Laserfiche WebLink
<br />Hydrologic modeling: <br />GIS databases can be used as preprocessors to establish internally consistent databases for the <br />hydrology. It is possible to represent the various hydraulic controls in a simulation mode to <br />represents the physical system in high detail. Rainfall distributions (from the archives) can be <br />placed over the study area watershed [in a dynamic mode?]. Then by looking at the different <br />scenarios of storm events that have actually occurred (either historic or nearby and with <br />adjustments for meteorology, rainfall imagery, etc. as we have talked about), this will help to <br />better define the spatial distribution of soil ( or surface) factors such as permeability. This will <br />lead to a better understanding of the response of the whole watershed to the modeled storm <br />events. Since we aren't completely sure of storm patterns it's a good idea to use various <br />scenarios for infiltration, etc to look at the range of possible outcomes. So with GIS and <br />computers we can look at not just one or a couple of "probable" scenarios, but rather a suit of <br />systems and events that will enable us to beat the problem to death. It's a task but computers <br />make it much easier and faster. <br /> <br />2D Hydrodynamic (Distributed) Model: <br />Operates on a DEM. Grid resolution is variable (lm to 1km, this example used 500m sq. grid). <br />It takes account of soil moisture, initial conditions, etc. that are appropriate for the watershed and <br />recovery between storms based on a relatively simple mathematical representation (kinematic <br />wave algorithms?). It's not appropriate for lowland floodplain modeling but is appropriate for <br />foothill and mountain watersheds. Lynn has used the 6- minute radar rainfall imagery (bias <br />corrected) to pass a storm event over a watershed and look at the watershed response. Results <br />were pretty good. The F2D model routes flows in the eight directions ofthe grid. It takes into <br />account the hydraulics of overland flow and channel flow (network topology), soil texture based <br />on the federal databases, soil moisture, % impervious surfaces, etc. [used DOA GLEANS <br />climate, temp, wind data ?] <br /> <br />F2D Model Output: <br />Showed simulation, distributed modeling for 24 hr cumulative 'infiltration' and rainfall-runoff <br />'output' for Buffalo Creek using NEXRAD data. They varied infiltration and surface runoff <br />parameters. Soil moisture is initialized daily. Nothing was done by hand (except "fine-tuning" <br />the sensitivity analysis based on fieldwork and "tweaking" some of the rain gauge data as <br />deemed appropriate using reasonable probability ranges (?). The weather service looked at the <br />results and found it to be reasonably accurate. <br />George commented: that the Dual Polarized data may give even better results. <br /> <br />F2D Model Calibration and Uncertainty: <br />Also, rain gauge data and runoff response for Ralston Creek and Goldsmith Gulch was used to <br />simulate 14 historic events. They ranI 000 "parameter set" simulations... but after 8 simulations <br />can eliminate the ones that give lousy results Uses a Monte Carlo approach called GLUE <br />(British model) to rank parameter sets as simulators of the basins response to soil infiltration <br />parameters, surface and channel roughness and slope. Essentially estimating probability <br />distribution functions. <br /> <br />One can, for multiple events, accept different "sets" of parameters from the 2D model that give <br />you reasonable results for the monitored runoff. With each run, by eliminating parameter sets <br /> <br />3 <br />