SN 16 day forecasts and WAMs

carpetman's picture
carpetman started the topic in Monday, 4 Nov 2013 at 12:43pm

Hi Ben,

As continued from can't view reports thread....

thermalben wrote:

16 day forecasts are very useful for tropical locations who receive long range groundswells (ie Indo, Sth Pac).

In Australia, the 16 day forecasts are pretty good for tracking long range swell trends in WA/SA/Vic and southern/western Tas.

These long range forecasts are less reliable for most of the kinds of swells that the East Coast receives - however we used them recently to track a small long range trade swell that dished up 2-3ft surf in SE Qld and Northern NSW. 

Overall, it depends on the kind of weather system you're looking at. But even high res models can get it wrong at a day or two out.. nothing is perfect.

carpetman wrote:

So for the places like Vic/SA/WA we'll have a good understanding of the underlying long range swell which is generally on the smaller side due to increased swell decay but will typically keep somewhere like Mornington 2-3ft for most days of the year. But should a more dominant localized swell pop up, the long range swell may in comparison be insignificant and not be felt?

Also, I know you can't give anything away but your WAMs give a really-really-really long forecast range. I thought the longest available WW3 was 180hrs? Do you get your data from other places or have you built your own to run for the extra 204hrs?

I assume, as discussed above, it's pretty good for tracking the long range swells but how reliable is it at forecasting a storm 2 weeks in advance?

Kiwi123's picture
Kiwi123's picture
Kiwi123 Monday, 4 Nov 2013 at 2:21pm

Hi Carpetman, yes, you are correct that currently ww3 only provides forecasts out to +180 hours.

In my previous post in this forum I asked if the forecasts were based off the NOAA WW3 model forecasts or not, and Ben said they using an in-house ww3 model.

So based on that, I am guessing they are running their own version of ww3 (which is a freely available model) and running it using the NCEP 16 days GFS weather model forecasts.

You can see an example of the 16 day GFS here http://www.weatherzone.com.au/models/?lt=wzcountry&lc=aus&mt=mrf . You will need to register a "pro" account, which you can a free one month trial for, and then click on the US 16-Day link.

From memory, the 16-Day GFS is essentially the same model used to drive the 10-Day WW3 forecasts, but just run for longer. It may have a slightly lower model grid resolution, but I am unsure off the top of my head.

carpetman's picture
carpetman's picture
carpetman Monday, 4 Nov 2013 at 2:59pm

Ah, ok got ya. Thanks. GFS runs out to 384hrs but at intervals of 12hrs instead of 3hrs, not sure whether there is a difference in grid resolution. I thought GFS was a completely different model putting out wind, pressure data etc. not wave data. But maybe the WW3 model is run on the GFS model?

Anyway, my question has been answered. Thanks.

Kiwi123's picture
Kiwi123's picture
Kiwi123 Monday, 4 Nov 2013 at 3:07pm

Yip correct again. The NOAA WW3 forecasts are based upon satellite altimeter and wave buoy (and most likely a few other sources) data to determine the present sea state. The ww3 model then forecasts the growth, transformation and decay of waves due to a number of processes such as wind forcing (from GFS) and effects due to the ocean floor such as diffraction, shoaling friction etc...

thermalben's picture
thermalben's picture
thermalben Monday, 4 Nov 2013 at 3:41pm
carpetman wrote:

I thought GFS was a completely different model putting out wind, pressure data etc. not wave data. But maybe the WW3 model is run on the GFS model? Anyway, my question has been answered. Thanks.

You got it - WW3 based on T384 GFS input data.

Kiwi123 wrote:

The NOAA WW3 forecasts are based upon satellite altimeter and wave buoy (and most likely a few other sources) data to determine the present sea state.

Not quite. Altimeter and buoy data is only used for verification. The WW3 model runs independently on GCM wind forcing, and creates a self-initialised restart file every model run which is then used to kick off the next run.

Kiwi123's picture
Kiwi123's picture
Kiwi123 Monday, 4 Nov 2013 at 3:45pm
e wrote:

Not quite. Altimeter and buoy data is only used for verification. The WW3 model runs independently on GCM wind forcing, and creates a self-initialised restart file every model run which is then used to kick off the next run.

I knew that it runs independently.

Ahhh thanks, it has been awhile since I last looked at the specifics.
So the initial conditions do not necessarily represent the actual global wave conditions at T+0 ?
What do you mean a self-initialised restart file?

thermalben's picture
thermalben's picture
thermalben Monday, 4 Nov 2013 at 3:55pm
Kiwi123 wrote:

So the initial conditions do not necessarily represent the actual global wave conditions at T+0 ? What do you mean a self-initialised restart file?

Yeah, initial conditions may not always be right (and that's where a lot of 'errors' in wave model forecasts originate from - poor analysis/initialisation). Where would they get wave data in the deep Southern Ocean from?

Re: self-initialised restart file - it may not be the proper technical term but what it means is that the model uses its own forecast data to create the next 'analysis' timestep. So if you're running the model every six hours, it'll use T+6 as the next run's 'analysis' data (which itself is the previous analysis time step plus six hours of wind forcing).

Kiwi123's picture
Kiwi123's picture
Kiwi123 Monday, 4 Nov 2013 at 3:59pm

Thanks for the reply. Ahhh, ok, so it is essentially using a previous forecast as the initial conditions.

I guess I was over-estimating the instantaneous global coverage of wave buoy and altimeter data.

Thanks

thermalben's picture
thermalben's picture
thermalben Monday, 4 Nov 2013 at 4:18pm

Yeah there's lots of buoy coverage in the Northern Hemisphere (particularly around North America and Hawaii due to to NOAA's incredible buoy program) but there's very little coverage in the Southern Hemisphere. And most buoys are typically located close to the coast and are therefore attenuated by local bathymetry, which makes it a little harder to quantify the accuracy of the wave models.

donweather's picture
donweather's picture
donweather Monday, 4 Nov 2013 at 4:33pm

Ben, up until earlier this year, didn't the Jason 1 Altimetry provide a way of calibrating modelled wave height versus actual wave heights?

thermalben's picture
thermalben's picture
thermalben Monday, 4 Nov 2013 at 4:35pm

It does provide some verification input ('assimilation') - as does ENVISAT - but the satellite tracks are very narrow and don't cover much ocean.