HECWFS Huron to Erie Connecting Waterways Forecasting System RMA2 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 5/13/2008 Stopped RMA2 script. HECWFS now beling modeled using FVCOM ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/31/2008 Added links to GLERL GLCFS, GLERL AHPS, Current Conditions, St. Clair R. WebCams, and Connecting Channels Forecast pages. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/24/2008 Added a companion page to display St. Clair and Detroit River products. Also, added GLOS link and logo to sponsor list. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/18/2008 Put hecwfs.sh on the cron. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/10/2008 Problem on DOY 042. High winds. Tried modifying the wetting/drying DE card. TI 0 8 0.0005 0.0005 DE .275 .6 4 originally from Eric: DE 0 0 4 This didn't help, blew up at same timestep: 5.4, but I left it as DE .275 .6 4 Cut winds in half for 1st cycle (hr 0-6). This worked. Reset winds at hour 6 and continued. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/10/2008 Had problems over the weekendon DOY 030. High winds and dramatic drop in water level at Gibralter. The model couldn't get through the day w/o blowing up. I increaded niter to 50 and both increased and decreased the convergence criteria (from .0001 to .005). Nothing worked. I didn't try increasing the number of timesteps per hour from 10 to say 20 because run_bc.pro is not set up for this. Dave suggested cutting off the winds for this day. I did so, setting ws and wd equal to 0 in run_bc.pro and running for 4 cycles. This worked. I then set cc back to .0005 and niter to 8 and restarted the catchup runs from DOY 031 00. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/6/2008 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Started running the model with niter=2. It blew up around day 10. Changed niter to 8 - model successfully got past trouble spot (day 10-11). I looked into the TI card: niter and convergence criteria. The manual states that the dynamic model typically takes 3-5 timesteps to converge. I set niter to 8. The manual suggests that the convergence criteria be set to 0.005 to 0.0001 for steady state, and between 0.05 to 0.001 for dynamic runs. Eric had the convegence criteria (cc) set to 0.0001. I reset to 0.0005. For a 6-hr run, the number of iterations is about 2-3, occasionally 4 or 5. The run time for these settings is about 19-22 minutes. The runtime for cc=0.0001 and niter=8 was about 36 minutes. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 3/3/2008 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Started running model from DOY 001, 2008. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2/29/2008 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ When using a hotstart file, need one extra boundary condition set at start BHL 2 999.999 BHL 1 999.999 BW 1 99.9 99.9 END This set is not used, it presumabbly corresponds to the last timestep in the hotstart file. The next set is used for the 1st timestep. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2/20/2008 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ orig code from: /zeus/home/anderson/stclair/rma2/ /zeus/home/anderson/stclair/nowcast/ Compiled using: lfc -O3 --parallel -o rma2.x *.f ../sou/rma2/rma2.x < c200805000.in > c200805000.out lower water level we use is Gibralter. The model extent goes further down to Bar Point, Ont. Therefore, we subtract 0.326 feet from Gibralter to estimate Bar Point (EJA & DJS) For near-realtime Bar Point water levels, see: http://www.glos.us/ http://www.glos.us/conditions.php#levels http://www.great-lakes.net/envt/water/levels/levels-cur/stclairwlc.html http://www.waterlevels.gc.ca/C&A/wldata/barpthis.htm For historical Bar Point data: http://www.waterlevels.gc.ca/C&A/gs_erie_e.html http://www.meds-sdmm.dfo-mpo.gc.ca/ http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Home_e.htm http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/TWL/maps/Maps_b.asp?IncludeFile=b_pt_glhb.gif http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/TWL/TWL_e.htm http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/TWL/TWL_inventory_e.htm http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/TWL/TWL_station_list_e.asp?user=MEDS®ion=CA&tst=1 http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/TWL/TWL_station_period_e.asp?user=MEDS®ion=CA&tst=1&no=12005 01-JAN-2008 to 01-JAN-2008 http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/TWL/TWL_data_display_e.asp?user=MEDS®ion=CA&tst=1&no=12005 water levels in feet. winds in m/s. wd=270-wd wind direction is direction toward which the wind is blowing, measured in degrees counterclockwise from the positive x-axis. So wd=270-wd. For example, an E wind (90) blows toward the W which would be 180 degrees counterclockwise from the x-axis. ;niter=2 ;no. of iteration per timestep, calmer periods niter=8 ;no. of iteration per timestep, high wind periods water temp=10 (set to constant in rma2) Put this file in directory with other in/out files to increase print buffer size. Default values cause the execution to fail. r2memsize.dat user MND MEL MR1 MFW NBS MPB MCC MCCN MXSTRM MXUU MWTS5 50000 20000 150000 1200 60000000 100000 150 300 1 100 10 TR 0 -10 0 0 to sace ascii results to frl file after every 10 time steps (hourly) TS 0 0 -60 to save hotstart only at end of 6 hr run (60 time steps) 1 hr takes 7:35 minutes of run time, niter=8 908.801u 0.894s 7:34.97 199.9% 0+0k 0+0io 0pf+0w 1 hr takes 2:49 minutes of run time, niter=2 337.193u 0.765s 2:49.06 199.8% 0+0k 0+0io 0pf+0w Missing node reports... when cell dries out, the frl file will not have a value for that node. Also, the srl will not have a timestamp and value for that node (if it was one of the saved nodes via the TRN card). The output file may report total number of inactive cells for each timestep (this may be the number of dry cells).