Beyond Mapping III
|
Map
Analysis book with companion CD-ROM
for hands-on exercises and further reading |
Harvesting an Understanding of GIS Modeling — describes
a prototype model for assessing off-road access to forest areas
Extending Forest Harvesting’s Reach — discusses
a multiplicative weighting method for model extension
A Twelve-step
Program for Recovery from Flaky
Forest Formulations — describes
a spatial model for identifying Landings and Timbersheds
E911 for the
Backcountry — describes
development of an on- and off-road travel-time surface for emergency response
Extending Emergency
Response Beyond the Lines — discusses
basic model processing and modifications for additional considerations
Comparing
Emergency Response Alternatives — describes
comparison
procedures and route evaluation techniques
Note: The processing and figures
discussed in this topic were derived using MapCalcTM software.
See www.innovativegis.com to download
a free MapCalc Learner version with tutorial materials for classroom and
self-learning map analysis concepts and procedures.
<Click here>
right-click to download a printer-friendly version of this topic (.pdf).
(Back to the Table of Contents)
______________________________
Harvesting
an Understanding of GIS Modeling
(GeoWorld, April 2010)
Vast regions of the Rocky
Mountains are under attack by mountain pine beetles and a blanket of brown is
covering many of the hillsides. Dead and
dying trees stretch to the horizon. In
five years there will be just sticks poking up and within twenty years the
forest floor will look like a game of “pick-up sticks” with a new forest poking
through.
It’s an ecological cycle,
but it is both aggravated by and aggravating to many of us who live and play in
the shadows of the mountains. Is there
something we can do to contain the spread and hasten the regenerative
cycle? One suggestion is to remove the
dead wood to speed forest health and convert it to useful products to
boot.
This appears attractive
but just knowing there are giga-tons of beetle-gnawed biomass awaiting “wood utilization”
solutions isn’t a fully actionable answer.
What products are viable? Where
and how much harvesting is appropriate?
These two basic questions
captured the attention of combined graduate project teams at the University of
Denver. A “capstone MBA” team focused on
the business case while a “GIS modeling” team focused on the geographic
considerations. Their joint experience
in identifying, describing and evaluating potential solutions provided an
opportunity to get their heads around a complex issue requiring integration of
spatial and non-spatial analysis, both at a macro state-wide level and a micro
local level. The experience also
provides a springboard for a short Beyond Mapping series on GIS modeling (scar
tissue and all).
Our outside collaborators
(a non-profit organization and a large energy company) narrowed the
investigation to biomass for augmentation of base-load electric energy
generation—first lesson, always heed the client’s interests. This assumption narrows the macro considerations
as haul distances from a plant are critical.
Considering mountainous travel, buffering to a simple geographic
distance is insufficient and travel-time zones were recommended—second lesson, clients
love the on-road travel-time concept.
Figure 1.
Relative harvesting access is determined by availability of forest lands
as modified by intervening conditions.
The concept of modeling
off-road access, on the other hand, is a bit harder to appreciate. It was decided that a micro level
“proof-of-concept prototype model” for assessing forest access would be
developed. Figure 1 depicts the map
variables and basic approach taken for a hypothetical demonstration area—third
lesson, never use real data for a prototype model if you want clients to
concentrate on model logic.
The first phase of the
basic model determines Availability of lands for harvesting
activity. Legal concerns, such as
ownership, stream buffers and sensitive areas must be identified and
unavailable lands removed from further consideration. In addition, physical conditions can become
“absolute barriers,” such as steep slopes beyond the operating range of
equipment. A second phase characterizes
the relative Access of available lands by considering intervening
conditions as “relative barriers,” such as increasing slope in operable areas
increases costs of harvesting.
It is important not to
“over-drive” the purpose of a Prototype Model as a mechanism for demonstrating
a viable approach and stimulating discussion—fourth lesson, “keep it simple
stupid (KISS)” to lock a client’s focus on model approach and logic. Anticipated refinements should be reserved
for a “Further Considerations” section in the presentation describing the
prototype model.
Figure 2. Flowchart of the basic model involves four
base maps and ten processing commands.
If model refinement
accompanies prototype development, there isn’t a need for a prototype.
But that is the bane of a “waterfall approach” to GIS modeling. You can easily drown by jumping off the edge at
the onset; whereas calmly walking into the pool with your client engages and
involves them, as well as bounds a
manageable first cut of the approach and logic … baby steps with a
client, not a top-down GIS’er solution out of the box. Fifth lesson—there
is a sweet spot along a client’s perception of a model from a Black box of
confusion to Pandora’s box of terror.
Figure 2 contains a
flowchart of model logic for the basic Availability/Access prototype
model. Only four base maps and ten
commands are involved in a demonstrative first cut. A Slope map is used to derive slope impedance
where ranges of steepness are assigned 1 (most preferred)= 0-10%, 2= 10-20%, 4=
20-30% 7 (seven times less preferred)= 30-40% and 0 (unavailable)=
>40%. The other maps of Ownership,
Water and Sensitive Areas are used to derive binary maps where 1= available and
0= unavailable lands. The final step
calculates the acreage of accessible forests within each watershed.
The four calibrated maps
are multiplied for a Discrete Cost Surface that contains a zero for unavailable
lands (any 0 in the map stack sends that location to 0) and the relative
“friction values” based on terrain steepness are preserved for available areas
(1 * 1* 1 * friction value retains that value).
In turn, this map is used to generate the relative access map using a
“Least Cost” approach that will be discussed in next month’s column that “lifts
the hood” on technical considerations (see Author’s note).
Figure 3. Different effective “reaches” into the
accessible forested areas can be generated to simulate varying budget
sensitivities.
Figure 3 provides an
early peek at some of the output generated by the basic Forest Access
model. The left inset shows the relative
access values for all of the available forested areas with warmer tones
indicating a long harvesting reach into the woods; light grey, unavailable and
dark grey, non-forested. A user can
conjure up different “reach” scenarios defining accessible forests as a means
to understand the spatial relationships from grabbing just the “low hanging
economic fruit (…err, I mean wood)” that is easily accessed (right inset), to
increasingly aggressive plunges deeper into the woods at increasingly higher
access costs.
Also, consideration of
human concerns, such as housing density and visual exposure, might affect a
practical assessment of the access reach.
Finally, locating suitable staging areas (termed “Landings”) for wood
collection and the delineation of the forest areas they serve (termed
“Timbersheds”) provide even more fodder for next couple of columns.
_____________________________
Author’s Note: For a discussion on “Calculating Effective Distance and Connectivity” see the online
book, Beyond Mapping III, Topic
25, posted at www.innovativegis.com/basis/MapAnalysis/.
Extending Forest Harvesting’s
Reach
(GeoWorld, May 2010)
The previous section
described a basic spatial model for determining relative harvesting availability
and accessibility of beetle-killed forests for harvesting. The prototype model was developed by
“capstone MBA” and “GIS modeling” graduate teams at the University of Denver. A non-profit organization and a large energy
company served as outside collaborators and narrowed the focus to the
extraction of biomass for base-load electrical energy generation.
State-wide analysis
involving on-road travel was proposed for assessing hauling distances of wood
chips to power plants where the resource would be further refined and mixed
with coal. Adjusting for mountainous
travel along the road network, some beetle-kill areas simply are too far from a
plant for consideration.
Local level analysis
involving off-road harvesting is considerably more complex. In summary, this processing determines the
relative accessibility from the landings into the forest considering a variety
of terrain, ownership and environmental considerations. Adjusting for off-road access, some
beetle-kill areas are unavailable or effectively too far from roads for
harvesting.
The Basic Access Model
outlined in the top portion of figure 1 demonstrates the types of factors that
can be considered in assessing off-road access.
The processing first identifies absolute barriers to harvesting
based on ownership, environmentally sensitive areas, water buffers and terrain
that is too steep for equipment to operate.
These factors are represented as binary map layers with 1= available and
0= unavailable for harvesting activity.
Relative barriers to forest access are rated from 1= most preferred
to 9= least preferred. In the prototype
model, slopes within the harvesting equipment operating range are used to
demonstrate relative barriers with increasingly steeper slopes becoming less
and less desirable. Multiplying the
stack of map layers identifying absolute and relative barriers results in an
overall preference surface for harvesting with values from 0 (no-go), to 1
(best) through 9 (worst). The final step
uses grid-based effective distance techniques to determine the relative
accessibility of available forested areas from roads (see author’s note).
As an extension to the
basic model, human concerns for minimizing visual exposure and housing density
are outlined in the lower portion of figure 1.
The procedure first derives a visual exposure density surface
identifying the number of times each location is seen from houses and roads and
then calibrates the exposure from .5 (low exposure) through 1.0 (high
exposure). Similarly, a housing density
surface identifying the number of houses within a half mile radius was
calibrated from .5 (low density) to 1.0 (high density). The two adjusted maps are averaged for an
overall weighting factor for each map location.
Figure 1. The
Extended Access Model develops a multiplicative weighting factor based on
housing density and visual exposure of potential harvesting areas.
When the multiplicative
weight is applied to the preference map stack, it improves (lowers) preference
ratings in areas with low visual exposure and housing density, while retaining
the basic ratings in areas of high visual exposure and housing density. The effect on the model is to favor reaching
farther into available forested areas in locations that are less
contentious.
Figure 2 compares the
results with the left side of the figure tracking the results of Basic Model
and the right side tracking the results of the Extended Model that favors
harvesting in areas of low human impact.
The effective distance to the farthest available forest location is
reduced by a third from 116 to 76. The
3D plots on the bottom of the figure (insets c and d) depict the results as
bowl-shaped accumulation surfaces with the lowest value of 0 “cells away” from
the road in the lower center portion of the project area. Note the considerable easing (lower values;
flattening of the surface) of the relative proximity at the circled remote
location.
Figure 2. Comparison of Basic and Extended model
results.
Figure 3 illustrates a
couple of techniques for summarizing related map information using a binary map
of accessible forest areas. A
region-wide (zonal) overlay operation can be used to “count” the total number
of acres of accessible forest in each of the three watersheds (e.g., 374 aces
of accessible forest in Watershed 3).
Also, by simply multiplying the binary map times the vegetation map
identifies the vegetation type and area for all of the accessible forest
locations (e.g., 964 acres of accessible Lodgepole pine).
Figure 3. D. Summarizing accessible forest areas by
watersheds and vegetation type.
The ability to repackage
all beetle-kill areas into those meeting harvesting availability and access
requirements is critical. Just knowing
that there are giga-tons of biomass out there isn’t sufficient until they are
mapped within a comprehensive decision-making context. The next section explores procedures for
determining the best set of staging areas, termed “landings,” and the
characterization of the potential wood chip supply within each of their
corresponding “timbersheds.”
_____________________________
Author’s Note:
For a discussion on “Calculating
Effective Distance and Connectivity” see the online book, Beyond
Mapping III, Topic 25, posted at www.innovativegis.com/basis/MapAnalysis/.
A Twelve-step Program for Recovery from Flaky Forest
Formulations
(GeoWorld, June 2010)
The last two columns
described a basic spatial model for determining forest availability and access
considering physical and legal factors that, in turn, was extended to include
human concerns of housing density and visual exposure to harvesting
activity. This column builds on those
procedures for a further formulated model that 1) identifies the best set of
staging areas for wood collection, termed “Landings” and 2) delineates the
harvest areas optimally connected to each landing, termed “Timbersheds.”
The model involves
logical sequencing of twelve standard map analysis steps that are described
using MapCalc commands that are easily translated into other grid-based software
systems (see author’s note). The top
portion of figure 1 uses the five “binary maps” created in the basic model to
generate a map of potential landing areas.
The maps are calibrated as 1 = available and 0 = not available for
harvesting, and when multiplied together (1. Compute) results in 1 being
assigned to all roads locations passing through available forest areas—
1*1*1*1*1= 1; if a zero appears in any map layers it results in a 0 value (not
a road in an available forest area).
Figure 1. Identifying candidate
Landing Sites that are along forested roads
in gently sloped areas (steps 1-3).
The lower portion of
figure 1 depicts using a neighborhood/focal summary operation (2. Scan)
to calculate the average slope within a 100-foot reach of the each forested
road cell. The third step (3.
Renumber) eliminates potential landing areas that that are in areas with
fairly steep surrounding terrain (> 15% average slope). The result is removal of over two thirds of
the total number of road locations.
Figure 2 shows processing
steps 4 through 9 used to locate the best landing sites. In step 4, the Discrete Cost map indicating
the relative ease of equipment operation created in the basic model is masked (4.
Compute) to constrain harvesting activity to just the forested areas. The Accumulated Proximity from roads is
calculated (5. Spread) resulting in an effective distance value for each
forest location that respects the intervening terrain conditions from forested
roads.
The optimal path from
each forest location to its nearest road location is determined and the set of
paths are counted for each map location (6. Drain) resulting in an
Optimal Path Density surface. The insets
in the upper-right portion of figure 2 shows 2D and 3D displays of this
less-than-intuitive surface. Note the
yellow and red tones where many forest locations are optimally accessed—with
one road location in the southern portion of the project area servicing 785
forested locations. The long red path
leading to this location is analogous to a primary road where more and more
collector streets join the overall best route.
Figure 2. Locating the best Landing Sites based on optimal path density
(steps 4-9).
The lower portion of
figure 2 shows the steps for isolating the best landing sites. The highest levels of optimal path density
are isolated (7. Renumber) and then masked to identify the forested road
locations with the highest optimal path density (8. Compute). In turn these locations are assigned a unique
ID value (9. Clump) and summary statistics on each of the “best”
potential landing sites are generated.
The summary statistics,
along with expert judgment is used to identify an appropriate final set of
landing sites that is suitably dispersed throughout the project area (10.
Renumber) as depicted in the upper portion of figure 3. These final locations for Landings are
used to derive new effective distance values for each forest location
considering intervening terrain conditions (11. Spread) in a manner
similar to step 5. Finally, expert
judgment is used to limit the reach in each of the Timbersheds to a manageable
distance (12. Renumber).
Figure 3. Identifying and characterizing the Timbersheds of the best
Landing Sites (steps 10-12).
To put the spatial
analysis into a decision context, a “thumbnail” estimate of the wood chip
resource for Timbershed #15 is 164ac * 40T/ac = 6560 tons. At $15 to $30 per ton this converts to 6560T
* $22.50 = $147,600. From another
perspective, assuming 6000 to 8000 btu per pound of woodchips the energy stored
in the biomass translates to 6560T * 2000lb/T * 7000btu/lb = 91,840,000,000
btu. At 3412 btu per kilowatt hour this
converts to 91,840,000,000btu / 3412btu/kWh = nearly 27 million kilowatt hours
…whew!
Any way you look at it
there is a lot of energy locked up in the giga-tons of beetle-gnawed biomass
blanketing the Rockies. GIS modeling of its
availability and access is but one of several critical steps needed in
determining the economic, environmental and social viability of a “wood
utilization” solution.
_____________________________
Author’s Note: See http://www.innovativegis.com/basis/MapCalc/MCcross_ref.htm
for cross-reference of MapCalc commands to other software systems. An animated
PowerPoint slide set of this 3-part Beyond Mapping series on “Assessing and
Characterizing Relative Forest Access” and materials for a “hands-on” exercise are posted at www.innovativegis.com/basis/MapAnalysis/Topic29/ForestAccess.htm.
E911 for the Backcountry
(GeoWorld, July 2010)
One of the most important
applications of geotechnology has been Enhanced 911 (E911) location technology that
enables emergency services to receive the geographic position of a mobile
phone. The geographic position is
automatically geo-coded to a street address and routing software is used to
identify an optimal path for emergency response. But what happens if the call that “I’ve
fallen and can’t get up” comes from a backcountry location miles from a
road? The closest road location “as the
crow flies” is rarely the quickest route in mountainous terrain.
A continuous space
solution is a bit more complex than traditional network analysis as the
relative and absolute barriers for emergency response are scattered about the
landscape. In addition, the intervening
conditions affect modes of travel differently.
For example, an emergency response vehicle can move rapidly along the
backcountry roads, and then all terrain vehicles (ATV) can be employed off the
roads. But ATVs cannot operate under
extremely steep and rugged conditions where hiking becomes necessary.
Figure 1. On-road emergency response travel-time.
The left side of figure 1
illustrates the on-road portion of a travel-time (TT) surface from headquarters
along secondary backcountry roads. The
grid-based solution uses friction values for each grid cell in a manner
analogous to road segment vectors in network analysis. The difference being that each grid cell is
calibrated for the time it takes to cross it (0.10 minute in this simplified
example).
The result is an estimate
of the travel-time to reach any road location.
Note that the on-road surface forms a rollercoaster shape with the
lowest point at the headquarters (TT = 0 minutes away) and progressively increases
to the farthest away location (TT = 26.5 minutes). If there are two or more headquarters, there
would be multiple “bottoms” and the surface would form ridges at the
equidistance locations in terms of travel-time—each road location assigned a
value indicating time to reach it from the closest headquarters.
The lower-right portion
of figure 1 shows the calibrations for on-road travel by truck and off-road
travel by ATV and hiking as a function of terrain steepness and recognition of
rivers as absolute barriers to surface travel.
The programming trick at this point is to use the accumulated on-road
travel-time for each road location as the starting TT for continued movement
off-road. For example, the off-road
locations around the farthest away road location starts “counting” at 26.5,
thereby carrying forward the on-road travel time to get to off-road
locations. As the algorithm proceeds it
notes the on- and off-road travel-time to each ATV accessible location and
retains the minimum time (shortest TT).
Figure 2. On-road plus off-road travel-time using ATV under operable
terrain conditions.
Figure 2 identifies the shortest
combined on- and off-road travel-times.
Note that the emergency response solution forms a bowl-like surface with
the headquarters as the lowest point and the road proximities forming “valleys”
of quick access. The sides of the
valleys indicate ATV off-road travel with steeper rises for areas of steeper
terrain slopes (slower movement; higher TT accumulation). The farthest away location accessible by
truck and then ATV is 52.1 minutes.
The grey areas in the
figure indicate locations that are too steep for ATV travel, particularly
apparent in the steep canyon area (lower left insert with warmer tones of Slope
draped over the Elevation surface). The
sharp “escarpment-like” feature in the center of the response surface is caused
by the absolute barrier effect of the river—shorter/easier easier access from
roads west of the river.
Figure 3 completes the
emergency response surface by accounting for hiking time from where the wave
front of the accumulated travel-time by truck and ATV stopped. Note the very steep rise in the surface (blue
tones) resulting from the slow movement in the rugged and steep slopes of the
canyon area. The farthest away location
accessible by truck, then ATV and hiking is estimated at 96.0 minutes.
Figure 3. On-road plus off-road travel-time by ATV and then hiking under
extreme terrain conditions.
The lower-left insert
shows the emergency response values draped over the Elevation surface. Note that the least accessible areas occur on
the southern side of the steep canyon.
The optimal (quickest) path from headquarters to the farthest location
is indicated—that is within the assumptions and calibration of the model.
Next month we will
investigate some alternative scenarios, such as constructing a suspension
bridge at the head of the canyon and identifying helicopter landing areas that
could be used. The bottom line is that
GIS modeling can extend emergency response planning “beyond the lines” of a
fixed road network—an important spatial reasoning point for GIS’ers and
non-GIS’ing resource managers alike.
_____________________________
Author’s Note: See
www.innovativegis.com/basis/MapAnalysis/Topic29/EmergencyResponse.htm
for an animated slide set illustrating the incremental propagation of the travel-time
wave front considering on- and off-road travel and materials for a “hands-on”
exercise in deriving continuous space emergency response surfaces.
Extending Emergency Response Beyond the Lines
(GeoWorld, August 2010)
The previous section
described a basic GIS model for backcountry emergency response considering both
on- and off-road travel. The process
used grid-based map analysis techniques that consider the spatial arrangement
of absolute barriers (not passable) and relative barriers (passable with
varying ease) that impede emergency response throughout continuous geographic
space.
While the processing
approach is conceptually similar to Network Analysis, movement is not
constrained to a linear network of roads represented as a series of irregular
line segments but can consider travel throughout geographic space represented
as a set of uniform grid cells. The
model assumes that the response team first travels by truck along existing roads,
then off-loads their all-terrain vehicles (ATV) for travel away from the roads
until open water or steep slopes are encountered. From there the team must proceed on
foot. The result of the model is a
travel-time map surface with an estimated minimum response time assigned to
each map location in a project area.
Last section’s discussion
described the key conceptual considerations and results of the three stages of
backcountry emergency response model—truck, ATV and hiking movement. The most notable points were that movement
proceeds as ever increasing waves emanating from a staring location that are
guided by absolute/relative barriers and
results in a continuous travel-time map (bowl-like 3D surface).
Figure 1. Flowchart of map analysis processing to establish emergency
response time to any location within a project area.
Figure 1 outlines the
processing as a flowchart. Boxes
represent map layers and lines represent analysis tools (MapCalc commands are
indicated). The flowchart is organized
with columns characterizing “analysis levels” proceeding from Base maps
(existing data), to Derived maps, to Interpreted maps, to Modeled map
solutions. The progression reflects a
gradient of abstraction from “fact-based” (physical) characterization of the
landscape involving Base and Derived maps, through increasingly more
“judgment-based” (conceptual) characterizations involving Interpreted and
Modeled maps expressing spatial relationships within the context of a problem.
The row groupings
represent “criteria considerations” used in solving a spatial problem. In this case, the processing first considers
truck travel along the roads then extends the movement off-road by ATV travel
and finally hiking into the areas that are inaccessible by ATV. The off-road movement is guided by open water
(absolute barrier for both ATV and hiking) and terrain steepness (relative
barrier for both ATV and hiking and absolute barrier for ATV in very steep
slopes).
Figure 2. Extended response models for new trails (left) and helipad (right).
Figure 2 identifies
modifications to the model considering construction of new ATV and hiking
trails and a helipad. The left side of
the figure updates the ATV and hiking “friction” maps with lower travel-time
values for the trails over the unimproved off-road travel impedances. The hiking trail includes a foot bridge at
the head of the canyon that crosses the river.
The revised friction values (ATV trail = 0.15 minutes; hiking trail =
0.5 minutes) directly replace the old values using a single command and the
model is re-executed.
In the case of the new
helipad (right side of the figure) the hiking submodel is used but with a new
starting location that assumes an 18 minute scramble/flight time to reach the
location.
The bottom portion of
figure 3 shows the three emergency response surfaces. Visual inspection shows considerable
differences in the estimated response time for the area east of the river.
Current access requires
truck travel across the bridge over the river in the extreme SW portion of the
project area. Construction of the new
trails provides quick ATV access to the foot bridge then easy hiking on the
improved trail along the eastern edge of the river for faster response times on
the east side of the canyon (light blue).
Construction of the new helipad greatly improves response time for the
upper portions of the east side of the canyon.
Figure 3. Emergency response surfaces for the current situation,
additional trails and helipad.
The next section’s
discussion will focus on quantifying the changes in response time and
developing routing solutions that indicate the type of travel (truck, ATV,
hiking, helicopter) for segments along the optimal path to any location.
_____________________________
Comparing Emergency Response Alternatives
(GeoWorld, September 2010)
The last two sections described
a simplified backcountry emergency response model considering both on- and
off-road travel and then extended the discussion by simulating two alternative
planning scenarios—the introduction of a new ATV/Hiking trail and a
Helipad. The conceptual framework,
procedures and considerations in developing the alternative scenarios were the
focus. This section’s focus is on
comparison procedures and route evaluation techniques.
The left side of figure 1
depicts the minimum expected travel-time from headquarters to all locations
within a project area under current conditions.
The river in the center (black) acts as an absolute barrier that forces all
travel to the southeastern portion across a bridge in the extreme southwest. This makes the farthest away location more
than an hour and a half from the headquarters, although it is less than half a
mile away “as the crow flies.”
Figure 1. Subtracting two travel-time surfaces determines the relative
advantage at every location in a project area.
The inset in the center
of the figure locates a proposed new ATV/Hiking trail. The first segment of from the road to the
river enables ATV travel. A light
suspension bridge crosses the river to provide hiking access to an improved trail
along the southern side of the canyon.
While the trail is justified
primarily for increasing recreation potential within the canyon, it has
considerable impact on emergency response in the canyon. Note the introduction of the green and light
blue tones along the river that indicate response times of about half an hour
as compared to more than an hour and a half (purple) currently required.
The right side of figure
1 shows the difference in travel-time under current conditions and the proposed
new trail. This is accomplished by simply
subtracting the two maps—where 0 = unchanged response times (light grey), values
= difference in the response times (red through blue tones). The red area between the road and the
suspension bridge notes that ATV access is slightly improved (less than 2
minutes difference) with the introduction of the new trail. The greens and blues show considerable improvement
in response time with a maximum difference of 68.0 minutes.
Draping the result over
the elevation surface shows that the south side of the canyon bottom is best
serviced via the new trail. The more important,
non-intuitive information is the dividing line of best access approach (red
line) halfway up the southern side of the canyon. Locations nearer the top of the canyon are
best accessed via the current truck/ATV/Hiking utilizing the southern bridge.
Figure 2. The optimal path is identified as the steepest downhill route
over a travel-time surface.
(see Author’s Note)
Figure 2 extends the
analysis to characterize the optimal path for the most remote location under
current conditions. The first segment
(red) routes the truck along the road for approximately 19 minutes to an old
logging landing. The ATV’s are unloaded
and precede off-road (cyan) toward the northeast for an additional 15 minutes
(19 + 15= 34 minutes total). Note the
route’s “bend” to the east to avoid the sharply increased travel-time in the
rugged terrain along the west canyon rim as depicted in the travel-time
surface.
Once the southern side of
the canyon becomes too steep for the ATVs, the rescue team hikes the final
segment of 62 minutes (violet) for an estimated total elapsed time of 96
minutes (19 + 15 + 62 = 96). A digitized
routing file can be uploaded to a handheld GPS unit to assist off-road navigation
and real-time coordinates can be sent back to headquarters for monitoring the
team’s progress—much like commonplace network navigation/tracking systems in
cars and trucks, except on- and off-road movement is considered.
The backbone of the backcountry
emergency response model is the derivation of the travel-time surface (right
side of figure 2). It is “calculated
once and used many” as any location can be entered and the steepest downhill
path over the surface identifies the best response route from
headquarters—including Truck, ATV and Hiking segments with their estimated
lapsed times and progressive coordinates.
Figure 3. Comparison of emergency response routes to a remote location
under alternative scenarios.
In addition, alternate
scenarios can be modeled for different conditions, such as seasons, or proposed
projects. For example, figure 3 shows
three response routes to the same remote location—considering a) current
conditions, b) new trail and c) new helipad.
In this case, the response is much quicker for the new trail route
versus either the current or helipad alternatives.
It is important to note that
the validity of any spatial model is dependent on the quality of the underlying
data layers and the robustness of the model—garbage in (as well as garbled
throughput) is garbage out. In this
case, the model only considers one absolute barrier to movement (water) and one
relative barrier (slope) making it far too simplistic for operational use. While it is useful for introducing the
concept, but considerable interaction between domain experts and GIS
specialists is needed to advance the idea into a full-fledged application …any
takers out there?
_____________________________
Author’s
Note: See www.innovativegis.com/basis/MapAnalysis/Topic14/Topic14.htm#Hiking_time
for a more detailed discussion on deriving off-road travel-time surfaces and
establishing optimal paths.