INR Fuels Project: Rich Gwozdz

2010.09.16

rich gwozdz — Thu, 16 Sep 2010 22:05:34 +0000

Finished fixing bugs in PresideOut and ran new FVS-based Path simulations for the WCR models. Compared results to expert opinion models – look pretty similar with the exception of WCR_fss. Am waiting for feedback from Miles, Josh, and Emilie. Also continued FVS procedure for the other west Cascades models.

Also continued working on the Path Graphing software requested by Miles. I combined all the classesSummary.csv files from the WEC Path simulation. I removed the unnecessary fields and any timesteps greater than 100. The resulting file was about 1.6 gb, which is small enough to link to in MS Access. Not sure how many timesteps the Path simulations are suppose to be, but hopefully not more than 100.

2010.09.02

rich gwozdz — Fri, 03 Sep 2010 05:53:06 +0000

Completed preside step for two PVTs – WCR_fmc and WCR_fmt. This step is a bit tedious, but it should be complete by Tuesday. Now I probably just jinxed it…

2010.08.31

rich gwozdz — Wed, 01 Sep 2010 04:27:26 +0000

Finished all of the FVS natural growth runs for the WCR. One last step in this process and that is the Preside steps. Looks like it will be a bear, but with any luck I will be done with it sometime next week.

2010.08.17

rich gwozdz — Wed, 18 Aug 2010 00:35:33 +0000

Had a FVS setback today. Due to a quirk in the way the DNR data is stored, a single stand could have been simulated multiple times for a given PVT. The caused havoc with the calibration statistics, etc. Luckily I caught the error in step five of Eric’s procedure. It wasn’t exactly easy to fix, but after some python scripting and SQL querying I’m back on track. But I have to start the procedure over to ensure data quality.

2010.08.16

rich gwozdz — Tue, 17 Aug 2010 03:22:05 +0000

FVS-Preside work continues. I’m doing each of the PVTs in parallel. In this way I hope to save time and avoid the “how do you do this again?” questions that might arise if I go through the entire process of each PVT separately. Currently I am at the step where the stand max density and BA get set.

2010.08.09

rich gwozdz — Mon, 09 Aug 2010 22:51:46 +0000

Day was split between trouble shooting FVS and building the summary documents for the expert opinion models. Finally was able to get FVS to run the SLF formatted stands — there had been a series of path problems that had prevent the data from getting loaded correctly. However, I am still unable to get the database held stands into FVS for simulations. I have Erin Crosland at the DNR giving me a hand trouble shooting – hopefully will receive some insight soon — ah wait, Erin just got back to me. She has given me something to go on!!!

I have not yet finished the summary documents for the expert opinion models of western Washington. But they should be done soon.

Rich

2010.08.06

rich gwozdz — Fri, 06 Aug 2010 22:23:18 +0000

Not much progress on FVS today – still getting all 0′s for all stand runs. hopefully will be able to talk to Eric on Monday.

I did run the expert opinion models for the WCR, WNC, and WWC today using equal area initial conditions. I loaded all the results in SQL Server and generated the summary plots with R. Will make official documents on Monday.

2010.08.02

rich gwozdz — Mon, 02 Aug 2010 23:43:22 +0000

Still trying to figure out fire tranisition probabilities for the WCR. Josh pointed out that the 0.5 multipliers probably weren’t carried into Path form VDDT. Indeed, they were not. Reran the simulations with these multipliers. they just finished so, I will evaluate tomorrow in the AM.

Spent the other portion of the day working with the DNR FVS stand data. Wrote a python script that pulled each stand’s ecoclass code out of its database and wrote to a file. Was able to see that there were some ecoclasses that we had not developed crosswalks for, so I sent those to Miles.

Then I wrote a second python script that will update all the DNR databases with the appropriate VDDT model. The script works and is ready to go once Miles returns the completed crosswalk.

2010.07.30

rich gwozdz — Fri, 30 Jul 2010 22:56:50 +0000

Some what of a difficult couple of days diagnosing the strange problems with the rollup tool. The source seems to be coming from the strata aggregation tables produced by EBG’s R Code. If you run those codes numerous times on the same data, the results are usually not the same (but can be) — more simply, you get different export tables when they should be the same. EBG tried to fix quickly on Thursday – and it looks like the problem is indeed less severe. But unfortunately the code is still not producing repeatable results. Will have to wait to fix until EBG returns from leave.

On other fronts….another glitch in FVS simulations. DNR data is in access format while CVS data is in SLF format. This may not be such a big deal. Weikko put me in touch with Erin Crosland at the DNR and they suggested a work around. Will hopefully have time to try on Monday.

Re-ran the WCR PVTs that MH said looked strange – changed fire probabilities per MH methods but they produced some very strange results. Also thing there is a problem in the MCM file that JH produced for me. It is causing WindHiSev transitions more often the WindLoSev transitions, even though the respective probabilities would suggest otherwise.

2010.07.29

rich gwozdz — Fri, 30 Jul 2010 22:35:11 +0000

Quite a long hiatus for me on the blog. I’ve been getting so wrapped up in IT issues here – seems like it has been one thing after another. But in the midst of all of that, there has been some progress on the western WA regions.

WCR:

GNN crosswalks completed, and rollup successful (minus aggregation)
some model testing and review by Miles, which is still in progress

WWC and WNC:

GIS team delivered the data layers
WWC GNN crosswalks complete and added to rollup;
WWC rollup tested and functional
Will do initial testing on models soon, pending adjustment suggested by MH for WCR