2012 Site Updates

Some links and images are no longer available in our older site updates. Additionally, some information is no longer applicable.

For the latest site updates, click here.

Today while rewriting some of the RECCO decoder I came across some unfortunate errors. As a result, RECCO data has been incorrect at times when it comes to the conditions along the flight route, type of icing and in rarer cases where in some instances "/" was incorrectly interpreted by Perl as a "0". The other decoders have undergone more vigorous testing while this one was our least used decoder. Therefore, the errors went unnoticed. Because of the errors I have removed all recon data from our site until it can be reprocessed. This includes Google Earth data as it includes decoded RECCO data. (I did reprocess the data for Sandy.) The online RECCO decoder has been updated to fix these errors and any additional errors will be fixed as I find them.

As a result of removing historic recon data, I might make significant changes to the recon decoder sooner than I planned. I did not want to interfere with the recon archive but now I am forced to remove it anyway. There may be significant issues with the recon system at times if I proceed.

On October 27th we let people know about our Airborne eXpendable BathyThermograph (AXBT) decoder that is currently being tested which is now available in our manual recon decoder. The decoder is not yet part of our live system and remains highly experimental. We made it available so people could see some ocean temperature readings in Sandy from NOAA aircraft. The AXBT decoder will be more stable when our recon system is rewritten in 2013. (AXBT format examples: Newer NOAA format | FM 63-XI Ext. BATHY (JJVV) | Example of seemingly older NOAA format) We hope to be able to have Air Force data at some point in the future too, perhaps on a delayed basis.

The rewrite of the dropsonde decoder is still going well. I still need to update the dropsonde diagram some and then also make some updates to be a little more compatible for a small amount of messages in the 1990s that had remarks after that main part of the report. The new decoder handles things differently than the old one when it processes the data so I need to make sure that it can still be as versatile. The new version of the decoder is not yet available.

The rewrite of the dropsonde decoder is going well. I still don't know if I will release it in beta form when I am done or wait until next year when the next version of the recon system is ready for beta testing. I still have to redo the dropsonde diagram and a few other various little things. The new decoder will decode GlobalHawk dropsondes when it is released.

I am rewriting a lot of the dropsonde decoder. In trying to allow dropsondes from the Global Hawk, I realized the entire decoder needed to be rewritten. Although the code worked, it was more complex than it needed to be. It may take a month or so to complete. All the testing is being done offline and once well tested it will be made live at some point in the future. I might just wait until I rewrite the other parts of the recon system.

I have been very busy this week doing something non tropical and have not yet finished updating the dropsonde decoder in my offline testing to handle Global Hawk dropsondes and therefore I have not yet updated the recon system to handle those sondes. I thought about rushing to release it so that it would at least display everything below 100 mb, but given the very complex nature of the dropsonde decoder, I would rather not release anything until it is well tested. There is very little information that I can find on decoding the "51515 10190" group above 100mb. It would be nice if someone knew what 70mb, 50mb, 30mb, 20mb, 10mb altitude's decoded to. I have not yet got extremely deep into that testing though, so I may yet figure it out. It's kind of reverse trial and error. I am trying to create a fake dropsonde AVAPS file to run through ASPEN to try to see how it treats the actual heights of those levels in the "51515 10190" group so I can try to figure out what the three digit height should decode to. It's just hard to create the fake data to do that. 70mb and 50mb seem to be used by Global Hawk so I definitely need to figure that out at the very least. If anyone can give me a raw AVAPS file (.avp) from each of those sondes it would be great! (One for the Global Hawk when the "51515 10190" group is at 70mb and the other at 50mb.) I still have not tried cross checking it with flight level data nor have I actually done the decoding of the significant levels yet above 100mb so that may yet help me to figure it out. Plus, I have to adapt the online system to handle a new agency, NASA. I'm still not sure how involved a task that may be. I will update the dropsonde decoder to handle NASA when I get to that and then eventually I will add it so that if NASA ever released any of the tropical products the Air Force releases or any of the other products NOAA releases it would also be available in the live system for those products as well. Given the work left to do, I doubt I will be able to display Global Hawk sondes in the live system before the HS3 mission ends in early October. Maybe, because I would like to test it live, but I make no promises.

While dropsonde data from the NASA Global Hawk does currently come across with the header UZNT13 KWBC, our site does not yet decode it. It will take some time to upgrade the decoder to handle the sonde data since it includes groups XXCC (mandatory pressure level data above 100 mb) and XXDD (significant pressure level data above 100 mb) which our site does not yet decode. Once the manual decoder can handle these additional groups, it will also take some time to add a new agency, NASA, to the live system. I am currently rewriting the entire recon system. I'm not yet sure when the Global Hawk dropsonde data will appear in the live system. The rewritten recon system will not be released until 2013 and because the Global Hawk missions as part of the Hurricane and Severe Storm Sentinel (HS3) program end in early October, I might just wait until 2013 to add Global Hawk dropsonde data to the live system. Flight level Global Hawk data will not appear until likely 2013 in the archive and I'm still not sure if flight level data will be available in real time next September due to the large file size of the raw data that would need to be downloaded.

For the rest of the year and into early to mid 2013, the recon system will be undergoing updates that will reduce the stability of the system at times. This process will eventually make the recon system more efficient, consolidate separate recon systems for the Atlantic, East / Central Pacific and West Pacific, add new features and improve existing features. This upgrade will be similar to how the consolidation of the model system occurred. Current testing is being done offline.

Most every existing link to the previous recon system will change during this process. If you link to any part of our recon system, consider the link to be temporary. It is best to link to the main page of the recon system, http://tropicalatlantic.com/recon/, as this will probably be just about the only page that does not change.

The system is also undergoing this upgrade in order to be more portable in the future. When the upgrade is completed next year, the system will likely be less demanding as it has been, meaning the recon system might be able to appear on backup sites. Other aspects of the changes may make the system more demanding, but more reliable, if it was run on another site with the resources to use the new options planned. By mid 2013, the code of both the model system and recon system will likely be made available to the public.

Some undocumented new parts of the recon system may be available at times before they are officially released into beta testing. These features are highly experimental and will be marked as such in the web based part of the system.

If you have any suggestions for the new recon system, please contact us.

This afternoon 92L did not properly rename to 07L on our site. The reason was due to the NHC accidentally releasing the renumbering file with the wrong name. It was "invest_RENUMBER_al072012_al922012.ren" instead of "invest_RENUMBER_al922012_al072012.ren". The first part is what you are renaming from and the second part is what you are renaming to. As released, the file name said rename 07 to 92 which is why our site had a problem. Our site has been corrected so that it will bypass this error, and still properly renumber the file, from now on.

Usually our site's Twitter messages (tweets) are also posted to Facebook. As of August 1st something has changed on Facebook and tweets can no longer be posted automatically to Facebook. Twitter is aware of the problem and working on it. I will keep checking back to see when the functionality can be added back.

I am still working on various tropical projects. I got sidetracked from trying to figure out the NHC error cone in Google Earth and am working on something related that I need for historical error cones. What I am working on may actually become its own little side project. I need historical error cone sizes and since I can't find most of them I need to derive what they were as best I can from the NHC official forecast error database. I may also generate charts like the one seen here at the NHC.

Today, the model system stopped updating properly. I have tracked down the error and fixed it. The listing of the "tcweb" directory on the NHC's server is coming through slightly differently now on our site. Previously, the first day of the month came through as "01", the second day as "02", and so on for the first nine days of the month. Now it comes through as " 1" or " 2". Security checks in our system therefore rejected the data in the "tcweb" directory listing and exited the system. I think my host may have made a change as our backup site at HurricaneCity still worked fine.

I am still working on some new tropical products for next year and after that will continue work on rewriting the recon system. There is a lot of work to be done!

Our site is now running in normal season mode. The check for new model data for the Atlantic, East Pacific and Central Pacific occurs every half hour (:00, :30) which is actually the same as it was during the offseason. Recon data is now checked every 5 minutes and will update more frequently, if needed, when I notice recon scheduled for a storm. (In the off season it was every 10 minutes) The recon schedule: Atlantic (:02, :07, :12, :17, :22, :27, :32, :37, :42, :47, :52, :57) and then for East / Central Pacific (:03, :08, :13, :18, :23, :28, :33, :38, :43, :48, :53, :58). The Pacific recon system is running live on a trial basis. If resources are needed elsewhere by the site, it will be adjusted to every 10 minutes (:03, :13, :23, :33, :43, :53) or turned off until there is an actual storm that I notice with recon.

Tropical Atlantic now has a new official backup site at HurricaneCity.com. If our site ever fails for any reason, you can access Atlantic, East Pacific and Central Pacific model data there too. You can also check out an exclusive feature we added to their site, a best performing models map.

We are working on a lot of different tropical things this year for this year and next year as well. We'll have more on that in the coming months.

NHC model data on our site from 1851-1930, 1938, 1980-1981, 1988, 2004 (excluding invests), 2010 (excluding invests) and 2011 (excluding invests) has been recreated to reflect any updates in best track data. Any storms that had been updated in the NHC's ATCF archive were recreated on our site. The NHC specifically noted (site is often hard to access) changes from 1857-1930 and 2011 but since there were other files that had a newer date I decided to recreate those as well for various years.

When 91L came out my site was not checking the site where the model and best track data actually came out. Our system was configured to download ATCF files from NOAA's faster server at "ftp://ftp.tpc.ncep.noaa.gov/atcf" and instead the files were published only to "ftp://ftp.nhc.noaa.gov/atcf/", which had been an older location which our site was not checking. We are now checking "ftp://ftp.nhc.noaa.gov/atcf/" for now. Older model data will be added for 91L when it ends. Model error data for 91L is not accurate because it is missing older model data.

I have been doing some work on another project in April not related to this site. By the end of this month I will return briefly to doing some more code cleanup on this site then I will complete a short project for someone else. After that I will return to the project I was working on in April and sometime in May I will again return to the model system to do code cleanup and revisit the model error section that contains the extra info that is in alpha testing. I have yet to revisit that alpha feature yet so use that area with great care as there are very likely mistakes. After I do that and some code cleanup on the main model script I will then work on yet another project not related to this site before I finally have time to return to the recon system to continue where I left off at the end of last year. That will probably begin in June and will probably take much of the rest of the year.

The web based interface of the model system will continue to see some updates in April when I return to reviewing it. For now though, a lot of work is now being done on the main script that generates data for the model system. That file, which got up to 317KB, has now been separated into a dozen other files. The goal is to only load portions of the system that need to be loaded each time. Most of the time there are no storms and the vast majority of the system does not need to be loaded. I am also reviewing the code to clean it up a bit.

The new version of our model system is now live. Changes will continue to be made to the system in March. The only major change was to how error trend data for intensity error is calculated. The new "Extra Info Popup" in the model error section, currently off by default, probably contains some errors and is considered to be in alpha testing. It may be rewritten in the future. If you see any errors, please report them to us.

All of the model data has been recreated to reflect the minor modifications to the Saffir-Simpson Hurricane Wind Scale that become effective May 15, 2012.

Model and best track data from 1851 through 1988 has been reprocessed to reflect the minor modifications to the Saffir-Simpson Hurricane Wind Scale that become effective May 15, 2012. Our site will continue to reprocess data for the remaining years over the next week or so. The later years take a whole lot longer to process and upload.

The Saffir-Simpson Hurricane Wind Scale is changing slightly. ( http://www.nhc.noaa.gov/aboutsshws.php ) As a result, much of the data on our site must be reprocessed. The scale exists in at least some form in the files in the model system and in the recon system as well. The model system will probably be updated within the next month. The recon system will not have its data updated until much later this year when that system will see major changes to everything except the Google Earth part. I am still working on the update to the model system. I still have more to do with it. I have not even started working on code cleanup in the main script yet. That may come later however, after I release the layout update to the web based interface. (which includes code cleanup within that part of the system) The layout update is a lot more involved than I originally intended. (I still call it a layout update because that is the most visible difference, but it goes way beyond that.) It will probably be done sometime in March. (The update is being worked on offline so until I release it there is no change to the model system currently online.) I need to revisit the model error section which I completed a rough draft of. I may redesign how I do the info popups that will be part of the new update for the error page. The new method adds so much text to the page. I may work on a JavaScript solution, possibly, if the file size of the page would be significantly reduced. Otherwise, I have completed, for the most part, the rewrite of the web based part of the model system. I keep rewriting things here and there though. I also need to work on the new layout script for each basin that allow for a more customized look for each basin depending on what site you are on. (Eventually, I will have additional content on our East and Central Pacific sites, so the navigation on those sites will need to be different than on our Atlantic site within the model system.) I will also continue to go through and see what other improvements I can make here and there. Since I will be recreating all the model files now is the time to decide whether I want to make any changes to the file structure at all. Recreating all the files is a two day task so I don't want to do it and then find out I could have done something differently that would have worked better with the changes I am making. I have made some small changes to the main model script that already will have an impact on some of the generated files. These changes I obviously want to test better before I recreate all the model data. I don't think the changes will be a problem even if I didn't recreate all the data, but anytime I make changes I have to be cautious for unintended side effects.

In the last week I have been working a lot on improving the code that runs the page that shows the Google Maps model plots. That page will load better once the updates are applied. The code has been improved as well as having been significantly minified.

I am still working on the model system. I am changing how the error trend works for model intensity error. I am going to revert back to the old way I did it. I will update the product description when I eventually release the update to the model system within a few weeks.

Work still continues on the model system. While I am updating the layout of the model system to match our site's new layout I am also making various other little improvements here and there along with code cleanup to the code that runs the web based interface of the system. The changes are also going to be a bit more involved when it comes to the model error feature. While it will still work the same, there will be a new feature available so that when you hover your mouse over a cell's data information will popup that further explains what you are looking at in that cell. This feature is very involved and as a result will be in alpha testing when it is eventually released. It may be off by default, perhaps only at first, though because there is so much additional text that has to be loaded to the page I might not have it on by default. However, it goes a very long way to explaining how in the world the model error feature actually works by giving you information that is completely specific to the cell you are looking at.

It will likely still be at least several weeks before the model system is updated to reflect the new web standards compliant layout and various updated features. After that I will take a bit of a break and then return to working on code cleanup for the main script of the model system that generates all the data. I also want to fix a long standing issue where on rare occasions where a small bit of old data gets left behind. (an administrative thing I only need to fix before the system is made more widely available to media organizations) Next I will be working on another project for someone else for a short time before I then return to the recon system which will be worked on for most of the rest of the year as that system gets a major overhaul. (though again, the Google Earth part will not see any changes)

Work continues on updating the layout of the model system to match our site's new layout. While there are no new features at this time and everything looks nearly the same as it did, there are significant changes to the HTML coding of the pages. Additionally, there is a lot of cleanup being done to the code that runs the system. This process will probably take at least a few more weeks before the changes are ready to go live.

Since our model system is more efficient than our old one, which no longer operates, the system has been running every half hour during the offseason. This means that data for 90L is coming through as it would if it were during the regular season.

Meanwhile, work on our new layout continues. I took the last week, and will take several more days, to work on creating something not related to this site. I am creating something that web developers can use for a site layout when building CSS based websites. Rather than use table tags for layout I am now starting to use pure CSS. I am creating a little generator that web developers can use when building their site to do that. That feature is something that will help me in the future as well. It will help finish the site layout for the model system. There are no changes planned for the model system right now, just updating it so that the layout matches the rest of our site. I may add some graphical charting of various data in the future however, but that would be later in the year as I want to redo the recon system first.

Our site has unveiled a new layout. The new layout has been designed so that it can work in Internet Explorer (IE) 7 and above. I'm not sure what it will look like in previous versions. Older versions have serious security risks (Even Microsoft wants you to stop using IE6 and earlier), so I didn't worry about making the site work in IE6 or older. Special code helps to make the new layout work in IE7. Some of the new features that will be coming to the recon system later this year will only work in IE9. (More powerful Google Maps mapping for individually decoded messages) The bad thing about that is that IE9 cannot be installed on Windows XP and earlier, which a lot of people have. The good news, the new features should work in other browsers regardless of what operating system you have. (Hint, hint... Firefox, Google Chrome, Safari, etc.) If your browser can't use the full power of the new features, it still allows you to use the product without the new features that would not work but does still have some new features that would work. The recon system will be something that I will get back to in February. It will likely be March or later before I get an idea of whether I can bring the live system efficiently enough into Google Maps to make it possible for us to offer without it crashing our site.

First, more changes will be coming as our site's model system will also be getting the new site layout, likely in the next week or two. There will also be some other changes behind the scenes to that system, though there will be no new features. Some old working files get left on rare occasions and I want to try to write something to fix that. I also want to do some code cleanup. The updates to the system will likely add another configuration file for each basin so that the site layout can be more easily managed across basins, one each for the Atlantic, East Pacific, and Central Pacific. Eventually, this script will be available again to media organizations. I just need to make it easier to work with. That was the goal but the new system is becoming as sprawling as the recon system, though thankfully the model system, unlike the recon system, was built to be as manageable as it can be and for the features it has, it's not impossible for media organizations to customize.

Today I retired our old logo which had been around since June 2006 when our site first started.


The old logo still appears in places but it will eventually be replaced everywhere. The new logo appears below with some of the logos for some of our other current and future sites.


The site is probably going to get more and more unorganized over the next few months before it starts to get better. Throughout 2012 the site will undergo a complete makeover, along with many features that will be redesigned. I hope to have a lot of it done in six months and then start to develop and test new features in the last half of the year. But, that's a long way away yet. Testing continues to go well on the redesign to the existing Google Maps areas of the recon system. It is still not yet available to the public however and that might still be a long time away. The recon system is going to undergo a massive redesign. While the Google Earth mapping will remain the same, the changes to the other areas will be considerable. A lot of it will be behind the scenes as I attempt to consolidate the recon system so that it is more portable to other sites that are approved to use it. I also have some plans to make it more efficient. I might hold off on releasing any changes to it until I am ready to replace it with the new version.

Behind the scenes a lot of work is under way on revising some of our products. To start off 2012 I will be redesigning the recon system. To what extent I don't know yet, but work has been under way for many months. The recon system has always needed a redesign. It also needs to be centralized enough that it can be easily adapted to other sites that I allow to use it. Not only for other sites, but for our own sites as well, like the East/Central Pacific and West Pacific recon systems at Tropical Globe. The only enhancement I know will be released for sure is the Google Maps mapping that is available for a single decoded message. It might be awhile before that gets released, but the enhancement does work. Unfortunately it will only work in newer browsers because the technology used in part of the new mapping system uses the <canvas> element in HTML5. For Internet Explorer users, you must have version 9 or higher. For those running Windows XP, which does not run IE9, you will not be able to use the new features unless you use another browser. I must create an alternate version, like the really basic version that exists now in our Google Maps system, so that older browsers can show the mapping. Basically, I am trying to extend some of the features available in Google Earth or the Google Earth Plugin to Google Maps. And eventually, though I do not know if it is feasible yet, I hope to be able to have the live recon system run in Google Maps. I will likely not know for months if that will be feasible. That will take many months to try to develop as there is a lot to consider when it comes to trying to do that. It is important to note that the Google Earth recon system will probably not be undergoing any changes. After that, it is on to trying to develop a system to handle NHC advisories, which I have had nearly done for five years. It's all about trying to get the NHC cone to work right. And looking even more into the future, like years into the future, we just bought a couple new domains: Tropical West Pacific (simply links to our West Pacific recon decoder, available only when we are notified of recon) and another for the Indian Ocean (it will be at least years before this site has a chance of being developed). So that is a preview of what is in development for 2012 and beyond. Have a suggestion? Contact us!