 Before we move on further, I just want to remind you again that if this seems overwhelming, I highly recommend going to the Phosphor G Academy. They have five free courses of which the Introduction to Remote Sensing would be a great place for you to get a lot of background with us. There's free lectures, free videos. The Introduction to Geospatial Technology will get you started if you're brand new at this. They have labs and lectures that you can access online. The most important of these would be the Understanding, Remote Sensing and Aerophotography module, which is here. And it walks you through the basic principles and understanding the quality of images, what types of error that you might have, and so on. So again, I highly recommend these resources as a good place to start if you're feeling a little overwhelmed by it all. As well, I should mention that while the commercial services such as SecureWatch can be very important for those who have the budget to use it, there are open data programs through businesses such as Maxar. And Maxar is consolidating many data products from many satellites together and gets that 30 centimeter resolution. So here's their open data program, which is mostly for disaster response. And by disaster, they mean not political, typically, but natural disasters, although they do have all parts of the world. So you see here, Papa Indonesia, South Sudan, you can go in and you get excellent 30 meter resolution imagery of an entire location that's impacted by a natural disaster. So that's an opportunity for you to take advantage if you are monitoring a situation that involves a natural disaster. For any other type of humanitarian disaster, whether political or natural disaster, there is something called the hot OSM tool. The hot OSM tool is based at export.hotosm.org. This is the OpenStreetMap tool that allows you to take out any vector information, vectors, things like roads, houses, buildings, locations for getting people together, allows you to take all that data out and put it into the GIS. So again, this is, we've moved beyond just having a website to show stuff, to having our inability to export things. Now they've got many different tutorials on how to use this. You need an account with OpenStreetMap, which will then allow you into the hot OSM export tool. You can export in many different file formats. Again, KML allows you to look at it in Google Earth if you don't have other specialized GIS. Garmin would allow you to put it into different sorts of GPS units and analyze that in different software. Shapefile and GeoPackage are probably the most common for GIS programs. So Shapefile comes from Esri, and the GeoPackage is a newer, open, licensed version of a vector file that allows you to bring all sorts of data in, like as I've mentioned, buildings, commercial areas, communications, education, emergency finance, healthcare, humanitarian, so on. So this is an opportunity not necessarily to monitor a situation, but to get information about what's there. As well, there is an entire group of people through hot OSM that creates information based off of emergencies. So basically, if there is an emergency somewhere, say Nepal, when they had earthquakes and a lot of buildings falling down, people will come together to use remote-sensed data, whether that comes from being or other locations, to actually digitize things like roads. So this is a great place to get that type of information. I'm focusing a bit more on the imagery analysis in this workshop, but I just wanted to mention this in case that you're actually looking at it. So I did run an actual export from Basant Char, and I exported it as Google Earth as a Shapefile and as a GeoPackage. And if you export one region, so here I've selected an area of interest, and it just identifies Basant Char as a floating island, but if you export one region, having multiple data formats can be good, as several websites and software platforms use the data differently, so it's nice to have the data that can go multiple ways. Those are vector sources, but in order to get access to the largest database of different sorts of imagery types, you can always use USGS Earth Explorer. So USGS Earth Explorer is just earthexplorer.usgs.gov. You need to set up an account. Once you've set up an account, you can search anywhere in the world to get imagery. There's always the option of going somewhere in the world and then zooming in so that you can get to where you want to be. This satellite image, we're lucky in that it shows our area of interest. If you have a KML or Shapefile, then you can just upload that and keep that consistent across platforms. One way of creating a KML or Shapefile is through a website called GeoJSON.io. So I'll just show you that quickly before we get into USGS Earth Explorer. So here we have basically a website that allows us to produce automatically a GeoJSON. And you see again here we have some issues with our area of interest not being shown. So here it is on Earth Explorer and here it is in geogson.org. This is largely to do with the background map, the vectorized map that does not include our area of interest. So again, it might be as easy for you, it might be much easier for you to just use, say, zoom into a place. Here you see two different, actually three possibly different satellite images being overlapped, which is why the island looks different. It might be to zoom in a place and use the map. It selects the coordinates for that map. Or it might be easier for you to actually use a manual adjustment. So here I'm placing dots using my left clicker on my mouse to say this is my area of interest and you see how that populated the Latin loan. Or it might be easier for you to actually upload a KML or Shapefile. This also can be useful, the geocoder, but a lot of the places that we deal with in humanitarian and other types of complex political emergencies, we actually don't have good geocoding in those areas. So if you needed to do a KML or KMZ, say you had one from elsewhere, you could try to make one here. Again, not great because we don't have a perfect overlap of our, we don't actually have any representation of our island. So let's just click here. So I'll go ahead and make a straw polygon, click and drag a rectangle. And we'll hope that that covers the whole area. So that would be save as KML or save as Shapefile or save as GeoJSON, depending on other programs. But KML and Shapefile are the most common in a lot of circumstances. And then you would upload that here, and that would show this area. Again, I mostly rely on the actual satellite imagery to mainly outline an area. But for consistency's sake, you might want to be doing the import of Shapefile or KML. So once I've done that, then I go into the datasets. Now Landsat 8 is great, gives 30-mean resolution. It has all the bands that we need to find out a bunch of information. However, Sentinel is better in terms of resolution. So we'll look, you just type into the dataset search Sentinel. You'll get Sentinel-2. Let's go ahead and select that. Press OK. This is pulling basically the same information that you would get through the European Space Agency in terms of Sentinel-2. It does not pull all of the other Sentinel products. But again, we're not that interested in those other ones for our purposes. It's looking, it will look for any place that's within the area of interest that Sentinel has overlap. So if your area of interest contains too much space, you should probably adjust it so that you're making sure you're not capturing other tiles. The problem with capturing other tiles is that you will suddenly have multiple large images to download rather than one. So it's good to get your space down to exactly where it needs to be. There is also the issue that this island, for example, has changed over time. So you may want to give a buffer. But these are all things that you need to think about in terms of your own research and what you know about the site itself. So we've got Sentinel-2 selected. We'll look at the additional criteria. Let's say we want less than 20% cloud cover. You can make many other adjustments here. We'll look for all Sentinel products and we'll just go to the results. Oh, I should have set a date. I should have set a date really quick. Let's put it December 1st until today. So there we have our dates. Let's look at our results. So we have an image. This is captured. Our location has less than the amount of cloud coverage that we specified. You may get more images if you adjust your cloud coverage. So let's say that all our results. Now we have three images. Again, you need to be keenly aware of whether your location is actually 100% covered. You see this image here, I'll pump it into the map. It is basically by this show browse overlay. So we've got the footprint. Here's the show browse overlay. See what we get there. It's actually clouded. It's not showing up, but it's too cloudy, as you can see from here. So you see the cloud coverage within the metadata. If you actually click on the image, that shows the datum, the map projects, and the cloud covers 88%. So that's pretty useless for us. This one, hopefully it'll visualize. Visualizing, but this one, it actually doesn't cover quite the area that we need it to. It says it does, but it's actually only covering here. So you can look at how this image over here cuts off our main location. And this one pretty much does everything we need it to. So we download this one. You have to sign in to download it. There's two options. So full resolution browse and geotip format. This is an image, which is much smaller than the tile. This image you would download just so you can preview things. The tile, which is here in this case, 448 megabytes, will provide us with much more detailed information. In fact, it includes several bands, which we then need to actually analyze our image. Now, you may remember that the previous sites we looked at actually had later information. So while USGS usually keeps all the landslide information, it keeps much of the Sentinel-2 information eventually. It's quicker to get the most latest information from sites that are dedicated to Sentinel. So some of those sites, again, are the Sentinel hub. So you see here, again, we downloaded that image from December the 18th. And if we want to take this down, we click over here. And we can choose our coordinate system. We should probably choose the WGS 1984. You want high resolution. And of all these here, you would take the TIFF. The TIFF is the one that provides the most information for us. We definitely want it to be georeferenced. These are geotips, so they're georeferenced. You can choose a specific visualization or not. Or we can choose raw bands. And if we choose raw bands, then we get this raw information that we can then put together within QJS and other programs. So these are the raw bands. These are specific visualizations. You can do both if you wanted. You can have a visualization and then have the raw bands to do your own analysis. Band 2, 3, and 4 are also what is known as red, green, and blue. So 4, 3, 2, red, green, blue. They are combined to make a true color image. And band 8 is the new infrared, which we use for things like NDVI. Again, if you're brand new to this, this is a lot of information to take in. But I'm giving you the big overview so that you can actually understand what you're getting into. If you were actually trying to do some analysis on particular things like burnt houses again, you would want to get specific bands in order to do that analysis. And often the near infrared will provide, in the shortwave infrared, will provide more than just the visual bands that are presented here. 4, 3, 2 in true color. Another option is to look at the Copernicus Open Access Hub. This is scihub.copernicus.eu. The Copernicus Open Access Hub provides access to the most up-to-date Sentinel data. So in this case, again, we're focused. I tried to search for Basanshar. I didn't actually find it. You can see again that the map itself here is not showing it. If I go into the cloudiness, I can see that my island is there, but it was not showing on the map. And now I'm selecting the actual area of interest. If I click over here, if I am ready to do my search, I have to click on Mission Sentinel 2. There are various options here. Given the short amount of dates that we have, I'm happy just to leave these options unselected. But they're about the type of platform and then about the type of product. We'll be able to see that within our search results. So once we've done all of that, we've got an area of interest. We've got Mission Sentinel 2 selected. We'll do our search. I need to get rid of that. Let's do our search again. So last time it selected Basanshar and the actual footprint. And then when I got rid of Basanshar, it just went for the footprint. So here we have S2A, which is the Sentinel Sentinel 2 data. These download URLs will cause us to download a large amount of data. So we have 700 megabytes here, 572, 710. It's good to know about the data products we're downloading. The dates are embedded within the actual description. So here you have 2019, 1223, 2019, 1228. So just a few days ago, the cloud cover you can see over here makes those images not worth it in terms of our ability to do a lot of the analysis that we want. We can go down here. Let's look at, again, our 1218. That's looking pretty good. Our 1223, we might be able to get some information on that. And here's another from 1223. So let's go ahead and check those two 1223s out to see which one might provide us the information we need. Admission, Sentinel 2, instrument MSI, the sensing date, and the size. So let's go ahead. This will give us a zoom to the product. This will actually see a little product detail, which will tell us more about the cloud cover and other possible data issues that we might run into. So here is our cloud cover. Again, not great for us. We've pretty much lost our ability to do much detection here unless we know how to black out some of the cloud cover and use the other sensors. But our 234 bands will be severely occluded, and those are our 10 meter bands. We're almost unable to use this. You can go through this, the gear. This is all of the data that would be in that 700 megabytes. You can actually see within the granule. So these tiles are called granules. If you click through, you can see that there's specific bands. And here they've grouped these bands by 10, 20, and 60. So as I mentioned, the 10 meter bands are mostly the band two, three, four, and eight, and those are the ones used for a lot of this type of high resolution analysis. I guess it's now medium resolution, but this is what we used to call high resolution 10 years ago. So if you downloaded those bands in the last image there, you can open up QGIS and load these bands. There's a simple drag and drop. If you're looking for where the bands are, you'll have to know how to unzip the files. Once you've unzipped the files, then you'll have folders. Typically, if you get information from Sentinel-2, it labels it as safe. So even this one here that I had, when I unzipped it, it had a sub-directory called safe. But if you download them directly from the website like I did for that 1218 image, then I just get the bands that I download directly, the two, three, four, and eight. I've dropped all these bands in here as well as their accompanying information. They have kindly given us a true color image, TCI, at 10 meters, as well as you can see band two, three, three, and four, and, oh, sorry, that's two. That's eight, and that's three, and that's four. You can see how different sorts of features really stand out in these different bands. And again, this is all 10 meter resolution. So what's the point of getting actual bands within something like QGIS? Well, you build a virtual roster so that you can actually detect things that you want to detect. So this particular virtual roster is bringing together, so I've got the symbology here, a multiband is putting red, green, and blue so that we have this visual image. If I change any of these, I'll get slightly different images that bring out different components. What I like to do in QGIS is to use the layer styling to really explore these type of images, right? So we can go in and actually change the saturation, the contrast, and the brightness to really bring out some features. I might be interested in what these dots here are actually representing and do some investigative research. So you can, just by simple color changes, you can use your naked eye to bring out some of the features. You could also, using these different layers, especially if you've downloaded all of the layers, then you can use these different layers to start building things like NDVI. So let's look quickly at what an NDVI might look like. So this is an NDVI and I've given it a specific color scheme that really makes the healthy vegetation stand out in September of, sorry, this is November of, November 18th of this year. So this is an older image, but you can see how actually studying this vegetation might give you some clues, gives you better clues about where the water's at, might give you some clues about new buildings that have come out. There are many, not too complex, but more advanced processes that you can use to combine, or things that you can see by combining these bands. And if you use the roster calculator, you can do things like create this NDVI. So it's a little bit beyond the scope of probably what you're looking at, but I want you to know the possibility is here. And again, if you just wanted to look at NDVI over several dates, you could use those previous websites that I showed you in previous slides or earlier in this workshop. And just to show you the difference between the Sentinel data and the Landsat data, here is a really beautiful Landsat image. It's a virtual band, which I created, virtual brands can be created by using this function in QJS. It's under roster, miscellaneous built virtual roster. But this is a quite good image in terms of cloud cover, very little of it. But again, you can see that our ability to see detail is much more limited by the Landsat. You might be able to use something like this to compare to the Sentinel data that you get another month later or a few days later. It's good to be able to sort of think about how to compare the different data sets, especially if you're using the naked eye, it's not that critical that they be exact same resolution, but you're trying to build evidence database for changes that might happen. Now, if you're interested in moving forward with learning more about actual data analysis and the sense of doing supervised or unsupervised classification, there's two main resources that I would recommend to you. While you can again pay for the commercial platforms that provide you high precision, 30 centimeter images, as well as automatic conversions to specific road sensing indices or use those websites I showed you earlier, the us.com or the Sentinel hub or other ones to actually render things online. If you're going to use QGIS, you really need to learn how to use what's called the SCP plugin. So this entire workshop here, you can download, it has the answers as well as the instructions. You can download this all using the link here. So just go to http.dot slash bit.ly slash remote SCP. Now this will show you how to develop signatures, spectral signatures within QGIS using the SCP plugin. This is beyond the scope of this workshop in the sense that this is probably about a three to six hour workshop, this one here, if you go and download it, that you'll be able to do. If you have problems with completing this or downloading anything, I would highly recommend, and even if you don't have problems completing this or downloading anything, I would highly recommend going to, from GIS to RS.blogspot.com. This is an incredible resource for doing semi-automatic classification and manual classification through QGIS. It's a wonderful plugin and it has many, many tutorials that you can walk through in order to get yourself up to speed as well. The user manual is really a goldmine of knowledge. So if we just take a quick look at that user manual, it comes as either in many languages, as either PDF or in links down here below. And just to be clear, I really believe that you should ground yourself some of the basics of the science before you move forward. So if you're going to do this, I highly recommend reading at least this chapter before even starting the tutorial that I showed you in Google Drive or any of the tutorials here on this website for the SCP plugin or semi-automatic classification plugin. That third chapter there walks you through basics of GIS as well as several multi-spectral satellites where you can see as I referred to these bands and their resolution as well as what wavelengths they are using. So if you do need to use multiple satellite types or data from multiple satellites, and you will know that these bands don't always correlate exactly to the same. For example, Band 8 here in Landsat is panchromatic and Band 8 and the Sentinel-2 is near infrared. And then you have your short wave infrared here, things like the Cirrus Band, which helps you actually pull out clouds if you get more advanced. So these are important tools and data sources for you to realize this is not just something you should try to do over a night or over a weekend. This is a pretty serious endeavor to get into and to do correctly. I'm happy to respond to questions if you have them. I'm also happy to put you in contact with certain resources in terms of people, human resources, as well as to point you to online free resources. I hope that you learned a lot from this workshop. I hope that you are able to, we're able to follow. And that you go and do use the links that I put into the notes on this lecture slash workshop to take yourself further and to use this amazing tool to support human rights. Thanks a lot. My name is Arthur Gil Green and let me know if you have any questions. Bye-bye and happy New Year, 2020.