Corbitt Yes, Bob Jensen produces his own post-processed NDBC data. He starts from the raw spectra and integrates to derive significant wave heights, peak and mean periods, etc. The WRITE statement at the end of the e-mail below should describe the format of his files. We use: HS : Column 21 TPPFIT : Column 23 (equivalent to TPS from SWAN) TMEAN : Column 24 (equivalent to TMM10 from SWAN) WAVDVT : Column 25 (equivalent to DIR from SWAN) I hope this helps. Casey Dietrich dietrich@ices.utexas.edu ---------- Forwarded message ---------- From: Jensen, Robert E ERDC-CHL-MS Date: Tue, Jun 8, 2010 at 8:59 AM Subject: RE: To: Casey Dietrich Cc: Joannes Westerink , "Jensen, Robert E ERDC-CHL-MS" Casey Here's the deal. The NDBC buoy data I use is derived from the original archived TDF-291 data set we receive from NDBC on a monthly basis. The same data goes to NODC (NOAA's National Ocean Data Center) and is NOAA's archiving center for all data. I then decode those data, and for the directional wave buoys use the alpha1, alpha2, r1, r2, and the C11 data records to generate a directional spectral estimate using MLM, at a directional resolution of 1-deg. For the non-directional locations I use the C11 data records, and then compute all of the wave period estimates. If you only pull down the data records from NDBC (which I see you do), the AWP is defined as the "Rice Mean wave Period." I think it's like sqrt(m0^2/m2) or something like that. I really don't like that definition and use the inverse first moment which makes more sense to me. I do believe I generate a mean wave period from WAM using the same definition. As for the DWP defined from NDBC I think it's the discrete frequency band containing the maximum energy density. I also don't like using that either since we now have a conflict between the measurements and the model results derived from the discrete spectral frequency banding. A better approximation for that value (and gets us into the "continuous domain") is a parabolic fit of the peak frequency around the max energy density. That particular definition came from JONSWAP. It's a second order fit and weights the resulting value of the peak frequency based on the energy contained in the two bands on either side of the discrete spectral peak. So attempting to keep my comparisons in the "apples to apples" I just start from scratch so to speak, using the frequency spectra, and an estimate of the directional spectrum. Just to see what the differences are you might want to look at the attached file. I think I have 7 different definitions of the wave period at last count. I had to add the DELFT definitions because Roland Ris complained about what I originally had and wanted me to generate SWAN type outputs... I don't use them when I do validation, but write them out anyway. Bob WRITE (10,320) STAT,IYEAR,IM,ID,IH,IMI,LA1,LA2,LA3, * LO1,LO2,LO3,depth,ALT,SP,GU,DIR,AIR,SEA,BAR, * HS,TP,TPPFIT,TMEAN,WAVDVT,WAVDVFM,WAVPD,WAVDMX, * delfthmo,delfttp,delfttm,delfttm1,delfttm2, * delftfspr,delftmdr,delftpdr,delftsprd,sumefx, * sumefy,summfx,summfy -----Original Message----- From: Casey Dietrich [mailto:caseydietrich@gmail.com] Sent: Tuesday, June 08, 2010 8:34 AM To: Jensen, Robert E ERDC-CHL-MS Cc: Joannes Westerink Subject: Re: Bob I've been looking closer at the NDBC buoys where we thought the WAM mean periods were too high. For example, in the 'NDBC_TM01_100513a.png' image attached, the NDBC measured (gray circles) and SWAN modeled (green line) values show a peak in the mean period of about 8s, while the WAM modeled (red line) values show a peak of about 12s. However, in the 'timeptBAS42019_200808.png' image that you prepared, the NDBC measured values show a peak in the mean period of about 12s, and thus WAM is not nearly so high relative to the data. I've attached the 'NDBC_42019.txt' file from which I am working. If you scroll down to 2008/09/01, then you'll see the mean periods peaking at almost 8s. So I think I am plotting the data correctly. The question is do I have the correct data? Do these NDBC measured data match what you are working with? Thanks for your help. Casey Dietrich dietrich.15@nd.edu On Tue, Jun 1, 2010 at 1:22 PM, Casey Dietrich wrote: > Bob > > We are using the TM01 periods from SWAN, same as from the paper last > summer. I'm actually surprised at how well SWAN is doing west of the > track. For Katrina and Rita, we were much higher than the NDBC buoy > data in the west half of the Gulf. > > Casey Dietrich > dietrich.15@nd.edu > > > > On Tue, Jun 1, 2010 at 1:20 PM, Jensen, Robert E ERDC-CHL-MS > wrote: >> Casey >> >> That's true. There are probably more definitions for wave period >> than anything else. Now I am also assuming you are generating >> something that's similar... >> >> Bob >> >> >> -----Original Message----- >> From: Casey Dietrich [mailto:caseydietrich@gmail.com] >> Sent: Tuesday, June 01, 2010 12:18 PM >> To: Jensen, Robert E ERDC-CHL-MS >> Subject: Re: >> >> Bob >> >> Okay, good. I wish you wave guys would pick one mean period and >> stick with it! :) >> >> We'll be sending a draft to you within the next day or two. >> >> Casey Dietrich >> dietrich.15@nd.edu >> >> >> >> On Tue, Jun 1, 2010 at 1:15 PM, Jensen, Robert E ERDC-CHL-MS >> wrote: >>> Casey >>> >>> Whoops, you were correct.. >>> >>> Bob >>> >>> >>> -----Original Message----- >>> From: Casey Dietrich [mailto:caseydietrich@gmail.com] >>> Sent: Tuesday, June 01, 2010 12:13 PM >>> To: Jensen, Robert E ERDC-CHL-MS >>> Subject: Re: >>> >>> Bob >>> >>> Sorry, I was using column 14 for the periods. I've attached a >>> figure with column 23, and now the WAM periods are too low. >>> >>> Casey Dietrich >>> dietrich.15@nd.edu >>> >>> >>> >>> On Tue, Jun 1, 2010 at 1:09 PM, Jensen, Robert E ERDC-CHL-MS >>> wrote: >>>> Casey >>>> >>>> One thing concerning the periods. Did you use column 23 from WAM? >>>> And I appreciate you removing WAM results when they are really not >>> applicable. >>>> >>>> Bob >>>> >>>> >>>> -----Original Message----- >>>> From: Casey Dietrich [mailto:caseydietrich@gmail.com] >>>> Sent: Tuesday, June 01, 2010 11:35 AM >>>> To: Jensen, Robert E ERDC-CHL-MS >>>> Cc: Joannes Westerink >>>> Subject: Re: >>>> >>>> Bob >>>> >>>> I've attached the figures for HS, TM and DIR that we would like to >>>> include in the paper. In the figures, the red line is WAM, the >>>> green line is SWAN, and the blue line is STWAVE. >>>> >>>> At the stations you indicated: >>>> >>>> 42007: We have not plotted WAM. Jane provided STWAVE results at >>>> this buoy, so those results are used instead. >>>> >>>> 42035: The WAM periods were much too high at this buoy, probably >>>> because of the mesh resolution of the bathymetry, so I have not >>>> included the WAM results on the plots. >>>> >>>> 42019, 42020, 42055: The WAM periods are also high at these buoys. >>>> Should I remove the WAM results from these plots? >>>> >>>> 42099: The WAM results look good, at least over the period of record. >>>> >>>> Thanks again. >>>> >>>> Casey Dietrich >>>> dietrich.15@nd.edu >>>> >>>> >>>> >>>> On Tue, Jun 1, 2010 at 11:46 AM, Casey Dietrich >>>> >>>> wrote: >>>>> Bob >>>>> >>>>> Thanks for sending this data. I downloaded it from the CHL FTP >>>>> site without any problems. I will add it to our plots, except at >>>>> the buoys you mentioned, and send around the plots for review. >>>>> >>>>> Casey Dietrich >>>>> dietrich.15@nd.edu >>>>> >>>>> >>>>> >>>>> On Tue, Jun 1, 2010 at 10:58 AM, Jensen, Robert E ERDC-CHL-MS >>>>> wrote: >>>>>> Casey >>>>>> >>>>>> Looks like my mail box is full so the files I have created are on >>>>>> chlguest/jensen/4CASEY. There are 3 files, one "*.onlns" >>>>>> containing the time series for the stations you have indicated. >>>>>> The second is a "zip" file containing my time plots. The third >>>>>> is a flat ascii file containing statistics from the run. >>>>>> >>>>>> There are a few things that I would like to see. First off buoy >>>>>> 42007 should NOT be compared since the grid resolutions in that >>>>>> area are >>>> quite different. >>>>>> I am running with a 6-minute grid, whereas you have something >>>>>> that I would think is higher in resolution. This should also go for 42035. >>>>>> Basically I generate BC's for the STWAVE run, and anything >>>>>> landward of those boundaries would be considered as comparing >>>>>> "apples to oranges." I did use depth mechanisms, >>>>>> shoaling/refraction, bottom friction, and depth limited wave >>>>>> breaking. The latter mechanism (Batjjes and Janssen, 198?) works >>>>>> sometimes, and I have a strong feeling that I am in water depths that are much shallower than the buoy. >>>>>> >>>>>> One other thought is that for the Texas locations, 42019, 42020 >>>>>> and >>>>>> 42055 are to the left of the hurricane and really should not be >>>>>> considered for this study. And I didn't plot the 42099 since I >>>>>> know it had significant problems as Gustav approached. One of >>>>>> the internal circuits burned out (I think that was the problem, >>>>>> but don't >>>> quote me on it) so that data may be in question. >>>>>> As for 42003 we all know it also went belly up but did report >>>>>> some good data prior to that. >>>>>> >>>>>> Bob >>>>>> >>>>> >>>> >>> >> > Corbitt This e-mail may also be helpful to you. Casey Dietrich dietrich@ices.utexas.edu ---------- Forwarded message ---------- From: Jensen, Robert E ERDC-CHL-MS Date: Thu, Jul 24, 2008 at 10:36 AM Subject: RE: Validation Data for Katrina and Rita #2 To: Casey Dietrich Casey The word doc only describes the WAM output. Here's the NDBC file WRITE (10,320) STAT,IYEAR,IM,ID,IH,IMI,LA1,LA2,LA3, * LO1,LO2,LO3,depth,ALT,SP,GU,DIR,AIR,SEA,BAR, * HS,TP,TPPFIT,TMEAN,WAVDVT,WAVDVFM,WAVPD,WAVDMX, * delfthmo,delfttp,delfttm,delfttm1,delfttm2, * delftfspr,delftmdr,delftpdr,delftsprd,sumefx, * sumefy,summfx,summfy ALT anemeometer elevation SP wind speed DIR wind direction (met) AIR air temp SEA water temp BAR barometric pressure HS wave height TP peak discrete spectral period TPP peak spectral period (parabolic fit of the e(f) around the spectral peak) TMEAN inverse first moment of E(f) mean period WAVDVT overall vector mean wave direction (I think its also met but have to look) WAVDVFM vector mean wave direction at the spectral peak. WAVPD wave direction at the spectral peak and discrete direction band. Don't worry about the others.. Bob -----Original Message----- From: Casey Dietrich [mailto:caseydietrich@gmail.com] Sent: Thursday, July 24, 2008 10:07 AM To: Jensen, Robert E ERDC-CHL-MS Subject: Re: Validation Data for Katrina and Rita #2 Bob Yes, thank you, I received your e-mail. I am looking over the files now. In your "OUTPUTdocument.doc," I can see how the first set of format descriptions applies to the "VALID_TESTS..." files. Does the second set of format descriptions apply to the *.onlns files? It isn't obvious to me what is contained in the columns of those files. Thanks again. Casey Dietrich dietrich.15@nd.edu On Thu, Jul 24, 2008 at 11:02 AM, Jensen, Robert E ERDC-CHL-MS wrote: > Casey > > Let me know you got my last email.. There were a ton of attachments.. > > Bob >