How to optimize Wx & PrecipAccum grid generation in the forecast process

Forums (Forecast Builder)

RSS (Opens New Window)

Kathleen Torgerson, modified 7 Years ago.

How to optimize Wx & PrecipAccum grid generation in the forecast process

Youngling Posts: 6 Join Date: 3/20/15 Recent Posts

I am running into an issue with FB in that it seems I am either not using FB efficiently, or perhaps the tool itself could be optimized. So here is my scenario that I need help with...

I go through and edit all my foundation grids, including ProbIcePresent for drizzle/freezing drizzle (on my plains), SnowLevel, T grids, SnowRatio, QPF etc, then run through the remainder of FB, editing my Potdrizzle, PotFreezingDrizzle and PotFreezingRain grids as necessary (mostly to remove it out of the mountains/valleys where I expect Rain/Snow) and add some PotFog or whatever. The tool eventually finishes, and I look at my SnowAmt grids and usually find areas where I need to make non-trivial improvements due to terrain effects. This means I need to go back and make changes to foundation grids (QPF, or SnowRatio, or SnowLevel or whatever), then re-enter the tool at step 4. But Step 4 overwrites my PotDrizzle, PotFreezingDrizzle grids so I need to redo all the changes to those Pot___ grids for my mountains vs plains which takes up valuable time. And (unfortunately) I may find that I forgot to hit that stratiform button the second time through step 4, and/or perhaps I'm still not happy with how my SnowAmt grids turned out in some complex terrain locations and need to make further edits, thus requiring me to re-enter the tool again and re-working all those Pot_____ grids that get overwritten each time. So my thought is, would it be possible to separate the sections of FB that compute the wx grids, from the sections that compute SnowAmt, IceAccum etc, so that the user can be more efficient? Or is there a method that I haven't figured out yet to better utilize the tool in its current form? Right now it is adding too much time to my forecast process in complex weather scenarios, and I don't appear to be getting any faster with it.

Hopefully this description is understandable. Thanks in advance for any insights.

Kathy Torgerson

NWS Pueblo, CO

Andy Just, modified 7 Years ago.

RE: How to optimize Wx & PrecipAccum grid generation in the forecast proces

Padawan Posts: 99 Join Date: 6/2/15 Recent Posts

Kathy,

What you have found is the concept behind ForecastBuilder - that you perform much of your work with the Foundation Grids (and top-down for those offices utilizing that approach), then let the precipitation type probabilities, SnowAmt, IceAccum and Wx all fall out based on various statistics and science. This set up was installed because if you start editing those precipitation type probabilities, your forecast can get out of whack meteorological consistency wise real quick! Plus it's kinda messy. So what might help with your situation is for the ProbIcePresent grid, ensure values < 100 are only in locations where you expect freezing rain/drizzle. If when you see the Snow/Ice accums and they're not what you like, as referenced in the training, go back to the foundation grids and adjust and collaborate as necessary.

Don't worry too if you struggle a bit on this process. I know this was something offices were trying to do as well in the early stages last winter. It takes a bit of getting used to, and it helps ensure both meteorological and office to office consistency.

Let me know too if there are any tools/techniques you might need for ProbIcePresent. It sounds like perhaps there are typical areas that should stay as rain/snow. You could have that defined as an Edit Area (store it under one of the 1-7 numbers in the upper right of GFE), then call it up when editing ProbIcePresent and assign 100.

Kathleen Torgerson, modified 7 Years ago.

RE: How to optimize Wx & PrecipAccum grid generation in the forecast proces

Youngling Posts: 6 Join Date: 3/20/15 Recent Posts

Hi Andy.

So, if I understood correctly, I should have repeated step 4 of FB to re-work my ProbIcePresent grids until my Pot____ grids came out asdesired before proceeding to rework QPF(and other relevant foundation grids) to get my SnowAmt grids in order? I think I understand now. We were still dealing with the bug with the "QPF value separater" so I did have to edit the Pot____ grids some. But now I can see this logic. Its not clear to the user in step 4 which calculates first, the precip accum grids or the Pot____ grids. The way I was doing it, I was undoing my work each time I re-ran FB at step 4, but I do see now that if I could have perfected my ProbIcePresent grids from the start, I might not have run into this problem.

So my case was actually a little more complicated than my initial description, and I was using the tool in the short term (which I know is not the expectation at this point, but I had a good case of ongoing mixed precipitation to put it through its paces). We had bands of snow spreading off into the adjacent plains as deeper moisture and lift associated with an upper trough was advancing overtop shallow upslope. So I had bands of heavy snow very near areas of drizzle/freezing drizzle across the I-25 corridor for a few hours before the event switched to all snow. I had a difficult time getting my ProbIcePresent grids reflective of the current situation, and didn't have my gradients in exactly the right places (though I did try). Not to mention, where it was snowing, it was snowing hard! So updating my QPF/SnowRatio etc to calculate snowAmt grids in the same step was a bit mind boggling in that fast paced environment. But it led me to this thought...

What if we could have a "back" or "recalculate" button on the FB procedure gui which shows up at the end of step 4. That way, one could make changes to the foundation grids, and rerun step 4 without having to watch the grids disappear/reappear in the grid manager each time we cancel, then restart FB. Of course in researching this idea tonight, I discovered the "Go to Main Menu" button at the end of step 4, which appears to do pretty close to what I'm describing in that at least I can refire step 4 without having to watch all my weather elements unload and reload. Unfortunately, if you've clicked around on the derived Pot___ grids or SnowAmt grids then hit "Go to MainMenu' and re-run step 4, it doesn't run the tool over the same time range the procedure was initially launched with. This seems like a gotcha. I would think it should preserve running the tool over the initially selected time range since you haven't really exited the procedure. Is this a bug? I'm also wondering if could we add all relevant foundation grids (for the calculation of the Pot___ and accumulation grids) to the grid manager during step 4. (ex: Snowlevel, SnowRatio, ProbIcePresent, T grids, etc). That could save time in not having to bring up the weather elements browser to load those grids and rework them before hitting the "Go to Main Menu" option then re-running step 4.

Anyway, sorry for the long post. I have 2 mixed precipitation cases (and 4 shifts) under my belt, and I'm still trying to hone this process down in my mind in order to get more efficient with the tool. Any additional thoughts/guidance much appreciated!

Kathy