Vegetation Mapping of the
Middle
Mark Horner
CE 547
Spring 2006
Background
The Upper Rio Grande Water Operations
Review and Environmental Impact Statement (URGWOPS) is a watershed level
analysis that seeks to operationally balance the myriad of competing water
interests in the basin through existing authorities held by the joint lead
agencies (JLAs) of the U.S. Army Corps of Engineers,
the U.S. Bureau of Reclamation, and the New Mexico Interstate Stream
Commission. These federal and state
agencies are not only responsible for the day-to-day operations of the major
dams and conveyance facilities in the basin, but also share in an range of
ancillary roles dictated by the Endangered Species Act (ESA), Rio Grande
Compact, and various flood control responsibilities. Given the complexity of such legal,
functional, and environmental interactions, the JLAs
formulated an approach whereby modeled effects of differential reservoir
operations upon the various resources in the basin could be assessed from
long-term planning perspective. In
summation, URGWOPS and the JLAs are attempting to
maximize water resources in the Rio Grande basin while minimizing the potential
adverse effects imparted by the proposed operational regime(s) upon a
frequently drought stressed river system.
For more information or to obtain a copy of the draft Environmental
Impact Statement, please visit the URGWOPS website at: http://www.spa.usace.army.mil/urgwops/. For more information on the Upper Rio Grande
Water Operations Model (RiverWareÔ) please visit: http://www.spa.usace.army.mil/urgwom/
or http://cadswes.colorado.edu/.
Riparian resources are clearly an important
part of the system and thus a compulsory consideration of the URGWOPS
project. As a result, an extensive
vegetation mapping effort was completed in late 2003 and was largely patterned
after Hink and Ohmart (1984). This
mapping sought to characterize the woody species dominance structure between
the levees from
Figure 1.
URGWOPS Vegetation Mapping Effort
Methods
Hink and Ohmart Vegetation Classification – The Hink and Ohmart (1984) classification
system was formulated to characterize woody species dominance and was largely
predicated on wildlife use. The
methodology sections a given stand into canopy and shrub layers (separated by
“/”) with a trailing numerical value to indicate a bounded overall canopy
height (1 = > 40ft. and 6 = < 5ft.).
In addition, each layer can have up to four species, implying a minimum
25% level of code inclusion that is hierarchical (i.e. the first species in the code is typically more abundant but
can be a 50% member in the case of a two species mix). For example: C-RO/SC1 = cottonwood-Russian
olive canopy with a saltcedar shrub layer.
Stand height is > 40ft. Again,
the canopy layer can have a 50-75% contribution of cottonwood and Russian olive
is bounded by a 25-50% contribution with respect to the order in the code and
the minimum 25% inclusion level.
Successive iterations, including more species, are similarly structured. Again, this methodology was conceived to
characterize dominance and not intended to be an account of all species
present. For more information reference
Hink and Ohmart (1984).
Initial URGWOPS Vegetation Mapping
– Field maps and notes
were compiled, validated, and digitized in ArcInfo at
a nominal scale 1:5000 (no greater than 1:10,000). Snapping tolerances were 0.5 meters and the
digitization process was overlaid on August 2002, color infrared
orthophotography (also used for field maps).
ArcEdit sessions typically involved the
construction of polygons through arcedit:ef arc (edit feature arc) and clean/build commands
to produce a continuous polygon coverage.
Separate arc:tables
entries were limited to unique polygon identification numbers (polygon_ID). Field attribute information was collected
wherever possible and indexed to the polygon_ID for
subsequent joining in ArcGIS
8.2 (36 attribute fields). Projection,
datum, and units were UTM NAD27, meters as August 2001 photography was orthorectified under these parameters. This was deemed sufficient as the cylindrical
projection, with a central meridian of -105.00°, should preserve area well along the narrow riparian corridor
(bosque) of the
Exotic Species Extraction and
Characterization – Hink and Ohmart attribute codes of the 2002 mapping data were
used to extract polygons of exotic (saltcedar and Russian olive) dominance
using Structured Query Language (SQL) expressions and grouped into three
infestation categories (Heavy, Moderate, and Light). This phase was performed in ArcGIS 9.0. The
bounded categories reflect the relative position of the species in the
code. For example, SC/XX and SC-XX/XX
are both classed as “Heavy” saltcedar infestation as infestation ranges from
50-100%. This was applied for either
canopy or shrub layers. Additional
descriptions on all classifications can be found in Appendix A. Each infestation category was then
characterized for percent coverage of the total mapped area (10,689 hectares)
and preliminary evaluations of mean polygon size was also performed (Satterthwaite’s two tailed t-test assuming unequal
variance). National elevation dataset –
New Mexico (NED) digital elevation models (DEM) were downloaded from the New
Mexico Resource Geographic Information System Program at http://rgis.unm.edu. These data were used for visual and location
rendering. NED data has a cell size of
one arc-second.
Results
Saltcedar distribution in the study area is
given in Figure 2. Heavy saltcedar
infestation constitutes 53% of the total mapped area (10,689 hectares) and
fully 70% of the mapped area shows some degree of infestation. Non-infested area is all other species. Mean
polygon size of Heavy (9.5ha) and Moderate (7.4ha) infested stands were
significantly larger (p = 0.0000, df = 725; p =
0.0000, df = 230 respectively at a = 0.05 level) than those containing no
saltcedar (3.9ha). Light infested
polygons were not significantly different (5.1ha; p = 0.1, df = 97 at a = 0.05 level).
Figure 2.
Percent infestation of saltcedar over total mapped area (10,689ha)
Russian olive distribution is given in
Figure 3. Heavy and Moderate infested
areas were similar in proportion (14% and 9% respectively) while Light
infestation comprised only 4% over all mapped areas. Russian olive infestation on the whole totals
27% of all mapped area. Again,
non-infested areas include all other
species. Mean polygon size of Russian
olive Heavy infestation (5.2ha) is significantly smaller (p = 0.02, df = 822 at a = 0.05 level) than non-infested polygons
(6.7ha) although not overwhelmingly.
Both Moderate (5.8ha) and Light (8.0ha) Russian olive infested polygons
were not significantly different (p =
0.2, df = 368; p
= 0.3, df = 71 at a = 0.05 level respectively).
Figure 3.
Percent infestation of Russian olive over total mapped area (10,689ha)
Generally, saltcedar infestation is both
more centralized and abundant in the southern portions of the study reach
(Figs. 4 & 5). This pattern holds
for virtually all areas south of the San Acacia diversion dam and polygon size
also tends to increase as latitude decreases. Russian olive infestation
exhibits a greater abundance and centralization north of the San Acacia
diversion dam and more so in the
Figure 4.
Example of southern study reach showing saltcedar infestation
Figure
5. Saltcedar infestation in the vicinity
of Socorro, NM
Figure 6.
Russian olive infestation of the
Figure
7. Russian olive infestation of south
Conclusions
Saltcedar and Russian olive infestation of
the middle
Russian olive infestation, while not nearly
as widespread, still comprises roughly 27% of all mapped areas at various
infestation levels. Mean polygon sizes
do not suggest that this species has the same ability to exclude others and
merge into larger areas. In fact, it
appears to nucleate into small, dense areas as polygon size grows successively
smaller from Light to Heavy infestation; however, this conclusion may be
falsely influenced by the larger nature of saltcedar polygons included in the
non-infested Russian olive. Russian
olive tends to be centered and more abundant in the northern areas where
drought tolerance is less of a factor.
Conversely, this pattern may be the result of saltcedar’s overall
abundance in the south and the inability of Russian olive to obtain a
foothold. While both hypotheses are
plausible, at this point in the analysis, it is difficult to conclude or
exclude either case, but again, patterns of infestation require a more pointed
attention and understanding.
Future Work
Russian olive and saltcedar categories are
mutually exclusive. In other words,
characterization and enumeration of these exotic species do not account for the interaction of multiple exotic species in a
given polygon. For example, a Light
saltcedar and Light Russian olive are indeed counted twice, but how does the presence of both confer a
greater level of infestation? Do two
Lights equal a moderate? I would propose
that a set of rules be derived to account for this interaction and that this be
combined in some way to the singular characterizations attempted herein as well
as the inclusion of other exotic species not addressed in this analysis.
Attribute information of these datasets
include a percent relative cover term.
So, for a given polygon, there could be a corrected area calculated that
reflects both the code and percent cover terms.
For example, a 20 hectare polygon may only be 10 hectares of vegetative
cover (percent relative cover term) and be only five hectares of saltcedar
(code term). Once validated on the
orthophotography, the corrected area would be a more accurate measure of effort
or funding for removal as well as the implications upon the ecological
understanding of infestation patterns.
Either past or future commensurate mapping
efforts need to be completed in a GIS framework. By past, I suggest the original Hink and
Ohmart (1984), which is currently hand drawn on Mylar, be orthorectified. In either case, past or future, additional
mapping could provide valuable insight into the rate of infestation and the
transitions from Light to Moderate to Heavy classes. Clearly, this has planning implications on
removal projects and their budgeting schedule.
As it stands now, most removal efforts are proceeding without broader
objectives or sound guidance. In the
end, this could have profound impacts in our efforts to reclaim infested areas
and restore the bosque…
Lastly, more in depth and formal spatial
analyses should be performed when additional processing on the 2002 data and/or
additional data sets could be obtained.
This may provide further insight in the patterns of exotic infestation
and spread.
References
Hink, V.C. and R. D. Ohmart. 1984.
Middle
Appendix A
Exotic Categorical
Classification
&
Suggested Rules for
Corrected Area Calculation
Categorical
Classifications and Suggested Rule Set
q
Heavy
Salt Cedar Infestation:
-Any code combination whereby either canopy
or shrub layer is 50-100% relative cover.
1) XX/SC = Saltcedar present in shrub 76-100%
2) SC/XX = Saltcedar
present in canopy 76-100%
3) XX/SC-XX = Saltcedar present in
shrub 50-75%
4) SC-XX/XX = Saltcedar present in
canopy 50-75%
For
future work of code and cover corrected area (see Future Work section):
Mean value multiplier of items 1 & 2 is 88% and items 3 & 4 is
62.5%.
q
Moderate
Salt Cedar Infestation:
-Any code combination whereby either canopy
or shrub layer is 25-49% relative cover; 25% is included here
because the two-species stands with saltcedar being the last constituent (see
green and blue entries below) could be from 25-49% saltcedar whereas saltcedar
in the first position is 50-75% (see green and blue entries above).
1) XX/XX-SC = Saltcedar present in
shrub 25-49%
2) XX-SC/XX = Saltcedar present in
canopy 25-49%
3) XX/SC-XX-XX = Saltcedar
present in shrub 34-50%
4) SC-XX-XX/XX = Saltcedar
present in canopy 34-50%
For
future work of code and cover corrected area (see Future Work section): Mean
value multiplier of items 1 & 2 is 37% and items 3 & 4 is 42%.
q
Light
Salt Cedar Infestation:
-Any code combination whereby either canopy
or shrub layer is 25-33% relative cover.
1) XX/XX-SC-XX = Saltcedar
present in shrub 25-33%
2) XX/XX-XX-SC = Saltcedar
present in shrub 25-33%
3) XX-SC-XX/XX = Saltcedar
present in canopy 25-33%
4) XX-XX-SC/XX = Saltcedar
present in canopy 25-33%
5) XX/SC-XX-XX-XX = Saltcedar
present in shrub 25% (in any one of the four shrub positions)
6) SC-XX-XX-XX/XX = Saltcedar
present in canopy (in any one of the four canopy positions)
For
future work of code and cover corrected area (see Future Work section): Mean
value multiplier of items 1-4 is 29% and items 5 & 6 is a fixed 25%.
Note: For saltcedar present in
opposing layer: SC-XX (canopy = 50-75%)/SC (shrub =
76-100%) then default to highest % relative code cover; in this case 76-100%
(88%).
Example Code and Cover
Corrected Area Calculation
Multipliers for percent relative cover of
Saltcedar:
62.5
& 88% = “Heavy Infestation”
37
& 42% = “Moderate Infestation”
25
& 29% = “Light Infestation”
To obtain the corrected aerial (two-dimensional)
extent of saltcedar infestation as described by the categories Heavy, Moderate,
and Light:
Multiply the mean percent relative cover code values
(from above) and the mean cover
proportionality term (in attribute information) with the area as
mapped. Again, for saltcedar present in
opposing layer such as SC-XX (canopy = 50-75% = 0.625)/SC
(shrub = 76-100% = 0.88) then default to highest percent relative cover
(76-100% = 0.88). This should capture
the extensiveness of infestation without over estimating area.
For example, consider the polygon XX/SC
that has a 25-75% shrub cover (as noted in the attribute table) and is 10.0
hectares:
0.88 (mean single species
relative cover value in shrub layer) x 0.50 (mean of
25-75% shrub cover) x 10.0 (total acreage of polygon) = 4.4
hectares of heavily infested SC of the original 10.0 hectare polygon.
For Hink and Ohmart (1984):
Multiply the mean
percent relative cover values with the area of the given polygon
ONLY – in order to be able to compare and statistically test
against the 1984 mapping. This will
overestimate the total saltcedar infestation as it ignores any open areas of
the polygon and assumes full and uniform vegetation coverage over the polygon
(i.e. percent relative cover information was not collected as part of the
original Hink and Ohmart (1984) data; this would be a significant shortcoming
of the 1984 data set. Percent relative
cover should be estimated through contemporary orthophotography for any direct
comparisons of area.
For example, consider the same polygon
XX/SC and is 10.0 acres:
0.88 (mean single species relative cover
value in shrub layer) x 10.0 (total acreage of polygon) = 8.8 acres of
heavily infested SC of the 10.0 hectare polygon.
Appendix B
Contact information for data request from the URGWOPS project
(vegetation mapping, orthophotography, etc.):
Mark W. Horner
Project Manager/Operations
Division Senior Biologist
4101
Tel: 505-681-2743
or
Lesley A. McWhirter
Project Manager
4101
Tel: 505-342-3678