COD vs Twisterdata vs HAZWX ... Is there a difference

[Andrew Clope]

Personally I like HAXWX just because it's easy to navigate, it's overlaid on a googlemap setting, and it's just easy to use.

But I've noticed that even running the same model/same initialization times that they don't necessarily agree.

Which is most accurate in you guys'/girls' opinions?

 

[David Williams]

I like CoD. They are still working on the site, but I rely on it predominately. Also, it's free. I can't compare it to HazWx since I don't have a subscription, but I've heard of the many convenient features.

I just started using TwisterData more often; however, it's definitely a supplemental/secondary source for me after CoD.

 

[Jeff Duda]

Each of those sites has advantages and disadvantages. I can't say one is better or worse than the others.

CoD
Advantages:
-Pretty good looping system (although the labels for forecast hours for the GFS are still messed up)
-Point-and-click model soundings with indices and parcel paths AND the ability to switch between surface-based and mixed-layer parcel paths AND hodograph availability. This is HUGE for this site. Unfortunately I have noticed many times that the indices like CAPE and SRH in the window that comes up with the sounding does not agree with the value obtained by examining the contour plot.
Disadvantages:
-Missing some useful products:
-Why can't I see 2-m dewpoint with wind barbs and MSLP? The streamlines and 30-mb mixed dewpoint doesn't make much sense to me and has always bothered me.
-General bugs that have not been fixed since they revamped the site over a year ago
-Lack of products for SREF, 4 km NAM, and HRR
-I would love to see 0-1 km shear and lapse rates for all models, not just those mentioned above

TwisterData
Advantages:
-Quality graphics library
-Point-and-click model soundings and hodographs
-Expansive suite of dProg/dt graphics (CoD has those, too, but only the four most recent runs)
The above includes a satisfying archive of recent runs (goes back a few days for GFS)
Disadvantages:
-Can't zoom in on any particular area. There are occasionally chase setups that have a very narrow window of opportunity and the spacing of the wind barbs makes it very difficult to tell where the sweet spots are from time to time. Also difficult to see what is going on in the immediate vicinity of fronts
-Animation/switching forecast hours is a PITA

HazWx
Advantages:
-Improves upon the Google Maps interface that TwisterData uses, including zoom and pan capabilities
-Offers TONS of products that you don't see on the other sites, including 0-3 km CAPE, VGP, convergence, frontogenesis etc.. This is the case for ALL models, including HRRR and 4 km NAM.
Disadvantages:
-Coarse resolution data when zoomed in (no tiling of forecast imagery)
-Like TwisterData, animation or switching forecast hours is a PITA
-No dProg/dt that I know of

There are other sites you should consider also. With the exception of CoD (which does a poor job at it), none of the above sites include probabilistic forecast information from the SREF, GEFS, or any storm-scale ensembles. The SPC site includes SREF and SSEO graphics, which are pretty cool and helpful. Really the future of forecasting is in ensembles and I'd love to see all sites moving towards probabilistic information. On finer and finer scales, a given forecast from a deterministic model like the NAM or GFS control member can be worthless on fine scale details. Also you would have no idea what the certainty of the forecast was, nor its dependency on forecast hour or location.

 

[Shane Adams]

Twisterdata doesn't have regional maps.
Haz-wx isn't free.

Therefore, CoD.

 

[Dan Robinson]

I'm still using UCAR's graphics for RAP, NAM and GFS: http://weather.rap.ucar.edu/model/index.php

I've always liked their presentation best. Since the UCAR images are GIFs with the same URLs for every new run, I was able to build my own bookmarks/display pages ( http://stormhighway.com/data/modelwall.php ) where I can compare UCAR's GFS/NAM output side-by-side as well as show multiple parameters (CAPE, precip, 500mb, 850mb, surface, etc) at multiple times in a grid format.

Aside from their lack of regional maps, the drawback of UCAR is that some helpful parameters are missing (their GFS output has no CAPE maps, for example), and nothing new has been added in many years. Still, I find that using their maps via the bookmarks pages allows me to see all of the information I need quicker - much faster than having to click back and forth constantly to view maps individually.

 

[Skip Talbot]

Another big advantage to CoD that hasn't been mentioned is their speed. The model graphics come out quickly and download quickly. That's part of why they only have a few products from the HRRR, so they can process each run quickly. Jeff mentioned that there are several bugs that need to be worked out. One to look out for that might bite you is old HRRR data getting labeled as the current run. You download what's listed as the current run of the HRRR, but it's actually using yesterday's data. It can be really misleading if yesterday had a similar pattern too.

Can't speak for HazWx, haven't tried it.

Don't forget Earl Barker's site if you're looking for some convective products that aren't found many other places:
http://wxcaster.com/smallfiles_central_svr.htm

That site's layout is brutal, and the main reason why I created my own viewer for it, like Dan made his own for UCAR's graphics:
http://skip.cc/chase/models/

A few of those are broken as links have changed and need to be updated. You can tell which models I use most often though.

 

[Royce Sheibal]

I've been using COD and Earl's website since I started chasing. Both are invaluable to me. COD works smoothly on android, which is primarily what i use to assist me on the road. Earl's website, however oldschool, loads incredibly fast and uses very little data when you're stuck out in the boonies. The 3 sites always i have bookmarked and pre-saved on every computer and phone I own are COD, Earl's, and stormtrack.

 

[rdale]

On the note about HazWx changing times - once you select a time just use your arrow keys. Not quite such a PITA

 

[Rob H]

CoD - uses a crappy routing system where everything is dynamic and based off the same URL. It would be awesome if you could do something like weather.cod.edu/forecasts/nam/latest/surface/dewp or something instead of having to use their cumbersome navigation system every time. Their product selection, sector views, and comparisons are awesome. Having fewer products for things like the HRRR are a feature in my mind, since it gets delivered blazing fast. My go-to forecasting site.

TwisterData - I like that the 240hr and 384hr runs of the GFS are smooshed together. Data seems reliable, if a little slow, compared to other sites. I can't ever recall seeing TD with old/bad data. No regional sectors is crappy, and the Skew-Ts and hodos are very minimal and borderline worthless. This isn't a horrible choice for a site to look at, but it doesn't really do anything better than any other sites.

Earl Barker - an abomination of a site, but has those hard to find charts like ESRH, LLL3, etc. Spend 4 hours digging through every nook and cranny to find what you want, bookmark and label it well, and never use anything besides those bookmarks. The only certainties in life are death, taxes, and Earl's site layout.

Haz-Wx - I understand paying for models like ECMWF, I don't understand paying for free, gridded data overlaid on Google Maps. The screenshots I've seen look really pretty, but in forecasting I like easy to find and lightweight images in minimal interfaces. I need to quickly churn through lots of products. I don't want to download Google Maps API and run a ton of scripts on a tablet in the Nebraska sandhills just to see a chart of 500mb winds.

There are other sites you should consider also. With the exception of CoD (which does a poor job at it), none of the above sites include probabilistic forecast information from the SREF, GEFS, or any storm-scale ensembles. The SPC site includes SREF and SSEO graphics, which are pretty cool and helpful. Really the future of forecasting is in ensembles and I'd love to see all sites moving towards probabilistic information. On finer and finer scales, a given forecast from a deterministic model like the NAM or GFS control member can be worthless on fine scale details. Also you would have no idea what the certainty of the forecast was, nor its dependency on forecast hour or location.

For spaghetti plots of the GFS, I'll either roll my own in IDV or use NCEP's model guidance site. It's nice for looping and seeing how the GFS converges on a solution over time, especially in regards to the jet stream and troughs/ridges: http://mag.ncep.noaa.gov/model-guidance-model-area.php

SPC's SREF is used quite a bit by me as well: http://www.spc.noaa.gov/exper/sref/

I personally like ensembles as well, but I haven't seen a visualization that includes something crucial - how the data is trending. If the position of a warm front keeps trending north, the mean position is kind of worthless.

 

[Jeff Duda]

I personally like ensembles as well, but I haven't seen a visualization that includes something crucial - how the data is trending. If the position of a warm front keeps trending north, the mean position is kind of worthless.

Time lagged ensembles, my friend. We have discussed this in private before. It may not be the "traditional" method of presenting ensemble forecast information, which is where dProg/dt comes in and why I built the GFS TLE ensemble products on my site. I stopped running it because I didn't find it to be all that useful. A given forecast (including a forecast the day of) can vary wildly from the trend, just like the verification can be wildly different than the ensemble mean from a given traditional forecast. That's why it's important to look at ensemble forecast uncertainty

 

[Rob H]

Time lagged ensembles, my friend. We have discussed this in private before.

Yeah, and it went over my head then

In my mind, a simple "darker color is older, lighter color is newer" visualization would work well. Or opacity, or line thickness, or whatever makes the most sense for the data.

 

[Mark Blue]

Don't forget Earl Barker's site if you're looking for some convective products that aren't found many other places:
http://wxcaster.com/smallfiles_central_svr.htm

When it comes to forecast SkewT plots I go straight to Earl's website to get the information I'm after. There is a boatload of information below the SkewT and Hodographs as well. Precipitable Water, Supercell Potential, LIs, SRH, CAPE (all flavors), EHI (1 - 3 Km), VGP, STP, BRN Shear, LCLs, and even Skip's LSI is there. It really is a lot to digest from one chart! Here are the three I use in case you want to bookmark them.

Archived SkewT's: http://www.wxcaster.com/archivetext.htm

GFS, NAM, Meso-ETA 120 Hour SkewT's: http://www.wxcaster.com/model_skewt.htm

HRRR, NAM, WRF SkewT's: http://www.wxcaster.com/moreskewts.htm

Everything under the sun model-wise: http://www.wxcaster.com/models_main.htm

As far as the last link goes, you'll just have to hunt and peck to find the charts that work for you. This is the cumbersome part mentioned above by Rob. Just bookmark those you find useful to cut down on valuable time you'll need in the field. Hopefully this is another tool you can use for future forecasts.

 

[Michael Towers]

HRRR, NAM, WRF SkewT's: http://www.wxcaster.com/moreskewts.htm

Thanks, I didn’t know Earl had added new forecast soundings. As previously mentioned he has great some stuff but navigating through it all can be tedious so I’ve basically had my main resources of interest from his site individually bookmarked and haven’t visited his main page in ages. I prefer Earl’s forecast soundings above all others and the HRR soundings will likely become a useful tool in my chase day analysis. Time to visit his main page and see about any additional resources I may not know about.

 

[David Williams]

Besides a few brief mentions, I'm surprised no one bolsters their forecasts with anything from the SPC forecast tools page. They have data rich skew-t charts and hodos, some strange things like percentages for anticipated CAPE values ( which I'm not exactly sure why you would need) , and some ensembles... if you're into that sort of thing

Also they boast a pretty great archive. I particularly like the hourly surface observation maps. It seems like they have everything archived that you would want; although, I admit I don't go there very often for archived information.

 

[Shane Adams]

Besides a few brief mentions, I'm surprised no one bolsters their forecasts with anything from the SPC forecast tools page.

I use it during the morning hours of a chase day for monitoring surface trends, but once we hit the road, I just use UCAR's surface page. Just too graphic intense for a phone (at least mine) and doesn't update nearly as often. With UCAR I can just hit 'refresh' every 15-20 minutes and get a look at how things are setting up.