Dreamflows Bulletin Board
 
"I can't boat everyday, but I can always Dream"
 
Home River Flows Bulletin Board Custom Email Reports Alerts Signup Login Help About

 Topic:  Estimate accuracy 

Forums -> Flows -> Estimate accuracy Page 1 of 1
Posted to flow site: - Latest PostNext TopicPrev Topic
 Author  Message 
chris
Site Admin
Posted: Sep 9, 2008Post Subject: Estimate accuracy
Marc Hoshovsky, who was backpacking up there, reports as follows: "I was just in Cherry Creek Canyon above Cherry Lake this past weekend. There is no flow in this section now, except sometimes where we found a 1-inch wide trickle from one isolated pool to the next".

The Dreamflows estimate is still showing some flow (cycling between zero and about 70 cfs, for the most part). BUT - as you can see from the above post, it should be showing zero cfs. The formula used to estimate the flow came from Hetch Hetchy Water and Power and I assume they got it right. I assume therefore that the reason for the difference is that one or more of the many gauges in the formula is a bit off. With 1000 cfs flowing here, 800 cfs there, it's not too surprising there's something left over in the calculation that shouldn't be.

So the estimate isn't totally accurate, even though it is based directly on gauge data. It should still be useful during snowmelt for telling you when Upper Cherry is "in", though, by following the trend that you see on the graph.

guido
Member
Posted: Jul 7, 2009
I was in Upper Cherry from July 1-4. I used this gauge extensively (along with the T above HH) to plan my trip. I think it was useful in telling me the basic trends of the creek (although I noticed that there are often corrections to the gauge that appear after the fact), but I'm not real sold on much else. The sudden change in the character of the graph that occurred around July 2nd was not something that I observed on the river - it looks more like a calculation artifact...

without getting to involved, it looks like the calculation went from using two components to just one. While that might make sense somewhat when west cherry dries up, I don't think it dried up in a single day... Any thoughts?
chris
Site Admin
Posted: Jul 10, 2009Post Subject: Re:
guido wrote:
I was in Upper Cherry from July 1-4. I used this gauge extensively (along with the T above HH) to plan my trip. I think it was useful in telling me the basic trends of the creek (although I noticed that there are often corrections to the gauge that appear after the fact), but I'm not real sold on much else. The sudden change in the character of the graph that occurred around July 2nd was not something that I observed on the river - it looks more like a calculation artifact...

without getting to involved, it looks like the calculation went from using two components to just one. While that might make sense somewhat when west cherry dries up, I don't think it dried up in a single day... Any thoughts?


That's useful information, Eric, thank you. The Cherry Lake inflow is a calculation based on six (yes six) different gauges, and therein lies the problem. Especially since one of those gauges measures reservoir elevation, notoriously inaccurate. Especially since those gauges aren't aligned in time, so I have to adjust based on estimated flow travel times. Hetch Hetchy don't attempt a "realtime" computation of Cherry inflow even for their own purposes, they just do a daily average. But that's just an average and by definition always at least a day late, so I thought I'd use some sleight of hand to create a realtime computation.

Yes, corrections do occur after the fact, I've noticed that with normal single realtime gauges sometimes too ... if the owner of the gauge (or the gauge itself) decides to adjust something then what used to be reported as X now gets reported as Y. Dreamflows accesses flow data for the same date/time several times for various reasons, and if the data changes then the results of the computation will change too. However I personally haven't adjusted anything this year yet, so any changes have been due to the incoming data.

The biggest problem with the Upper Cherry computation is that Holm Powerhouse releases factor into it, and they are very sudden very significant (up to 1000-cfs) changes in flow. We get a double-whammy there, since the Holm releases affect Cherry Lake too, but with a time delay. The computation has been doing a credible job of smoothing those fluctuations out ... most of the time ... but now Holm is the major flow contributor to the equation, it's not doing so well. I'll work on it.

The later into the season we get, the more Holm releases will screw the computation up. What will help me is to know what flow range to "tune" the computation for. I've asked several paddlers who paddle Upper Cherry for information over the past year but no word has come back yet ... perhaps you can help? What flow range are you looking at? Or, if that's easier, can you let me know what date(s) you had marginal, optimum and high but okay flows at? That would really help.
Thomas M
Member
Posted: Jul 11, 2009
Hi Chris, I was at UCC on july 4-5. I would have called the flow low side of good. I am not very good at guessing cfs but it seemed somewhere around the 200 cfs range. I was told by others that took off July 2 that the flow was a bit high for the cherry bomb gorge.It seems like flow came down to a lowish level somewehre on July 3 or 4th. Good luck on estimating that one.
guido
Member
Posted: Jul 12, 2009
Just to follow up, we were actually on the river for 4 days waiting for Cherry Bomb to come in (July 1-4). As far as I know nobody ran it before July 2nd. The late afternoon of July 2 and noon of July 3 were both on the high side of good for the gorge. I know some folks ran it on the morning of July 4, so maybe they can chime in on that level. If I was to guess a level for when we ran it (noon July 3) I would guess around 300 cfs. Flows seem to peak around noonish near the gorge.

Hope that helps
madmiguel
Member
Posted: Jul 25, 2009Post Subject: cherry bomb july 4
I was in there july3-4 with a few different groups. I did not man up enough to run da bomb my friends ran it july 4 and said it was on the high side of good not low side. One of them ran it again on the 5th and said it was perfect even though he liked the flow just as much on the 4th. One thing I noticed that the graph was not consistant with was the daily fluctuation. In reality(stick gauge)it dropped pretty steadily with minimal daily fluctuation. On the gage it seemed to almost double in flow every day. I think the T gage above HH is the best gauge to use. 500-600 on that gage seems closest to perfect.
chris
Site Admin
Posted: Jul 25, 2009
Thomas, Eric and Mike - thanks for your input, that's great. I'll work on this during the off-season and hopefully we'll have something much more usable and accurate next year. I'll keep you posted.
chris
Site Admin
Posted: Apr 25, 2010
Having studied the graphs and taking into account the preferred flows for Upper Cherry, this is what I've found. This time of year, it's easy to produce a very nice graph for Upper Cherry, with well-formed waveforms that coincide with the peaks and troughs predicted by Hetch Hetchy. But at the point in time you guys run it, the flow has dropped enough that the "noise" from the Holm releases makes a total mess of the graph, obscuring the important stuff. So, I'm discontinuing realtime graphs.

Instead, I'm computing a midday-to-midday daily-average flow as follows. The average flow from noon on day 1 to noon on day 2 is reported as the average flow for day 2. So when you see the average for today (Sunday) is 1000 cfs, what this means is that the computed average inflow to Cherry Lake from Upper Cherry from noon yesterday to noon today was 1000 cfs. Hopefully this new graph, together with the T above HH, will give you what you need to dial in when the flows are right.

Note that usually daily averages are computed midnight-to-midnight, however doing that means that you don't get the information until the following day (in the case of Dreamflows you wouldn't get Sunday's average flow until 5pm on Monday). A midday-to-midday computation is designed to get useful flow information to you sooner.

For reference, the averages computed in the new way for July 2009 were as follows - remember that "July 1" is really "noon June 30 through noon July 1":

July 1426 cfs
July 2353 cfs
July 3218 cfs
July 4204 cfs
July 5180 cfs
July 6169 cfs


As always, feedback welcome ...
Thomas M
Member
Posted: Jul 21, 2010
Hey Chris,
I just wanted to say that the estimated flows seem pretty close on for this year. We put on UCC on 7/19 and paddled out 7/20/10. I saw that the estimate said the flow rose slightly between the 19th and the 20th however we saw a drop of water lever from mon to tue. My opinion is that the flow on Mon was medium or the low side of medium and the flow on Tue was low.
Edited once.  Last edited by Thomas M Jul 21, 2010.
guido
Member
Posted: Jul 23, 2010
I also think the flow estimates were pretty close. We put hiked in the 20th, ran the put-in slide, paddled to flintstone on the 21st and then out the 22nd. I think there might be a little noise in the flows at that low level (seemed like the levels dropped slightly each day, not down to 65 then up to 130), but definitely in the ballpark. I feel like I would trust this gauge for planning a trip. The only downside is waiting the 24 hours for the flow to post; it can put you behind the curve a bit. Not sure if there is a solution though...

Funky year, really dropped out, but I guess that is to be expected on bigger snowpack years. For the record, the run was on the low side of low, but it made everything really low-stress. All the big drops went fine except for West Coast Gorilla (probably fine, but nobody felt like getting beat up) and Dead bear (very green landing). As a plus, you can paddle out of the pocket in Kiwi...
Edited once.  Last edited by guido Jul 23, 2010.
guido
Member
Posted: Jul 28, 2010
by the way, if you want to check out pictures at those flow levels - http://classvstout.blogspot.com/2010/07/ucc-how-low-can-you-go.html

Thanks again Chris for all the hard work!
Forums -> Flows -> Estimate accuracy Page 1 of 1