|
giswr
> events
> feb 2000 giswr
conference > general feedback
General Feedback
The following document summarizes the
feedback of the conference attendees to the breakout
session questions.
Synthesis of Feedback
GISWR, 24 February 2000
Notes Prepared by David Arctur, ESRI
- What is an object? Data that has structure, relationships
and behavior. An object might be a computation algorithm
with little/no persistent structure, or a policy for how
other objects might be built/used; not just a bit of data
with behavior.
- Should we be working on an Arc 8 model that incorporates
behavior needed for hydrological modeling into the database?
- Probably not initially; hydrology models too complex
to implement efficiently
- What about using component models
- Use geodb as a preprocessor to external models; possibly
even glue-ware for external processes.
- Want to explore interface between GIS data and an external
modeling system; ways of structuring relationships between
features
- What types of models are appropriate based on purpose,
scale
- Why develop a cross-sectional model view of channels?
Mainly to support hydraulic models, but it could be separated
and ignored from rest of core hydro model by users not
interested in the overhead (would take a little work to
remove, but is not fundamentally required by the rest
of the hydro model).
- FEMAīs DFIRM GIS model
- How will objects interact? Need to build object state
and interaction diagrams.
- Much data acquired from FEMA or contractors is out of
date on arrival and needs significant update how to
deal with this problem?
- Need to be able to support at least a few tiers of users
& applications, from heavy-GIS to novice-GIS to non-GIS,
and they should all come up with same answer to the same
questions on a given set of data. Should be able to apply
ActiveX components for use across different applications
and user levels.
- Objects / applications needed:
- Channel routing (open channels), incl. channel losses,
diversions, dams
- Channel hydrologic routing, scale dependent
- Track quantity of flow (q) down the river, something
different from Tracking Analyst since that just works
with ArcView. Write numerous methods on the flowline
for alternative algorithms. Routing paramīs from other
objects (eg, soil type from raster).
- Track water quality
- Track overland flow, rainfall gathering
- Generate hydrographs
- Atmospheric models for precipitation; where precipitation
occurs is as important as what is being modeled. Canīt
necessarily use landform-based aggregate rainfall expressed
as 25% urban, 75% rural
- Relation to OGC: this hydro group should coordinate
with OGC to see whatīs coming down in form of service
architecture, catalogs, metadata, feature models, etc.,
as well as to feed back to OGC the results of this groupīs
work
- Big issue is integration of geodb with legacy data &
processing
- Public libraries of objects built yet?
- Need to choose the Best Old Code as candidates for object-wrapping
in a GIS framework
- Mature core routines in Fortran, Army Corps of Engrīs
HEC LibHydro of watershed, stream & channel routing
components of HEC-1
- USGS MMS: Modular Modeling System categorizes types
of data, types of models for surface water analysis;
written in C/Motif to work on UNIX, Linux (NT in process);
supports C++, Fortran routines with Java wrappers. (Roland
Viger, USGS Denver)Ņ Versatile architecture: a consultant
has sold and value-added the model to the Pentagon to
model failures on jets.
- An effort is needed by qualified staff to design the
wrappers and related architectures, as well as to search
for additional packages to be adapted for use in this
framework.
- Need resources among participants in form of spatial/hydro/computer
programming skills to build these wrappers
- Kim Davis for Jack Dangermond: ESRI will write the
code, based on the Consortiumīs design, and ESRI will
sell the tools to use the hydro model.
- BUT Lori / Dale: much ESRI resource has been diverted
to ArcFM water/wastewater utilities, but hydrology group
could get some attention.
- Potential problem with FEMA submittals on proprietary
models solved if vendor will provide source code for
the proprietary model to FEMA (under non-disclosure
for vendorīs protection)
- Conversation with DaleH / ScottM / David Maidment: get
the data modeling straight first, before worrying too
much about adding object functionality
- Open Source vs. commercial software: validation routines
may be good candidate for open source
- Data transfer between models is important
|
