GIS Applications in

1
GIS Applications in Environmental Modelling
David E. Atkinson Laboratory for Paleoclimatology
and Climatology Department of Geography University
of Ottawa
2
The strength of Geographic Information Systems

• Quantitative combination of multiple data sources
  for a given area
• Analysis - insight into a specific problem
• flood analysis - will a specific area be flooded
• classic business example - where to place a store
• These take information about things like property
  location and river flood stages and combine them
  to determine if a given area will be flooded for
  a given magnitude of event
• Prediction - generate or model a result - a new
  type of data, usually an entire field
• in these cases information about processes known
  to control the parameter under investigation are
  combined in the GIS to predict occurrences of the
  unknown parameter
• My personal focus is the predicitive/modelling
  aspect of GIS various will be covered here

3
Example groups

• Temperature modeling
• My model for the Canadian Arctic Archipelago
• Chris Dalys PRISM model
• Dan Cornfords Minimum temperature model
• Watershed model - erosion, water balance
• U of New Hampshire Gulf of Maine project
• Permafrost model - occurrence, temperature, depth
• Fred Wright and Caroline Duchesne of the
  Geological Survey of Canada
• Vegetation model
• New Canada Plant Hardiness index map
• Cellular automata - not GIS per se, but could be
  implemented
• Mike Sawada and I model for dynamic ecological
  modeling

4

• Begin with the more detailed examples
• Better feel for how they work
• Then apply to the more generally described
  examples
• Begin with my work Estimation of surface air
  temperature over the Canadian Arctic Archipelago

5

• First must consider the reasons for conducting
  the modeling
• Guides the methodology, because
• Identifies what type of supporting data are
  required
• Controls complexity of approach, including issues
  of scale
• Sets acceptable error limits
• Usually reasons are either
• Provision of more accurate data field for input
  into other work, eg permafrost model
• Generation of more accurate data or investigation
  of processes
• My practical requirement improve temperature
  data accuracy
• But
• Theoretical side test hypothesis about the
  processes controlling temperature
• Situational context in the CAA governing why do
  this

6
Physical features and topography of the Canadian
Arctic Archipelago
meters ASL
7
A closer look at the Devon Ice Cap...
8
Weather stations of theMeteorological Service of
Canada in the Canadian Arctic Archipelago
January, 2000
9
Existing weather observing situation in the CAA

• MSC network in the CAA
• data paucity (low density) and bias (coastal
  locations)
• represents trends, synoptic ok, but cannot be
  used for meso/regional
• e.g. Cogley and McCann, 1976 or Ted Lewis, 1999
• This weakness has been recognized and attempts
  have been made to improve meso-scale resolution
• Maxwell (1980, 1982) - experience and sporadic
  historical stations to subjectively modify
  long-term means
• Alt and Maxwell (1990) - incorporation of
  short-term non-standard observing stations to
  provide information in key areas.
• Jacobs (1990) - strategic placement of AWS for
  several seasons, transfer function to AES
  stations, surrogate data generation for full
  record

10
So what can sporadic data do?
11
Cogley and McCann, 1976
12
Empirical modeling of climatic fields

• GIS approach to guide and improve climate
  parameter (temperature) interpolation in a
  data-sparse region
• Essential concept processes acting on
  temperature can be parameterized in aspects of
  location
• Method
• Upper air temperature observations. to guide
  estimates of temperature at elevation
• Low-level winds used to determine on-shore
  advective exposure
• Icefield cooling effect
• Verification Used MSC and PCSP surface
  observations for spot verification
• 65 of residuals within /-1.4 C

13
Basis for the temperature estimation
14

• Consider the scale
• Base spatial scale 1 kilometer grid
• What features are resolved, ie icefields,
  valleys, fiords
• What controls on temperature are possible?
• Which of these is important at this scale?
• Examples
• surface differences on the order of 10s of meters
• Effects of elevation
• Synoptic patterns
• Maritime flow off the ocean (ice)

15
Base temperature estimate

• Take observed temperatures and interpolate them
  over the domain
• This is the traditional approach

16
(No Transcript)
17
(No Transcript)
18
Recall the topography of this region
meters ASL
19
Base temperature estimate

• OR incorporate at least the fact that temperature
  changes with elevation
• This is not a new concept but now we can better
  quantify it
• How to put this in the model?
• Two ways
• Specification
• Empirical observation
• Specification input a set rate of change (a
  lapse rate), such as the Dry Adiabatic Lapse Rate
• Observation or can use observed temperature
  data, if available
• Set of instruments up a mountain (which is rare)
• Upper air observations made using weather balloon
• Observational approach is superior because it
  incorporates features that were present at that
  time and place

20
Base temperature estimate

• Temperatures at elevation estimated using
  upper-air data
• High-order polynomial fitted to ascent curves
• Upper air data are not always present at all
  levels using a polynomial allows us to smoothly
  represent the upper air profile, and thus the
  rate of change of temperature with elevation.

21
(No Transcript)
22
Base temperature estimate

• Weak surface inversion sometimes present
• All UA stations in the archipelago are within 1
  km of the ocean
• Assumption summer surface inversion caused by
  proximity to ocean and must be removed to
  accurately represent regions free of a coastal
  influence

23
and an example inversion removal
24
and an example inversion removal
25
Ascent profile for MSC Resolute Bay Mean
estimated profile and individual measurements for
July 6-19, 1985
26

• Now instead of interpolating a temperature over
  the entire grid, we can interpolate the
  polynomial coefficients over the image
• Recreate an equation in every pixel
• solve the for temperature, pixel by pixel, using
  the elevation value from that pixel

27
Schematic representation of base temperature
estimation
VB
28
Influence of wind

• Now we have a base temperature estimate for each
  pixel in the region
• But the maritime influence has been
  (deliberately) removed
• Now must re-insert a maritime influence, confined
  appropriately
• How to do this?
• Resolve winds from the upper air data, and
• Any coastal area with the wind blowing onshore is
  then subject to a wind-induced modification

29
(No Transcript)
30
Wind effect filter operation
31
Wind effect filter results

• Arrows indicate wind direction

• Local resultant winds are extracted from the
  800mb level
• A wind effect potential is calculated for
  exposed pixels

32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
PRISM Parameter elevation Regression on
Independent Slopes Model

• Chris Daly, Oregon State
• Selected for the new Climate Atlas of the US
• Matches parameters
• All pixels have parameter list generated
• Interpolation is guided such that pixels most
  similar to a station will take that stations
  value

38
PRISM Parameter elevation Regression on
Independent Slopes Model

• Parameters include
• Distance
• Stations farther away will be less similar
• Elevation
• Many climatic factors affected by elevation
• Cluster
• Geostatistical concept clumping tends to cause
  overrepresentation
• Vertical layer
• Within boundary layer or not
• Topographic facet
• Aspect facing
• Coastal proximity
• Distance from and orientation with respect to
  coast
• Effective terrain
• For precipitation

39
(No Transcript)
40
Dan Cornfords Minimum Temperature Model

• Parameters include (based on analyses)
• DEM derived altitude
• Distance to nearest coast
• Distance to nearest drainage feature
• Percentage tree cover
• Landcover derived radiative properties
• Percentage land within 25 km radius
• Difference between cell elevation and maximum
  within 5 km
• Distance to nearest urban feature
• Katabatic flow accumulation
• Mean altitude with 50 km

41
(No Transcript)
42
Gulf of Maine Watershed Model

• Parameters include
• DEM (USGS GTOPO30)
• 2-minute resolution river location grid (derived
  from DEM)
• Land cover (derived from NOAA data)
• Soil data
• Climate data
• Stream discharge data
• Atmospheric deposition
• Water quality data (Chemical inputs)
• Outputs include
• Runoff
• Evapotranspiration
• shallow groundwater
• soil moisture variations

43
Geological Survey of Canada Permafrost model

• Parameters include
• Elevation
• Temperature
• Surficial material
• Covering
• Generates
• Occurrence of permafrost
• Temperature at top of permafrost
• Thickness of permafrost

44
(No Transcript)
45
(No Transcript)
46
Thanks to Fred Wright and Caroline Duchesne of
the Geological Survey of Canada, Terrain Sciences
Division For the Mackenzie Valley Permafrost
modeling information
47
Agriculture Canada Plant Hardiness Zones

• Parameters include
• Canadian plant survival data
• minimum winter temperatures
• length of the frost-free period
• summer rainfall
• maximum temperature
• snow cover
• January rainfall
• maximum wind speed

48
(No Transcript)
49
Cellular Automata Modeling

• Grid-based approach to modelling dynamic
  environments with sub-grid level interaction
• Simple ecosystems
• 2 and 3-dimensional dispersion modelling
• Spatial extent is defined as a grid (although
  there are squares, triangles and hexagons)
• Each cell is homogenous and can take only one
  state
• Most famous CA is The Game of Life, a simple CA
  by Conway that appeared in Scientific American in
  1970

50
(No Transcript)
51
Cellular Automata Modeling

• Cell state is (partially) dependent upon
  interactions with neighbours
• Local-scale interactions
• Environmental gradients can also be modeled
• Simulate something like coastal-to-inland
  progression
• Note that gradient forcing can also be temporal
• Run by discrete time-steps
• system is evaluated, changes made, next time step
  executed

52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
(No Transcript)
57
(No Transcript)
58
Thank you!