Hi all.Just wanted to throw in my 2 cents... Tom and I have talked about the NAM nests, but till we get CONDUIT upgraded and can increase the LDM queue, it's just too much data. So in the spring we can talk about this addition.
Becky On 12/5/2011 3:23 PM, Ethan Davis wrote:
Hi Geoff, No, no buses. Thanks for the info. Yes, the Best Time Series does seem to be a popular view of the data. Once we start getting the data, we'll make it available as Best Time Series. Thanks, Ethan On 12/5/2011 12:04 PM, Geoff DiMego wrote:OK that is probably correct and I am jumping the gun. We wanted FWIS to go onto CONDUIT first to get the IMET folks started making the connections to/thru FXNET. This would mean the other 4 fixed location nests would have to come later. I hope I didn't throw anybody under the proverbial bus. That said, I do think what Rich would be able to use most often are Best Time Series from the nests which are fixed and provide guidance out to 60 hours. On 12/5/2011 1:13 PM, Ethan Davis wrote:Hi Geoff, Right. Sorry, I missed that detail and jumped right to FWIS for some reason. We aren't getting the new NAM nest data (4km CONUS, 6km AK, etc.) on our server. I checked with Tom Yoksas and he said it isn't on CONDUIT yet. Linda, do you know the status of that data for CONDUIT? Becky Cosgrove's presentation to the User Committee mentions that data on a "Recent Model Additions" slide but I don't recall the details. Thanks, Ethan On 12/2/2011 12:37 PM, Geoff DiMego wrote:Ethan - thanks for doing the update on NAM documentation. I think what Rich was asking for are Best Time Series for the fixed domain NAM nests, not the FWIS which as you say can be and often are in different places each run. On 12/1/2011 4:06 PM, Ethan Davis wrote:Hi Geoff, I've made the documentation changes you requested to the underlying files. I will push them out to our server shortly. The "Best Time Series" aggregation is generated by our FMRC (Forecast Model Run Collection) aggregation. The FMRC takes a time series of runs from a given model and creates several virtual aggregated datasets including a 2 time dimension, 5-D dataset and several datasets with a single time dimension. This AGU poster describes the different single time dimension datasets http://www.unidata.ucar.edu/staff/caron/presentations/FmrcPoster.pdf For all the NCEP model data we serve (http://motherlode.ucar.edu:8080/thredds/idd/models.html ), the underlying data is the GRIB data we get from NOAAPORT and CONDUIT. We read from those data files (for which we have built indexes) to satisfy requests for data from the virtual datasets. Since the spatial domain of the FWIS can change for each run, we won't be able to do a full FMRC aggregation. Each FMRC aggregation expects all the data for each run of the model to be on the same grid. So, we can serve the data from each FWIS run on its own. But we'll have to think about what else it makes sense to do, e.g., "Best Time Series". I'll touch base with Rich and ask what he is trying to do with the data, how he would like to access it. Geoff, can you explain how the location of each FWIS run is determined? Is it an automated process? Or does someone/group decide? Also, are there times when the location stays the same across runs? Is the information on the domain for each run available somewhere besides the grid description in the resulting GRIB files? Thanks, Ethan On 12/1/2011 7:51 AM, Linda Miller wrote:Hi Geoff, Comments inline below. On 12/1/2011 7:31 AM, Geoff DiMego wrote:Linda; Hope all is going well for you and yours especially now that the hectic holiday season is upon us ^_^ I hate to be a bother, but I have a favor (yes ANOTHER one!) and a question to ask.No apologies neededA USGS user had asked me whether we would be making "Best Time Series" aggregations for the new NAM nests (4 km CONUS, 6 km Alaska, 3 km Hawaii& 3 km Puerto Rico-Hispaniola). I pleaded ignorance. He pointed me at http://motherlode.ucar.edu:8080/thredds/catalog/fmrc/NCEP/NAM/CONUS_12km/catalog.html?dataset=fmrc/NCEP/NAM/CONUS_12km/NCEP-NAM-CONUS_12km-noaaport_best.ncd which I admit I was unaware of - my bad, but I recognize the unique& useful aspect of this product which you are constructing.If you ever get to Boulder, you should plan to visit Unidata. We could bring you up to speed with what;s going on here. You might find some of the projects of interest to NCEP.The favor is to update the 3rd Summary bullet under Documentation to: NCEP Nonhydrostatic Multiscale Model on B-grid (NMMB) running in the NOAA Environmental Modeling System (NEMS) infrastructure, and Gridpoint Statistical Interpolation (GSI) analysis. You could also add a bullet called NCEP Analysis documentaton / tutorial and link it to http://www.dtcenter.org/com-GSI/users/tutorials/11_2.php .I'm sure we can do that. I've cc'd Doug Dirks and Ethan Davis who can check it out and make the appropriate edits.My question comes from the one the USGS guy asked: are you planning to aggregate Best Time Series for the new NAM nests?I'll bet the "USGS guy" is Rich Signell, who is a liaison to our Users Committee. He definitely keeps us on our toes. I'll defer your question to Ethan. Hope your holidays are joyful! Linda