 to you guys about a software I'm working on for visualizing and analyzing data in the cloud. For those of you on your computers, go ahead and go to that website and play with some of the features of the software while I'm talking about it. Or for those of you at home, we didn't want to make the trip up here in the fog and snow. So let's talk a little bit about how we typically work with data scientists. Usually there's a lot of data out there sitting in the cloud and these data servers and the way we interact with that is downloading it. If you think of the cloud as an actual cloud, it's kind of like a very tremendous thunderstorm, precipitating gigabytes, terabytes, of data all at you at the same time. For some occasions it's great, others it's not so efficient. The benefit of doing a cloud-based solution like mysoft or other gist-based solutions is that instead of it does a lot of that work behind the scenes. And so to the user, all you're really getting rained on pretty much are image files have already been generated and so your files size transfer is only very small. This is of course useful in situations where bandwidth is a precious commodity like field campaigns, conferences, when you're sharing Wi-Fi with 17,000 other people, or outreach or international collaboration efforts. So let's take a look at what my software can actually do starting with a very basic event. We're just going to look at different seasons and different parts of the globe. Here's a temperature plot of Northern Hemisphere winter, service temperatures for the month of January. If you were to go in, you can actually do some dynamic overlays. So let's do a simple subtraction of summer versus winter from Northern Hemisphere. We'll do that now. This is all done in real time. It's not done ahead of time or preprocessed at all. And so if you can think of any sort of difference you want to look at, you can visualize it here all in the cloud. So of course we know that it's warmer in the summer in Northern Hemisphere and colder in the winter. So we'll overlay on top of that. There's solar radiation. This is just downward solar flux. You can see that there's some correlation there where the solar flux is positive or positive or warmer or more abundant during the July. It's warmer and vice versa. In order to look at some more in detail, we're going to actually use some interactive tools. Some of these tools are in addition to the basic plotting features. They're usually manifest as simple elements like markers, lines, and boxes. Now we're just going to do a simple zonal mean overlay of the entire hemisphere looking at the differences between July and January. You can see that the curves kind of line up a little bit, but there are some definite differences here, especially in the Southern Hemisphere because of course of all the oceans and all the activity down there. Once you create a figure like this, it's actually, it can bring it up as many times as you want. It's all summarized in the left bar there. And so we'll create another figure. So this is a vertical profile of the same element. And if you can go back and forth, whatever, there's a lot of different types of visualizations you can do all within the same dynamic interface. At the same time, we were looking at model data. We can also do more, say, observational data that might not be continuous. Here's an example of such. It's the same plop but with oceanic ship data from the ICoS database. You can see here that it's not necessarily continuous, but it still gets a point across. Or we can do some more complex things like looking at other planets, for instance. Here's an example of the same plop over a label for Mars. Obviously it looks a lot different than it does on Earth. With the thin atmosphere, it looks, the temporal variation that we're dominated by a diurnal cycle rather than a seasonal cycle like they are here. We're even looking at other model outputs, say, we'll say we're just a regional model. This is just an example of worth output over the Middle East, for instance. Or if you really want to get really high resolution, it's also potentially possible. The next slide is going to show the high resolution rapid refresh over the front range, just showing the temperature field. You can see that all the different minute features of the data over land top of the satellite imagery in order to really see where things are colder or warmer over the course of an evening. Because everything is done in real time as opposed to ahead of time, it's actually a very powerful tool for modeling in comparison. You don't have to individual create files for each model. This is an example of just a random CFS climate forecast system run versus the reanalysis data. You can easily visualize differences between two models without having to do anything ahead of time. The idea of the software is to put it in a format that is useful for the users in their context. Language localization is something that I'm working on a lot these days. Here's the entire interface in Spanish. Also, too, looking at cross-platform development. I'm a second ahead of myself. The idea is to be able to pull up the same imagery, whether looking on a cell phone or whether you're looking at a desktop computer. The developers, of course, haven't been left hanging. There are also a lot of flexibility ahead of myself again. A lot of flexibility in the APIs you work with. Here's an example using open letters as the mapping interface as opposed to Google Maps. There's also an effort to produce in multiple data languages as well. Despite all these features, actually, the software is not very heavy. It's actually pretty lightweight and ahead of myself again. Easy to deploy. All it requires is a simple lamp stack, very standard, and then the grads application. Actually, for those of you who were playing with the website while talking about it, it wasn't actually running on a data server. It was actually running on one of these guys here, Raspberry Pi. It really does not take much computing power to be able to put something like this together. With that, I'd like to thank you all for your attention. Here's a full project you can play with that have the full set of features for very specific topics. Thank you.