Site Updates

July 27th, 2020

Euro model data, from ECMWF, will not appear on our site again until we make some updates to that part of our model system. Normally that tropical cyclone model data is automatically incorporated into our model and best track system. They made some changes at the beginning of July to add wind radii data for model forecast points. We need to adapt our system to be able to decode that data. While the changes needed to get the Euro data working again is simple, simply adding two text files we get from them, and we could have waited on the wind radii data, we also found an issue on our end. To determine what basin a storm is in, before we download the file from ECMWF, we use the coordinates in the filename to determine if we should download the data for a particular basin. We then previously placed the file based on the basin our system determined from the coordinates in the file name. But this became an issue when Hanna crossed into Mexico and was about halfway between the Atlantic and East Pacific. Rather than use the basin we try to determine from coordinates to place the data into our system, we need to use the basin found in the file we download to place the data. While that is an easy fix, we are also reviewing the code in the system. Eventually all of these systems will be released publicly, perhaps in a year or two, so we want to make it ready for that. We will also adapt the code to handle the new wind radii released.

In the previous update I mentioned that sorting in the model error system wasn't yet available. I forgot, but I had already added that into the system. It is the last option among all the options on the model error page. Eventually I will likely add it to the column headings themselves, with arrows, but maybe I will do that when I convert that system to be mobile friendly.

July 25th, 2020

Someone asked about whether we will eventually allow sorting of model error in the model system. Eventually I plan to have that option, but it could be a year or two. I decided to write a little about some of the current plans for the next few years. This site is constantly being worked on. Every day of the year, progress is being made, though sometimes you may not see the results of that progress for a long time as the features are worked on offline and not implemented until they are completed and well tested.

The current tasks, somewhat in order, involve switching all pages other than the model and recon systems to a mobile friendly layout. A few pages have already been updated. I need to verify that all the links still work on the site. That may seem easy, but when you add the radar display into that too, there are a lot of links. Add option to view desktop view on mobile, which would be helpful for tables that might have to be broken apart in mobile view in some areas of the model and recon system to display data vertically rather than horizontally. It would become a list of data rather than a table on mobile, though I haven't got around to deciding if that is what I want. I might just let it be scrolled horizontally. The next model and recon system tasks may not be released until they are all completed. Changing diagrams from Flash to JavaScript. (see previous site update for example of that project) Rewrite center fix map for ArcGIS. Switch model and recon systems to mobile friendly layout. Make model system work in real time for all storms, globally. Add all historical best track and model data for all storms worldwide. Worldwide data in real time would be subject to availability from a NOAA server that may not always work. NRL has the data too, but their site restricts the public from viewing that same data I need. Add sorting options to model error. Looking more into the future, switch model map to ArcGIS, and maybe CesiumJS, and add satellite imagery option back using data from SSEC's RealEarth™. Also allow 3-D display in that and center fix map. Replace simple bar charts for HDOB messages in the recon system to the same kind of charting that will be deployed in the model system. Add additional charting options in recon system for data, such as for an entire mission. Eventually rewrite the system that had NOAA P-3 radar imagery. I don't know if I'll ever get that data in a format I can use again, but I want to have the old data in a better display, like the one I have for Bahamas radar imagery for Dorian. Improve "Model Viewer (experimental)" and make it as real time as possible, rather than something I have to update once each year. Add additional charting to the model system, such as charts that show various statistics about the best track data over the life of the storm, such as wind speed, rather than only a simple text table. Maybe also have charting that charts various data from all the storms in a single season. Improve live recon display. Add ability to save options in that display and also maybe model map display.

The current task I am working on is creating a display for four historical storms I saved radar imagery for in the "Historical Data" section of the site. It will have a panel that can be dragged around the screen, that allows you to slide through the imagery better. That display could be used in the future too for other storms I sometimes mirror radar imagery for. I will also add that draggable panel ability to other displays eventually.

Other things to work on at times, improving automatically selected satellite imagery used in the customizable NASA satellite display. Add more radar sites into the global radar database map. (Italy has been done since sometime last year, but I never did organize the data that goes into each popup so I haven't released it yet) That display is a lot of work because I have to research the exact coordinates for every single radar site and verify it in satellite imagery. (sometimes it might take many hours to find one radar site) Then I have to determine the range of publicly available imagery which can also take a long time. Improve options on that radar map, such as exporting all sites to a KML file that can be viewed in Google Earth. Maybe a table view of all the sites and an option to export a CSV file of the sites that could be viewed in a spreadsheet program, like Microsoft Excel. Review and annotate lines of code in the model and recon system for eventual public release.

I have had two people email me about where model imagery is now located on our site. The link at the top of some of the pages on our site has changed. I have been updating the site, making it more mobile friendly, which is why the model data link is a little harder to find on the pages that have been updated. The links to active storms at the top of the page now link to a storm page for an active system rather than a direct link to the model page like it used to. (older pages still have the direct link to model data) Google charges for map usage after a certain amount. Rather than make a link on every page go directly to the map, and use a map load, I wanted to add the extra step to avoid unnecessary clicks. To be fair, Google has never charged me. Every time I have neared the usage limits they have granted me credits that have allowed me not to be charged. Even so, eventually, maybe next year, I will switch the mapping to ArcGIS for that model page. Their usage is free and at that time I might add the direct link back. Since other sites also use our systems on their site, and eventually the scripts that run the model and recon systems will be made available to the public again for additional sites to use, I want something that others can use freely too. Right now I have a quota on how many times Google Maps can be used on our site per day. I have to watch that constantly during the season and adjust that quota as needed. Google only allows you to set a daily quota. You are charged if you go over a certain amount per month. You can have about 28,000 free desktop map loads each month. (which is $200 in free usage they give you each month) Map views on mobile are free and not counted in those map loads. While that sounds like a lot, before we switched some mapping to ArcGIS, such as on the front page of our model and recon systems, we did exceed those limits during busy storms, sometimes in only several days. To not be charged, you have to set your daily quota to 903 map loads per day. (28,000 / 31) Since for a busy storm we can easily be over that, we have to adjust that quota sometimes to be higher for a few days. But that then means that we have to make sure we watch the quota as the month progresses so that we don't go over it before the month is over. It's annoying to do. But again, Google has worked with us multiple times in the past during instances where we simply needed more credits. I just can't rely on always being able to contact them for it. Once you go over, you have to pay, and I don't want to have to do that since this site generates no revenue. (This site is owned by an LLC that usually doesn't make any money, but we are not registered as a nonprofit so we don't qualify for that status at Google)

I should also note that we will likely be using social media less. We often simply retweet tweets on Twitter, and share posts on Facebook, from the National Hurricane Center. We will continue to use Twitter for developed storms, though we will likely post less often. We may not use Facebook too much in the future unless a system poses a more significant threat. Due to how Facebook displays images on a page, allowing old imagery to appear just after more updated content, we don't post imagery to Facebook any longer. I didn't like having content that we may have created months ago appear between content that is more recent. I would rather not have images or videos posted among current posts and there is no option to turn that off. Therefore, any content that we specifically create, whether an image or video, such as of radar imagery for example, will only be shared on Twitter. We did this last year too, which is when we deleted all of the previous image and videos we had posted ourselves to Facebook. While Twitter does show the latest images posted, it does so less prominently, in a way that makes it clearer that the content may not be current.

July 16th, 2020

In late June I completed an early version of the JavaScript charting (using C3.js) that will eventually replace the Flash charts in the model system. (example) This is likely not going to be available until late in the year. It also works for AXBT water depth in the recon system. I have not worked on one to replace sonde charting yet in the recon system, though I may use this one. (default example)

Yesterday I started changing a few of the pages on our site to be mobile friendly. (our site update page for example) That process will be a long one. All pages, other than those associated with the model and recon systems, will be converted first. It might not be until late in the year, or next year, before the model and recon systems are available in a more mobile friendly format.

Older Site Updates:

2019 | 2018 | 2017 | 2016 | 2015 | 2014 | 2013 | 2012 | 2011 | 2010 | 2009 | 2008 | 2007