Tuesday, September 25, 2012

GOES Down! GOES Down!

GOES-13, also known as GOES-East went into a state of hardware failure a couple days ago. GOES-14 which has been on standby has come online and taken it's place as GOES-East.

A couple weeks ago I rolled out the replacement satellite/radar pages, touting that visible satellite loops would now remove a healthy chunk of "night-time" images. Following that I realized that a handful of west coast regions for 1km, 2km and 6km resolution were not properly removing night-time images.

This removal of night-time images was accomplished by taking advantage of the fact that when our images get generated, the file size directly correlates to the average brightness of the image. This means that with visible satellite images, there should be a very natural drop-off in file size for night-time images. Within the code that grabs the most recent set of images, I simply added a file size check for visible images and filtered out the images that fell below a certain file size. This effectively removed night-time images from mix without using any information on sunrise, sunset, time of year, or latitude. I was quite happy with that approach. It meant I didn't have to over think the scripting or tailor the code to each region. Basically; less work, more reward.

Getting back to those west coast regions not removing night-time images; It turns out that GOES-West (GOES-15) changes how it operates when taking images of the dark side of the Earth. It attempts to bring in as much light as possible, which gradually increases the brightness of an otherwise dark image. This results in the minimum file size occurring right around sunrise and sunset, with a relative maximum in file size overnight. Here are some highly sophisticated graphs to help further describe this situation.

Images from GOES-13 would have created a File Size vs. Time graph similar to the one above.

GOES-15 on the other hand creates a graph similar to the one below.

Because of this, I can't set a lower threshold for file size to remove night-time images because it would also be removing valid day-time images.

Initially, I was willing to just let it be since the overwhelming majority of the sectors we provide are covered by GOES-East. But now GOES-East is GOES-14 and wouldn't you know it... GOES-14 operates the same way GOES-15 does. Both satellites now variably adjust the amount of light coming in to help grab as much valid data as possible from the visible channel. But this means that when our images are generated... well, hopefully you get the idea by now.

So... for what it's worth, I got 2 weeks out of a pretty neat feature for our visible satellite loops. But at the moment, I'm not interested in getting excessively cute with the script that grabs the most recent set of images for a loop and make it remove night-time images. I will also not write it in some way that involves me editing the code constantly to adjust for variances in time of year and latitude. Waste of time.

When other, larger tasks are taken care of, I may return to working on the scripts for these pages and try to get clever. For now, "Sorry folks. Can't do it."

Monday, September 24, 2012

Short post...

I'm going to attempt to update this blog at least on a weekly basis. Tuesdays will likely be the day. So since today is not tuesday, this will only be a short entry.

I've held off on the idea of allowing mobile users to switch back to FLANIS based satellite loops. Yes, for those who have devices that aren't properly loading the overlays, JSAni is not very pleasant to deal with. But neither is making the necessary changes to the pages to allow a user-friendly re-route to the FLANIS pages and back again. This would only be a temporary fix anyways. Instead I've been in contact with Bill Bellon about the issue, and while he is busy, the issue is on his to-do list. Until either a more definitive dead-end or conclusive fix is reached, I will leave the pages as is. I apologize for the inconvenience, however there is too much else at the moment that needs attention. I never like leaving a product semi-broken. But we have limited time and a lot that needs to be worked on.

Monday, September 17, 2012

Dual Pol Sites and Tiny Fixes

This should be a week to get a lot done... a lot of little things.

The current mode of operation in my head is to take care of a few scattered issues with a few of the projects that have been worked on recently. Notably this means issues with mobile satellite products, a few numerical model quirks, and a few random other items. These little tasks need to be knocked out so that all of our attention can be put toward Surface and Upper Air products.

One of the bigger, "little things" I've taken care of today is allowing several sites to display Dual Pol data that have recently been upgraded. These dual pol upgrades tend to happen for only 2 - 4 sites at a time, often take a week or two go into effect, and then roughly another week before our scripts start properly creating the images. Because of this I have the php scripts on the web end restrict access to these products until we know the upgrade has taken effect. But it has been several weeks since I've been able to peruse our sites to find out what sites are now ready with dual pol data. Turns out many have come online recently, making me look like a slouch.

Here is a complete list of recently added Dual Pol sites; KRGX, KLRX, KNKX, KDAX, KMAX, KGGW, KHDX, KEPZ, KABR, KDLH, KMPX, KSRX, KLZK, KLTX, KGYX, KCBW, KTYX, KCCX, KILN, KFTG, and KDMX. A total of 21 sites.

If you ever catch a site that has recently received the Dual-Pol upgrade but we aren't allowing access to the products, let us know using our Feedback page. Never hurts to give us a heads up. I often have a dozen other things on my mind that might be distracting me from making that small change.

Other tiny fixes; I'm going to finally work up a way for mobile users to switch between flanis/janis loops. At the very least for ones that use overlays. Which basically means satellite and radar products (excluding NEXRAD data). I'll be fixing some broken links in our Links page.

Then, as mentioned, we start major work on analysis products.

A note about our staff: It may not be readily apparent, but our staff consists two full-time professors, two official part-time staffers, a (contracted? fuzzy on the details) part-time remote staff member, and a couple IT consultants and administrators. Of this small staff, only myself has been tasked with the primary priority of designing and maintaining the website. The rest of the staff have other priorities that at times take precedence over product maintenance and development. Faculty needs to teach. Only one of our system administrators is technically paid, the other graciously offers his time and help (sparingly, of course) for free. Needless to say, we all have our plates very full. But we all remain considerably excited about this website. Since our site offers access to vast amounts of meteorological data for free, where others charge the user, our work feels very rewarding. We do this because most of us are forecasters at heart and we enjoy having fun tools at our disposal. We also firmly believe that giving students access to this data is a critical part of their education.

Long story short, we all scramble like crazy to do work quickly and do it well to maintain a unique, multi-faceted, undergraduate meteorology program along with a one-of-a-kind meteorological data and academic website. It's a tall order. But we do it every day.

With that said... sometimes... an occasional burn-out is to be expected. ;)

That's why I attempt to keep communication with users open. That way we give you, for a moment, the opportunity to be part of our staff. Your feedback is always a welcome contribution.

Tuesday, September 11, 2012

Mobile Madness

In a slight continuation of the previous post, I wanted to bring up an issue that has been brought to my attention recently. Since Flash support is steadily dwindling in the web world, and was never truly there when it comes to the mobile web world; I ... We made the choice to implement JSAni based loops for all mobile devices. Since JSAni is based in Javascript, it's highly unlikely that virtually any web browser (mobile or desktop) could not load a loop, and load it properly.

It turns out that second assumption was a bit too much.

Apparently Android OS version 2.3.x (aka Gingerbread) will not properly load overlay files from JSAni. What this means is that our satellite loops will not display ANY overlays... including the all so critical "Map" overlay. This problem is not a simple one. It's something fundamental about how all browsers on Android 2.3.x interpret the javascript loader in JSAni. Since these devices still allow the flash plugin that is required to run FLANIS on Android web browsers, in the coming days I am going to implement a way for mobile users to switch back to the desktop/flanis version of the satellite products. It may be a bit more frustrating to use on mobile, but it will work.

It's upsetting because JSAni is much more stable when it comes to loading a set of images and is less likely to crash. It's also easily customizable so that the interface is easier to use on mobile. And to top it all off, it works just fine on newer versions of the Android OS and iOS.

It may all be a mute point in the end, because eventually Bill and Tom at U of Wis.-Madison will be working on HANIS, an HTML5 based replacement for FLANIS that will hopefully combine both the functionality and customization of both JSAni and FLANIS. On that note, I want to say that these tools make running a met site unimaginably easier. A common saying in the NEXLAB world is "It is nothing without the loop!". Having these versatile animation tools is critical.

Well that's all I had to say on that matter. Other things going on; my work on the site is quickly moving from 'one big project at a time' to 'many small projects all over the place'. I'm attempting to implement a few site-wide design changes (subtle changes), but also have to divert some attention to some class related web work. These students need a new satellite lab.

I should say that the one big project on the horizon (which was hinted at in the previous post) is major re-structuring and re-development of our Surface and Upper Air Analysis page. We have a lot of new products to offer, as well as vastly improved existing products. Some of the changes to existing products should go into place this week. But my part of the job, improving the web pages for these products, is perhaps a few days to a week or two off. We have a lot of the new images worked out; right now we are still in the brainstorming stage of "What can we change to make this data easier and more fun to access?". With that, I end this post an open the discussion to you...

... I say that, but I know I have virtually NO readers at this point. But that's okay. ;)

Monday, September 10, 2012

Upcoming NEXLAB Work

First official post on this page and there is an awful lot to discuss. Lets get started.

A couple weeks ago I finally got a bug up my ass to do something about the growing need for an alternative to FLANIS loops for mobile users on NEXLAB. Actually, the truth is I'm an Android user. I own a Droid 4, and until my phone pulled down "Ice Cream Sandwich" (aka Android OS ver. 4.0) I was willing to dodge the issue with the remark "Well it works for Android...". Turns out I was completely un-aware that Bill Bellon out at University of Wisconsin-Madison had worked up a Javascript-based counterpart to Flanis called JSAni, which I've come to refer to as 'Janis'. It just rolls of the tongue better. Anyways, I had already discovered a method with which to detect mobile users on our site; So with the advent of JSAni I decided to move full steam ahead on creating functional mobile counterparts to many of our more frequently used "loop-able" products. e.g. Satellite, Radar and Models.

When it comes to our Satellite and Composite Radar products, the current desktop version of those products have really been gathering dust. They don't always function correctly, the menus aren't as good as they can be, the navigational ability is practically non-existant.... the list of things to fix with them is rather long, lets just say that. So I opted to move toward a complete redesign with the new mobile version of these pages that would essentially be a test bed for the desktop version. Any of you readers who are also mobile users of our site (NEXLAB) hopefully have already noticed those new pages.

Well what I'm working on now is bringing what I've done for the mobile version of these products to the desktop version, while at the same time now tailoring the functionality of the page ENTIRELY toward the desktop user. Since I now have the products operate with two distinct mobile and desktop counterparts, I can get away with implementing nicer features. Not to mention the fact that I am constantly learning new tricks. On a side-note, that's almost one of my downfalls as a web-designer, because as I work on one project I learn about 3-4 new techniques that are immediately applicable to project I just finished work on. But I digress.

What I'd really like to get to now is something I hope to do as much as I possibly can on this site, which is to offer the curious (which you must be if you found yourself reading this) a glimpse of what I'm working on. If you haven't had a sense of it yet, I'll just come right out and tell you; I very much wish to keep the lines of communication between User and Designer open. I enjoy any feedback I can get and I like to give everyone the best idea of what to expect from what is being worked on behind-the-scenes. So with that said, here is our new 2km Satellite page...


As of (9/8/2012), this page is quite close to completion, but work is still being done on it. What's new?
  • Congealed menus with similar organization to changes made to our Upper Air Sounding and NEXRAD data (Product Menu/Sector Map)
  • Menus BEHAVE much more as one would expect. They open by clicking the tabs and close by either clicking the tab again, mousing out of the menu for a period of time, or clicking virtually anywhere other than the menu itself.
  • Menu will NOT close instantly, and will remain open if the mouse re-enters the menu.
  • The hard-set 10 min page refresh has been replaced with a USER-DEFINED Auto-Refresh feature (found in the Product Menu). Can be turned OFF, or opened up to a 30 min refresh time. Defaults to our 10 min standard.
  • Better image handling, especially with loops, which means there should be fewer instances where server/data problems result in this page breaking for extended periods of time.
  • Improved aesthetic look at feel. I'm really trying to wipe out THAT blue color. It was a poor, hastily made decision on my part made very early on, and across to much of the site to be quickly dealt with.

Apart from this, the rest of our staff has been hard at work on some MUCH needed improvements to many of our Surface and Upper Air Analysis products. I can't get into that too much. It's still in it's infancy at this point. But I can tell you we are very excited about it, so I'll save that for a future entry.

Thank you reader, and welcome to my inner sanctum of geekery.