Hydrologic Modeling and GRASS

[Chris Rewerts]

vieux at quercus.ecn.uoknor.edu writes:

>This is an informal survey of people using GRASS in some
>way to model hydrologicprocesses.  If you are using GRASS in this way
>please respond to the following questions.  

>1. What model are you using with GRASS?

ANSWERS (Areal Nonpoint Source Watershed Environmental Response 
Simulation)

>2. Which data structure is more suited to the process, Vector or Raster?

ANSWERS is a distributed parameter model, meaning it divides the watershed
area in a series of equal-area grid cells, much akin to a raster GIS. 
So in the case of the ANSWERS-GRASS integration, raster is the more suitable
structure. Of course, many of the raster maps used for inputs to the model
were once in a vector format...

>3. What tools need to be developed in order to make GRASS more useable for 
>	hydrologic modeling?

My integration is not a deep integration - in it I treat the ANSWERS program
as a 'black box' for which I use GRASS programs to provide interface for
the purposes of preparing inputs and processing outputs. This reduces the
time and expertise needed to use ANSWERS, since the most volumous and tedious
inputs and outputs are spatial which can be stored, manipulated, analyzed
and displayed by the GIS. 

So, this integration is not a cutting-edge GIS development. It does make
ANSWERS, a model that has been used/tested/validated/compared since the
70's, more accessable to today's GRASS user.

Thus, my answer to this question is to point to further development
of open GIS, like GRASS, to provide for (and I am borrowing this
term) a Geographical Modeling Framework. Such a framework can provide
for deeper integration of process modeling. Another general need relates
to data - which is often the most expensive component of a GIS - anyone
doing spatial modeling needs it and can benefit by better methods of
collection, delivery, conversion, standardization, documentation, and
so on. I may as well tag something else to this paragraph, to 
re-iterate an often-mentioned need for GIS abilities for dynamic
processes, including more sophisticated means for handling 
temporal spatial data, such as map formats that contain different views
of a map theme over time (or maps that you can make changes to
but that can re-create their previous states).

But now to a couple specifics. Hydrologic modeling requires good
elevation-based data, such as slope, flow directions, and/or slope
length. We have found that how a DEM is rendered into these inputs
can have a drastic effect on the predicted results of our models. This
points to a need for more exploration and testing of these processes.
There are a number of tools work with elevation data to produce the
maps that are needed with hydrologic modeling, but they all have
their foibles, IMHO.

I could go on, but I think it will suffice to say that there is
plenty of work in these areas to keep us GIS-types busy for as
long as there is a disire to abstract reality into computer-palpable
manifestations.

>Thank you for your time. 
>Baxter Vieux
>Environmental Modeling and GIS Lab.
>School of Civil Engineering and Environmental Science
>University of Oklahoma

By the way, the ANSWERS/GRASS integration is now on the moon
ftp server (moon.cecer.army.mil 129.229.20.254) in the 
grass/incoming dir - files are answers.cpio.Z and answers.README.


-- 
Chris Rewerts                     Spatial Analysis and Systems Team
Constructon Engineering Research Laboratory, Environmental Division
rewerts at zorro.cecer.army.mil            1.800.USA.CERL/217.352.6511