.100
0.098
0.096
0.090
0.088
0.074
0.071
0.058
0.052
0.046
An interesting sidelight, not actually used in the analysis, is the tabulation of the number
of times a given factor was ranked as number one by the group of MDC Regional
Supervisors and C.O. staff during the ranking process.
Criterion
Drinking Water Supply
Private Forest
Forest Health
Forest Patches
Priority Watersheds
Riparian Corridors
Natural Heritage Sites
# 1’s
4
4
3
3
2
2
1
GIS Analysis
The GIS data representing each of the twelve factors was converted to the ESRI Grid
format with a cell size of 45 meters, an area representing approximately one-half acre on
the ground. The Grid of each factor was converted to a 0, 1 format. For example, all the
45m grid cells that fell within the riparian buffers were coded as a “1”, while all the cells
that were outside the areas of the riparian buffers were give the value “0” in that layer.
The analysis process is relatively straightforward. Each Grid is multiplied by its weight
value, so that the cells coded as “1” take on the weight value while all the “0” cells retain
a value of 0. Because all twelve grids were derived from the same source, the grid cells of
each layer line up exactly with the cells from all the other layers. The overlay analysis
procedure uses this fact to create a final result Grid whose individual cell values equal the
sum of the values in the same location (on the same half-acre) from all twelve layers.
The maximum possible cell value in the final Grid is 1. Result Grid values for Missouri
ranged from 0 to 0.948. No single cell location was a “hit” in all twelve data layers,
though many were “misses” in all twelve layers. The grid cell with the highest value had
a value of 0 in the Developing Areas layer but was a “hit” for the eleven other layers.
The ESRI Spatial Analyst extension allows for the specification of an analysis mask. The
analysis mask layer described above was used in this capacity to exclude areas of
Missouri that don’t meet eligibility criteria for inclusion in the Forest Stewardship
Program (open water, public lands, urban areas, etc.).
To make interpretation of results easier and allow for computation of area statistics, three
data classes were established to group the continuous cell values: Low, Medium and
High Stewardship Potential. There are several possible methods for establishing class
breaks. The Project group decided to use the Natural Breaks classification algorithm
available in Arc View. Class definition values for the Missouri final results data are:
Low
Medium
High
0 – 0.178
0.179 – 0.394
0.395 – 0.948
The final result grid was reclassified to an integer grid where 1 represents Low, 2
Medium, and 3 High.
Stewardship Plan Tract Digitizing
Stewardship plans are not collected centrally in Missouri. Maps of each Stewardship
plan tract drawn on a 1:24,000 scale topographic map were collected from each Forestry
office in the state. Tract boundaries were then “heads-up” digitized. Each tract was
assigned a unique identifier based upon its legal description, and a database of plan
attributes was constructed.