The Atmospheric Data Modeling Workshop
in Seattle
Meeting Minutes
Workshop participants
represented NCAR, ESRI, Unidata, Meteorlogix, Raytheon,
Pacific Disaster Center, US Air Force, National Weather
Service, and several other NOAA labs (PMEL, NGDC, ATDD,
USDOC). During the course of the meeting, the representatives
from the atmospheric community worked closely with ESRI
development staff to identify GIS approaches for handling
atmospheric data more effectively and began the process of
designing a community data model. The first data modeling
meeting in Seattle centered upon the conceptual design of the
atmospheric data model for GIS. Click here to view workshop
agenda.
At the beginning Steve Grisé (ESRI) gave an overview of the
data modeling process with emphasis on a conceptual design and
presented some thoughts and ideas on bridging the gap between
atmospheric and GIS data.
Click here
to view Steve's power point slides.
The following points were brought up during a discussion:
- Cannot build all encompassing data model from the
beginning. Need to build an essential data model
first and provide tools for it to be extended for different
groups/purposes/directions.
- Timeframe: Usually it takes 2-3 months after a data
modeling meeting to develop a draft data model. A draft gets
distributed for feedback and can be tested on real projects
and use cases. After 6-12 months feedback is received and
suggestions are incorporated into the data model. It took 3
years to build ArcHydro (hydrological data model) and about
two years to build a Marine (oceanographic) data model.
- Use cases: 1) GIS users getting atmospheric datasets; 2)
Equation solving models using GIS data.
- Atmospheric science - think about parameters /
quantities, continuous sets of functions. How to bridge gap
between discrete objects and functions?
Other points of discussion:
- Analysis and data integration
- Interface with land surface properties
- Handling of interface between solid system dynamics
(earth) and fluid system dynamics (atmosphere)
- File systems as databases
- Do we need to have a data type to represent NetCDF in a
database or provide services to connect to data stored in
NetCDF?
- Data types and/or web services
- N-dimensional views; indices
Following Steve's presentation and group discussion, Terri
Betancourt (NCAR) lead a discussion about the scope of the
atmospheric data model and summarized
responses to the Conceptual
Data Modeling questionnaire
Click here
to view Terri's power point slides.
The discussion points included:
- Build infrastructure for handling atmospheric data
- Focus on the atmosphere and provide links to other
models for integrated GIS work
- Importance of scale of modeling: need to accommodate for
different scales, both spatial and temporal
In the afternoon presentation, Joe Breman (ESRI)
demonstrated current ArcGIS functionality using atmospheric,
oceanographic, and hydrological data. In the discussion of
ArcGIS functionality, the group identified the following
points:
- Analysis and display of vertical cross-section
- Time query/analysis (user defined; time synchronized)
- Vertical and temporal interpolation
After Joe's presentation the workshop participants formed
four working groups. Each working group was asked to discuss
the data model conceptual elements using use cases and focus
on: atmospheric and related elements, data types, data
formats, data processing functions, analytical functions, and
presentation functions. Working group's reports are summarized
below.
Report from Working Group
I:
Steve Grisé (ESRI) Terri Betancourt (NCAR) Tiffany
Vance (NOAA PMEL) Edward Dumas (NOAA ATDD) Brian Newton
(U.S. Air Force Weather Agency)
Thematic Layer Stack WG1 (Focus on the
mesoscale meteorology and common datasets used in operational
weather forecasting)
I. Satellite data (Geostationary, Polar Orbiting,
SRTM, Aerial photography, down looking sensors)
a. Raw data b. Processed data
Single Value Multi-channel - all at one or more Z
values Derived
II. Radar data
a. Raw data b. Derived products
Side note: Examples of radar coordinate systems:
Coordinate
system |
X, Y, Z
units |
CAPPI |
km, km,
km |
PPI |
km, km,
degrees |
Radial |
km, degrees,
degrees |
Important parameters to consider:
Precipitation
III. Weather data (at the surface and all upper air
mandatory levels)
Important parameters to
consider: Pressure Temperature Wind Speed Wind
Direction Height (altitude) Precipitation (amount,
type) Derived parameters and quantities
IV. Location of fixed observing stations
Example: Mesonet
V. Mobile observing stations
Example: flight path
VI. Surface data layer (to be linked with the
atmospheric data model but not included)
Parameters: Topography used by the weather
model Topography for visualization Land Use / Land
Cover Streets, roads, transportation lines,
airports Hydrography Soil types Vegetation
types Shorelines / Marine Air/sea interface,
fluxes Solar radiation
Side note: Common Temporal Operators Point in
time Interval Offset in time Buffer in time GPS
time vs UTC time Temporal bounding box Tracking/ageing
Report from Working Group
II:
Joe Breman (ESRI) Olga Wilhelmi (NCAR) Kathryn Hughes
(NOAA/ USDOC) Ted Habermann (NOAA/NGDC) Jeff Logan
(PDC) Jim Block (Meteorologix)
Thematic Layer Stack WGII: (General
representation of atmospheric sciences: meteorology,
climatology, impacts)
I. Remotely sensed observations
Radar data Satellite data
Examples of parameters: Lightning strikes
Precipitation Wind speed Wind
Direction Clouds Wind fields with barbs (should reflect
terrain features) Atmospheric chemistry
II. In-situ observations
Examples: Gauges Radiosondes Point
observations (Coop, Buoy, Metar) Mesonets Examples of
parameters: Temperature Precipitation
Currents Winds Sea surface temperature Side note:
can organize by "things that cumulate"(i.e., precipitation)
and "things that average" (i.e., temperature)
III. Numerical Models
a. Weather forecasting models b. Climate prediction
models (gridded data sets)
IV. Earth surface characteristics - land, sea,
ocean
Examples: Albedo Snow
cover Vegetation Soil types Urbanization and built
environment Land Cover
V. Socio-economic characteristics
Examples: Human dimensions elements Land use
Greenhouse emissions Population density
Watches/warnings
VII. Topography (terrain)
VIII. Boundaries
Example: Fronts
For all layers: important to address multiple dimensions:
X, Y, Z1, Z2…, Zn, M1, M2…,Mn, t1, t2, …, tn (multiple
attributes, vertical layers and times)
Side note: Need to address and represent
different scales:
Global Mesoscale Microscale Synoptic …. also
tools for downscaling and upscaling
Working Group 2 also discussed issues related to
- Differences in terminology between GIS and atmospheric
sciences (need a glossary)
- Potential uses of data model (hazards, public health)
- Data types, formats and semantic standards. To see power
point slides that Ted Habermann (NOAA/NGDC) put together for
the report click here.
Report from Working Group
III:
Lori Armstrong (ESRI) Zhumei Qian (ESRI) Ben Domenico
(Unidata) Ken Waters (NOAA/MWS) Scott Shipley
(Raytheon)
Working group 3 presented many interesting use cases where
atmospheric data were integrated into GIS. To view power point
presentation by WG3 click here.
Report from Working Group
IV:
Steve Kopp (ESRI) Simon Evans (ESRI) Nazila Merati
(NOAA/PMEL) Jack Settelmaier (NOAA/NWS) Edward Amrhiem
(US Air Force) Bonnie Reed (Raytheon)
Last but not the least, Working Group 4 was able to address
all assigned questions: from data types and formats to
analysis and representations. To see power point presentation
prepared by working group 4 click here.
After working groups' reports Steve Grisé lead a discussion
on the atmospheric concepts of the data model. The concepts
describing data collection process and further use of
observational data to derive new variables are shown on a
diagram below:
![](admwkshopmtgnotes_files/concepts.jpg)
This diagram represents some of the initial concepts that
emerged as relevant to an atmospheric data model. The diagram
is neither complete nor a UML representation, but an initial
identification of some key model concepts and their primary
associations.
Group discussion also addressed the following points:
- Interoperability and data formatting issues
- OGC WCS connector as a mechanism to provide web
services to gridded datasets
- New ArcGIS interoperability extension
- Efficiency and performance
- Users of the data model
- Potential case studies
- Experience of Marine data model with multidimensional
data
- Placeholders
- "Z-aware"
- "M-aware"
- Implications of representing grids as points
- Topology issues
Next steps:
- Draft data model to be distributed for feedback in the
next 2-3 months
- Writing conceptual framework design document (from first
NCAR/ESRI meeting in Redlands to the outcome of this
workshop)
- Conduct case studies to test data model and provide
feedback
- Document methods, data conversion issues, limitations
- Provide ESRI development team (contact Steve Kopp) with
figures describing GIS applications in the atmospheric
sciences along with descriptions on research questions, data
processing, analysis tools and what can/need to be done to
improve existing functionality.
- Next meeting will take place at the ESRI User Conference
in San Diego (present draft of the data model to the
Atmospheric SIG)
|