StreamStats 3

[RWF7437]

http://water.usgs.gov/osw/streamstats/ssonline.html

Has anyone used this program ?
What have been your experiences with it ?
Have you compared it to other regression equations used in your geographic area ?

 

[TerryScan]

Have used it quite a bit. I have found it useful. Regulators in my state are promoting its use to aid in stormwater off-site downstream analysis.

I have not compared it with other regression equations, as it uses the latest equations for my state (what regulators require anyway).

 

[francesca]

Ooh very nice. Thank you for pointing it out.

 

[RWF7437]

Thanks TerryScan & Francesca.

I've tried comparing it with older ( 1978) regression equations for my state (Oregon) with disappointing results. I've also tried comparing with published FEMA flood studies; which are significantly different. It is these differences which prompted my question.

 

[RWF7437]

In an attempt to stimulate this discussion here are some questions i plan to ask both the USGS and the Oregon Department of Water Resources:

Questions:

1. Why are there considerable differencea in calculated peak flows ?
2. Why are there considerable differences in the 2 year-24 hour rainfall?
3. Why limit the slope to 5.62° ? This is only about 7%. Most of the Willamette Valley, which contains about 90% of Oregon’s population, will probably fall below this threshold.
4. Why use slope in degrees ? Most engineers, surveyors and many others use slopes expressed in feet per foot or percent.
5. Why not use some other commonly used measure of steepness (slope), such as, the slope between the 10% amd 85% points along the stream alignment ?
6. Why use the 30m DEM “elements” to calculate slope?This is nearly impossible to check without expensive GIS software.
7. If the precipitation data has been updated with an additional 20 years of record, will this updated data be used to update the published isohyetal maps? Can it be used to get Oregon into the National Weather Service’s website?:

http://www.weather.gov/oh/hdsc/

8. Will the National Flood Frequency (NFF) program be updated to use the “new” regression equations (REQ)?
9. Has ODOT, or other Otrgon reviewing agencies, adopted these “new” REQ?
10. Has FEMA adopted the “new” REQ?
11. How would one account for the effects of urbanization using the “new” REQ?
12. Who, beside engineers, uses these “new” REQ”? Hydrologists ? Climatologists ? Others ?
13. Are the “old’( 1978) REQ now considered invalid ?
14. Would you consider including examples and a tutorial with the published StreamStat program ?

 

[TerryScan]

RWF,

I am far from being a USGS hydrologist, but here are some of my attempts at responses:
1.Do you have link to the older equations you were using? (In my state, the equations are based on different variables vs. previous equations, there is also new rainfall and soils data. Unfortunately, there is NEWER soils data that is not used...more later).
3. The only reference I see to a slope of 5.62d is where the document lists the range of mean slopes found in the guaged watersheds used to develop the equations. 5.62 is listed as the LOW end of the range for region 2b. I don't see it mentioned as a limit. Remember, this is mean basin slope not slope of individual land surfaces.
4.Good question- maybe it was easier to put into equations. In my state, mean basin slope is expressed as a percentage.
5.The study mentions earlier studies that used channel slope. I'm sure they claim the new study has better standard error numbers for the estimating equations.
6."without expensive GIS"...well you have free Streamstats!
7-12 (I know even less about your local issues...I'm in Delaware)
13. I have no practical experience in performing FEMA flood studies. I did take a 1 week course on flood studies and HEC-RAS. I was led to believe they want one to use current equations.
14.The tool is fairly simple. The USER INSTRUCTIONS on the main page are good to get you going.

I think the tool is great. Like you, I had more questions about the development of the equations. In my state, we have SURGO soils complete. The reqression equations use STASGO (as does Oregon...which does not have full SSURGO coverage). STATSGO is a huge homoginization of soils. It seems odd to use this as a predictor when better information is available in my state. STATSGO soils for a watershed compared to SSURGO soils can yeild huge differences of A and D soil percentages(like 2%vs50%).

 

[RWF7437]

Thanks Terryscan.

I'll try to answer your questions in detail later but meanwhile, I've gotten the following reply from USGS to some of my questions:

"The new Oregon equations already have been added to the replacement for NFF. The software is called the National Streamflow Statistics (NSS) program. The name change was required because the software now allows solving of all types of regression equations that are available in each state, not just peak-flow equations. NSS has not been formally released to the public yet because some documentation needs to be updated and regression equations for a few states still need to be added. You can get to the NSS site, however, at

http://water.usgs.gov/software/NSS/."

It appears the newer NSS may answer many of the questions. It will also include a tutorial, I'm told.

 

[TerryScan]

RWF-

Just curious - what part of Oregon? I was in southern Oregon in 2006 for the first time and will likely be visiting again this summer. I often think about civil matters in different areas when I travel. Most places are much different than where I work (Delaware coastal plain) both physically and regulatory wise.

 

[RWF7437]

Terryscan,

I live in Salem, OR which is in the Willamette Valley; midway between the Cascade Mountains and the Coast Range. This is where most ( about 90%) of Oregon's population resides. Climate is usually described as "modified marine" and characterized by mild but wet winters and warm dry summers. Most flooding occurs during the winter, wet, months of September through March. Rainfall in the Valley is generally caused by large low pressure systems sweeping inland from the Pacific Ocean causing long duration, low intensity storms. These may be combined with snow melt from the nearby ( 50 miles or more) Cascade Range. This was the case in 1964 when the entire west coast, from Alaska to Baja California, experienced major flood damage.

Some flooding may occur during summer thunderstorms but these are usually smaller, and rarer events.

I have long answers to your previous posted questions but I'm waiting for a response from the author of the Oregon Regression Equations, Dr. Cooper, who was nice enough to call me this morning. Meanwhile, I'll try to educate myself about the Delaware coastal plain, where my brother in law, PhD Civil Engineer, now lives.

 

[RWF7437]

Terryscan,

Let me attempt to answer your questions, and mine, in this post.

First, the most important questions we both have concern how to use these equations and internet tools in our engineering practice. Your earlier post suggested that some public agencies may be adopting StreamStats and the regression equations (REQ) as "standards". Once such things are adopted by agencies who view their role as primarily "regulatory" we are all obliged to live by them as though they were cut in stone and handed down from some mountain top.

This alarms me. I can only hope any such agencies will understand what it is they are adopting and what the limitations of such "standards" are. But experience with some agencies suggests that my hope is forlorn.

Case in point: Dr. Cooper, author of the Oregon REQ has pointed out to me that the equations should NOT be used outside their original data limitations. Doing so will lead to large, unpredictable errors. Because most of my work deals with watersheds which do not fit within the data limitations of Dr. Cooper's study his equations will be of very limited usefulness to me. Dr. Cooper's study is excellent work but it cannot, and should not, be used unconditionally.

Apply this to your own practice and State of Delaware. I suspect you'll find similar limitations. If your State adopts the REQs developed for Delaware be sure to read the background report from which the REQs were derived. Because Delaware is small, has a more uniform climate and few, if any, mountains your problems will probably be fewer than we have here in Oregon and other western states.

All of these things are changing. The NFF will soon be replaced by the NSS program. StreamStats too is being changed. I, for one, would like to see these things used for a while in real world practice, before anyone adopts them as the Holy Grail of Hydrology.

 

[RWF7437]

"1.Do you have link to the older equations you were using? (In my state, the equations are based on different variables vs. previous equations, there is also new rainfall and soils data. Unfortunately, there is NEWER soils data that is not used...more later)."

I can send you the equations for Oregon in a spreadsheet (Excel). So as not to incur the wrath of the website Gods, you can find me at rwf7437 AT comcast DOT net.( case sensitve)

"3. The only reference I see to a slope of 5.62d is where the document lists the range of mean slopes found in the guaged watersheds used to develop the equations. 5.62 is listed as the LOW end of the range for region 2b. I don't see it mentioned as a limit. Remember, this is mean basin slope not slope of individual land surfaces."

It is a limit in the sense that any flows calculated using a slope outside this range will contain large, unpredictable errors. It is also a limit in that most of the drainage basins I might need to study, those in the central Willamette Valley, are much flatter than this lower limit.

"4.Good question- maybe it was easier to put into equations. In my state, mean basin slope is expressed as a percentage."

Glad to hear that. For Oregon this slope is expressed in degrees. This is a minor disadvantage.

"5.The study mentions earlier studies that used channel slope. I'm sure they claim the new study has better standard error numbers for the estimating equations."

The "new " (2005) Study for Oregon has about the same peak low error range ( plus or minus 30% to 40%) as the "old" (1978) Study. So the "improvement" is not significant.

"6."without expensive GIS"...well you have free Streamstats! "

Yes, free but nearly useless for most things I might need to do.


"13. I have no practical experience in performing FEMA flood studies. I did take a 1 week course on flood studies and HEC-RAS. I was led to believe they want one to use current equations."

I've read hundreds of flood studies and written dozens. You have probably used at least as many. As far as I know, to date, FEMA has NOT accepted any of these programs or methods. Given their limitations, I would think FEMA ahould NOT accept them, if only to be consistent with past practice.

"14.The tool is fairly simple. The USER INSTRUCTIONS on the main page are good to get you going."

Yep.


"I think the tool is great."

Nope.

" Like you, I had more questions about the development of the equations. In my state, we have SURGO soils complete. The reqression equations use STASGO (as does Oregon...which does not have full SSURGO coverage). STATSGO is a huge homoginization of soils. It seems odd to use this as a predictor when better information is available in my state. STATSGO soils for a watershed compared to SSURGO soils can yeild huge differences of A and D soil percentages(like 2%vs50%)."

I don't know what the acronyms SURGO or STATSGO stand for and have never had a need for them. Because most studies I do are in urban or urbanizing areas, soild are less important predictors of runoff. When much if the area, say 50% or more, has been paved or roofed over, the soil can have much less effect on runoff.

Even in rural areas, other factors such as tree cover, crop cover, tillage practices, slope, and similar human and natural factors are possibly more significant.

All of which suggests to me that we should spend more time installing stream gages and calibrating hydrologic models than we do.

 

[RWF7437]

http://pubs.usgs.gov/sir/2006/5146/

 

[TerryScan]

When I first read your concern with the 5.62d limit, I thought you were implying your slopes were greater (my misread). Isn't 5.62 degrees closer to 10%? wow.

Have no fear. I have great misgivings regarding Streamstats being mandated as part of stormwater management. Expecially the ways being suggested here.

Aside from the limitations of the regression equations themselves, they are looking at mixing regression equations with a "adjusted" TR-20 calcs. This sounds like apples and orangutans (oranges seemed too similar).

They are looking at a method being set up in Marlyand:

http://www.gishydro.umd.edu/

You may find it interesting reading. It scares me a bit.

The concerns here are regarding downstream analysis. They are considering the "10% rule" of looking at the impacts of developement up to the point where a site is less than 10% of the contributing watershed (and analyze at the next channel junction downstream from that).

I assume Streamstats is then considered in this affair because it is easy. It would not be resonable to require survey and detailed modeling for area 900%+ larger than every development site.

 

[TerryScan]

Looking at the equation study for Western Oregon. It really does seem odd that the majority of area designated for 2b equations has slopes mapped at less than 3.22 degrees, but no stations were used less than 5.62d.

 

[RWF7437]

Still not heard from:
WA
ID
CO
IL
IN
OH
PA
VA
MA
CN
TN