VISUALIZING FORECAST UNCERTAINTY FOR MISSION PLANNING

1
VISUALIZING FORECAST UNCERTAINTY FOR MISSION
PLANNING
Abstract In determining the effects that
weather will have on a mission, the analyst
currently uses a single weather forecast combined
with a database of weather effects on assets.
Incorporating uncertainty into the determination
requires the additional elements of a
probabilistic forecast and, importantly, the risk
tolerance for the mission. In this poster, we
discuss the visualization of the integrated
ensemble weather forecasts, weather effects on
assets, and risk tolerance. Next Century
Corporation is developing an innovative and
comprehensive weather-based decision aid for
mission planning that incorporates this critical
dimension of forecast uncertainty into its
analyses and visualization scheme. The
development of the Weather Risk Analysis and
Portrayal (WRAP) decision aid is being performed
under a Phase II SBIR contract with the
Computational and Information Sciences
Directorate (CISD) of the Army Research Lab
(ARL). Next Century is presenting the current
results and status of this aid, including a
demonstration of the current version of the WRAP
prototype.
Introduction Next Century is developing an
innovative and comprehensive weather-based
decision aid for mission planning that
incorporates the critical dimension of forecast
uncertainty into its analyses and visualization
scheme. Development of this Weather Risk
Analysis and Portrayal (WRAP) decision aid is
being performed in support of the Computational
and Information Sciences Directorate (CISD) of
the Army Research Lab (ARL) as part of a Phase I
and Phase II SBIR contract. Next Century
presents the current results and status of this
initiative including a demonstration of the
current version of the WRAP prototype. WRAP
employs ensemble forecasts, user-defined mission
parameters and a weather effects rules system
(currently, the Armys Integrated Weather Effects
Decision AidIWEDA) to produce a visual portrayal
of the missions weather-related risks, including
determining the best time frame for the mission.
The open architecture for WRAP allows for
multiple deployment possibilities, including
integration into the DoDs Integrated
Meteorological System (IMETS) and IWEDA
framework. Uncertainty in weather A very large
number of forces and processes combine to produce
a weather event or to shape the course of the
weather. Because of the uncertainty in observing
or specifying each of these causes of atmospheric
evolution, a forecast will always have some
uncertainty. A key strategy in using uncertainty
is to calculate a number of forecasts (an
ensemble) using a distribution of initial and
boundary variables and the variations of the
model physics. By calculating enough forecasts
to represent the conditions in these probability
distributions, it is possible to estimate the
probability distribution of the forecast
variables. There are two benefits to this
approach. First, the average ensemble forecasts
are observed, over the long term, to be the best
single forecast. Second, the ensemble gives
estimates of the probability distribution of the
outcomes. Using probability forecasts offers
several benefits over single forecasts. First,
because the uncertainty in the forecast is
specifically expressed, the user is made aware of
that uncertainty and can use that information in
decision making. Ensemble forecasts can be used
with thresholds to make decisions, where the
thresholds can vary from user to user, based on
the uncertainties involved and their own
threshold for making the decision. Asset
Rules The final use of forecasts for mission
planning is an understanding of the weathers
effects on the assets used in the mission. The
Armys Integrated Weather Effects Decision Aid
(IWEDA) program uses a sophisticated rules
processing system to determine the effects of
weather on combat operations. Each military
system and piece of equipment can have multiple
IWEDA rules associated with it. Each IWEDA rule
can specify a weather effect threshold that makes
the use of the system either marginal or
unfavorable. These rules may be either simple
(i.e. dependent on a single weather variable) or
compound (i.e. based on several variables).
These rules are currently used with a single
deterministic forecast. The purpose of the WRAP
project is to incorporate an ensemble forecast
with these sorts of rules. Risk Tolerance Each
mission has a unique level of acceptable risk.
In training missions, the risk tolerance can be
low, because additional time and opportunities
may be available for the mission. The mission
planner then needs to know not only when asset
rules are exceeded, but where and when rules
might be exceeded with some low probability. The
WRAP tool is designed to give the analyst this
additional information and the weather is close
to the boundary. For critical missions, the risk
tolerance can be higher. In those cases, the
analyst needs to understand where and when the
weather is least bad, and where the limited
opportunities exist. It is not useful to the
analyst to simply show a solid red screen
instead, they need to see where the system is
close to the boundaries.
Dithering This visualization technique employs
the use of dithered patterns graded dithering
of green/amber and amber/red dots. When the
probability of exceeding a threshold is greater
than the risk tolerance, the higher threshold
color is dithered on the lower threshold color.
For example, when the expected value of the
distribution is favorable (lt9 kts), but the
probability that it will exceed 9 kts is greater
than the risk tolerance (say, 10), then amber is
dithered on top of the green, as shown below.
The greater that the probability of exceeding the
threshold, the greater the amount of dithering.
The dithering is grouped into 9 steps to improve
visual understanding of the results the 9 steps
are shown below.

• WRAP Status
• WRAP is currently in development. The following
  have been accomplished
• implemented the overall architecture and
  application framework.
• created web service for downloading ensemble
  data, only getting needed data
• implemented both sliding boundary and dithering
  calculations
• integrated with IWEDA rules database for asset
  rules
• incorporated GIS visualization tools and
  geographic shape files
• provided overlays showing grid results
• added Next Centurys innovative temporal
  slider, showing aggregate risk information and
  temporal context
• implemented drill-down capabilities for the
  analyst to determine why they got the results
  they did

The WRAP system provides the analyst with the
ability to see how close the weather is to a
boundary, where close is based on the
probability distribution and the users risk
tolerance.
Combining Risk Tolerance and Uncertainty with
Asset Rules The ensemble forecast provides a
statistical distribution of outcomes as a
function of time and location. For each grid
square and time, WRAP colors the resulting grid
based on the acceptability of using the
equipment. The color depends on whether the
probability of exceeding asset rule thresholds is
greater than the risk tolerance. The analyst
provides the level of risk tolerance.

• Data and Data Flow
• WRAP currently uses forecast information from
  ZedX, a partner in the Phase II SBIR. The
  underlying data comes from the Global Forecast
  System operated by the National Centers for
  Environmental Prediction (NCEP). ZedX uses 11
  runs in the ensemble, processes it, and
  calculates the probability distributions for
  values to compare to equipment asset rules. The
  processing reduces error through bias corrections
  and the expectation maximum method.
• The forecast has a 48-hour forecast period, with
  6-hour time steps, with three domains
• Global domain with 1 degree ( 110 km) spatial
  resolution
• Iraq domain with 10 km spatial resolution
• Washington State domain with 1 km spatial
  resolution
• In the future, WRAP will be using ensemble data
  from the Joint Ensemble Forecast System (JEFS),
  produced by the Air Force Weather Agency. WRAP
  has been architected so that it is able to pull
  information from diverse sources. An adapter can
  be placed within the WRAP system to pull ensemble
  information from either ZedX or JEFS. Additional
  adapters can be used to use asset information
  from alternative sources.

• Sliding Boundary
• In this visualization technique, the
  probabilistic forecast is combined with the
  thresholds and the analysts risk tolerance to
  determine the coloring of the grid with solid
  green, amber, or red
• In the first example below, the probability
  that the wind speed will exceed the thresholds is
  very low (1). The grid is colored green
  regardless of the risk tolerance.
• In the second example below, the probability
  depends on the risk tolerance. If low or medium
  risk is tolerated, the probability of exceeding
  the unfavorable threshold is greater than the
  tolerance however, if the tolerance is high,
  then the probability of exceeding the threshold
  (58) is less than the tolerance (70). Note
  that the probability of exceeding the marginal
  threshold (405898) is greater than the
  tolerance and so grid is colored amber.
• In the third example below, the probabilities,
  risk tolerances, and rules combine to provide
  three different results depending on the users
  risk tolerance.

WRAP Team