This page gives information about current issues with the guidance plotting system, as well as planned upgrades and improvements. If there is a problem or an outage, I will try to note the cause below.  

Here is the current status of the system:

Current Notes

14 Nov 2006, 1635 UTC: This site has been restructured into two sections: one which displays the TC model guidance from open sources (NHC, UK Met Office) and another section which displays the model guidance from protected sources (JTWC model guidance). The latter section is only available to users at CSU's Department of Atmospheric Science (ATS) and the Cooperative Institute for Research in the Atmosphere (CIRA). Also, the script system has been extensively revised and simplified. One feature of the new system is that each current storm will now be plotted based on the basin in which it is currently located (not where it originated). So a user should click on the 'frame' links under the basin where the storm currently is located. The file naming convention for the archive plots will still be according to a storm's originating Basin/Storm Number/Year (e.g. aal092006).  Also, all active storms which have updated plots will be displayed in the text box near the top of the page. Finally, the new system maintains the correct access privileges for the plots, regardless of where the storm is. So if a Central Pacific or Eastern Pacific storm crosses the Dateline into the Western Pacific, it's protected plots (derived from JTWC data) will only show up on the protected page.

Upcoming Changes

A new extended range product maybe added to the suite. Some models (the GFS ensembles in particular) go out to 384-h (16 days). While these forecasts are extremely suspect given how far out they go, it will still be interesting to look at!

A change is also planned for the image conversion method. The current method generates high quality images, but they are quite large - 170-400  Kb! The new method should retain most of the quality, but shrink the image sizes down to a more manageable 75 Kb or so.

Eventually, I'd like to do away with the frame links and go to a dynamic page structure, perhaps with drop down menus for the plots from each storm.

Ultimately, I plan to make an archive of model guidance for storms from previous years. This should be of use for researchers doing case studies on track or intensity, modeling, and for the training of TC forecasters.

Previous Notes and Issues

25 Oct 2006: The West Pac, North Indian, and Southern Hemisphere basin plots are now being made again. The plots for those basins had been down since July 31 due to a hard disk failure on the machine which provided them to the guidance plotting system. Since the plots for this basin are only available to CSU users, the outage only affected these users.

28 September 2006, 0300 UTC: A MAJOR revision has been implemented for both the driving script system and the plotting program. The script system has been streamlined for efficiency and reliability, and it now will drive the plotter program for all major TC basins worldwide (except the South Atlantic). Error checking and logging capabilities have been added, so if a problem does occur in the future, it should be easier to ascertain the cause.

The plotting program has undergone changes which involved hundreds of lines of code. It now reads the abr-deck files directly instead of relying on intermediary conversion code, leading to increased reliability. The new abr-deck format includes latitude/longitude specifiers so the plotter is now able to plot for TCs worldwide, irrespective of whether they cross the International Dateline, Primer Meridian, or even the Equator (should that ever happen). Previously, forecasts were only shown every 12-h. Because the full abr-decks are now being read, data can now be plotted for the off-standard intervals (such as every 6-h). The track lines may not seem as straight any more, but this more accurately reflects the results from each model's tracker. Another major change is that each model's tracks are not constrained to start at the operationally-assessed bogus location. The old plotter basically ignored the 0-h forecast positions. The new plotter plots these if they are different than the bogus location - the result is messier, but also more useful because the user can now assess the impact of initial position error on each model's forecast. If a given aid has no 0-h forecast position, then it is assumed to start from the bogus position and plotted as such.  

29 August 2006, 0300 UTC: Access restrictions on the plot images are now in effect. Plot access is now denied for all requests coming from external links (otherwise known as 'hot linking'). Embedding image tags from another site is tantamount to stealing. At the very least, such unscrupulous behavior is wasting CSU's bandwidth while not giving any 'benefit' to this site. So in effect, those who set up 'hot links' to images on other sites are stealing bandwidth. Given that I'd like to keep this resource available to legitimate users I have chosen this block this type of behavior. Bloggers or media who wish to download an occasional plot are still welcome to do so - but please host the image on your own site following the updated media guidelines for redistribution. 

Additionally, access is now blocked to the 'Hurricane Owl' widget (written by Walter Schulze). This widget downloads hurricane images from a variety of sites, including this one. It has a bug in the refresh rate, causing images to be downloaded every 15 sec to 2 min. Since the images on this site are actually only updated every 6 h, this is a HUGE waste of bandwidth (~1 Gb per day per instance if the user leaves it running on their desktop). Thus, I've blocked access to this widget. 

19 Aug 2006, 1400 UTC:    There was an access outage for all users starting at 0330 UTC on 18 Aug 2006. This was due to an inadvertent mistake on my part. It was fixed at 1400 UTC on 19 Aug 2006.

19 Aug 2006, 2030 UTC:    Some users experienced an access outage, starting at 2030 UTC 17 Aug 2006. I had implemented a robots.txt file to try to stem the rising tide of referral spam occurring in the usage statistics.  Referral spam occurs when nefarious bots hit a site while passing false referral information - the idea is to get listed in a site's referral list. When a search engine like Google indexes this list, now full of links to all over the web, the usage page may get a high page rank and therefore give free advertising to the nefarious sites. The robots.txt file prevents search engines from indexing my usage statistics in their index, thereby denying any advantage to the nefarious referral spammers. Apparently, some legitimate user agents were blocked when I first implemented the robots.txt file - this problem has now been fixed. 

NHC has updated their verification page and issued their verification report for 2005 (pdf).