 current slide you should see my contact information moving through several different documents and websites and programs today so you should you should hopefully find some links in the text below this video. Notes for the video will be found in the links. These notes provide a little bit of background as well as an overview of what the focus of this training is on. This training includes direct responses to the transitional justice working groups questions about remote sensing for human rights. Notes on moving forward with more technical modules in satellite data acquisition and analysis. And two modules modules on accessing data and a module on imagery processing. This video will overview will provide some background to the work that's done in both those modules. Finally the notes for this training include a list of useful links for moving forward as I know many people watching this training may have little to no background in dealing with aerial or satellite imagery or doing remote sensing. The videos and notes for this training as well as data which can be given to you upon request are openly licensed. So without further ado let's move forward into some of the actual responses to the questions from the transitional justice working group. What are the sources of this imagery number one? The primary types of images you'll use for looking at human rights monitoring are free Landsat or Centennial II images. If you have a budget for it you can purchase images from Maxar and other providers Maxar has bought Digital Globe which bought another group called Global Eye and therefore has access to some of the best imagery commercial imagery in the world which is a 30 centimeter resolution. Most of the free public imagery maxes out around 10 meters so it's a giant difference between 30 centimeters and 10 meters in terms of analysis and imagery but of course spatial resolution is not the only thing that's important for human rights work. We also need to look at things like temporal resolution i.e. how often images are being collected and spectral resolution and radiometric resolution. So these are issues that are some of the background hopefully that you can read about or that you already have in terms of understanding remote sensing. How would you go about contacting providers or getting the source information? One of the main places that I use and that I would recommend to people is EOS.com it's here my mouse is over now. If you are planning on buying small images you can get a lot here for less than a thousand dollars. If you are ready to take the next step and pay a large amount for what you're accessing because you need higher resolution then I would recommend going with Maxar's Secure Watch which you can see here is still linked under 2B under the DigitalGlobe.com website. So Secure Watch will provide you with about well thousands of images a day regularly updated and 30 centimeter resolution. So that's pretty much the highest class of what you get unless you're creating your own images using digital photography or paying someone to fly over a location to create aerial imagery. Now I know that's out of the budget of a lot of people and that's a one-off temporal resolution. It's a one-off event so it's probably not what you want to be doing. It's better to pay a provider than to try to arrange those types of events where you're flying someone over a region. How can people navigate existing open source imagery? I highly recommend Earth Explorer from USGS.gov so United States Geological Survey aggregates data from the European Space Agency and many other agencies around the world in its Earth Explorer and we'll go through how to access information in this tutorial from the Earth Explorer. Most of the information the data that you access through the Earth Explorer can be done for free. There are some things that you need to pay for. We'll go through options as we move forward. So what are some of the summary notes that you should know? Some things you should know before perhaps we take on this workshop. There are several ways to access imagery. You can create it and get it for free. You can pay for it. The best is going to be if you pay for it or if you create it and again if you create it you're probably going to be paying for it. There's a lot available for free. I will focus on that in this workshop because I know the budgets of a lot of human rights groups are not such that they can afford thousands of dollars a month to do monitoring. However, if you're willing to pay Maxar SecureWatch is one of the best. It's a company again that has consolidated digital globe and global eye with other services. No additional software is provided or sorry is needed. It's an internet platform so you can go on there and you can get a 30 centimeter resolution work which is great. You can also have some basic analysis done on the fly. The types of analysis that are done on the fly are things like NDVI so it was the normalize different and vegetation index so there's many indices like that. We have one for snow one for water and they take different bands from different satellite images and combine them in different ways so that these sorts of things like water, snow, vegetation are better represented in what we see. You can read through the rest of these notes on your own. I just want to take a moment to point out that there are two primary approaches to image classification. Some of the image classification that you do can be that you might want is can be done through these web interfaces but more specific looking at things like disturbed soil or burnt homes analysis that might have to be done on your own computer so I would recommend something like QGIS. Alternatively if you have the budget something like ARC GIS might help you work through these and actually develop spectral signatures and accuracy assessments that you can apply on the ground. Again it's not recommended for you to try to develop this these supervised classifications. Supervised classifications when you develop spectral signatures. It's not recommended for you to do that unless you have some background training or a very deep understanding of the science of remote sensing in GIS. I have linked here to something from the Phosphor G Academy. I'll highlight that now. I think anyone who does this workshop would benefit immensely by just going through their remote sensing and aerial photography module that they have. Primarily in this module or in this workshop we'll be using QGIS and websites. When we use QGIS we'll be using it with grass. You have the option of installing Sentinel toolbox. Sentinel toolbox is a great way of dealing with several different types of satellite imagery but specifically developed for Sentinel data. So if you can go get both of these programs but most importantly QGIS and install them on a computer then you will be able to follow along closely with the tutorial in the videos. So let's start with the most important aspect of our analysis. Where do you need to do it? So one of the most important aspects of our analysis is not just a place name but a scientific way to actually measure where something is in the world. We have grids and all those grids are coordinates. Here we're looking at Basant Char. So Basant Char is in Bangladesh. It's the potential site of a refugee camp for the Rohingya. So if we're looking at Basant Char and possible flooding we need to first find out where this is. So once we've found out where it is there's a couple things to to think about. First of all many maps will not actually show this location. They won't label it and they may not actually show it on their actual vector graphics file here. So I'll show you a couple examples later but if you look up Basant Char you may not actually see it on some online services. Google Maps is pretty good in terms of including these location names and places. Now once you've located Basant Char there's a couple different things you might want to do. You might want to check the satellite imagery. So we're doing this in Google Maps. You could probably do this in Bing or other services but this is the one I know the best and it's universally accessible. So you might want to look at the satellite imagery. So here's the satellite imagery. One is one of the immediate problems we notice with the satellite imagery. You can see that some things are blurred out. Some things are more detailed. Google's satellite imagery is combining sources from different from different providers. We zoom in a little bit more. We get some better detail but you can still see down here and around there there's it's missing a lot of detail. These are tiles which are provided in a roster format. We can't do a whole lot with Google Maps very easily and in fact you probably shouldn't rely on the reviews in Google Maps for certain places as this refugee camp has been rated 4.3 out of 5. So you know just be aware of the politics of what's going on behind the scenes what data is provided where in the world. Most important thing we need to find is the latitude and longitude and decimal degrees. So if I right click anywhere on the location that I'm interested in and right here we see the newly developed refugee camp the buildings. If I zoom in more you can see a little bit more data. We'll talk about why some of this imagery is much better than what you saw just a second ago. So if I but before we do that let's just look at how do we actually get a point that's interesting to us. We need to know where this is an exact space relative to the rest of the world. You can right click and Google Maps and say what's here. That provides us with the accurate reading of the decimal degrees and latitude and longitude. These are things you should copy elsewhere maybe in a text file or some sort of note-taking document some sort of note-taking program. So let's just copy that and we can put it in Google Keep. We can put it in just a simple text document so that we can access that later on when we need to know where in the world we're studying. So that's the whole point of using Google Maps. Now you could do nearly exactly the same thing using Google Earth. So here's Google Earth and because it's again based off of Google's naming system you can search for Basan Sharping the dash. It will take you in to the actual location. You can also here get the decimal degrees and the in latitude and longitude or the degrees minutes seconds. Right now my Google Earth is showing this in degrees minutes seconds of latitude and longitude. With a simple change of the settings we can choose that to decimal degrees. Again you see that some areas here clearly don't have much good data collected and if we zoom in you see that that land area has a lot more than the ocean area and this land area that we're looking at this camp that was recently built has some pretty high resolution imagery. You can see a pool outside of what appears to be a possibly an employee house there whereas you have your refugee camp over here where you're looking at construction largely. Now there's several interesting things about Google Earth that you should know right off the bat. Google Earth is easy. You download it you install it on a computer anyone can use it. You can give people things called KMLs and KMZs. These are types of files where you can actually show them images or locations in the world. You can develop tours which also shows people specifics about a location. You should also know that Google buys its imagery from specific companies or municipalities. So as I zoom in here and I get this really good imagery which we're looking at maybe a meter or less here. You see at the bottom you have a copyright from Maxar Technologies. So Google has bought this from Maxar which is the group I mentioned earlier that bought Digital Globe which consolidated also Global Eye and they provide some of the best commercial satellite in the world and Google in buying that can provide it through Google Earth. You can see make out boats here. You can even make out probably some of the people moving around on the on the coast. So this is um it's a great way of visualizing what's happening in a location. Now the temporal resolution of Google Earth is not that great. So if you go up here you can look at historical imagery. We know we're looking at something from January 11th 2019. So that's about almost a year back from now or in December of 2019. And if you go through time with Google Earth you'll see the changing landscape and changing resolution of images that have been collected. This is in August 15th of 2018. This is in April 8th of 2018. So you can see again a big change in terms of the construction. You this is all using the naked eye. Now this is a lower resolution image but still providing some of the the same details from 2018 March of 2018. If you go back to March 6th you keep going and now we're at November 19th of 2016. We still have a decent image but we can see that everything has been built since that time. We've skipped pretty much an entire year a year and a half to get from November of 2016 to March of 2018. So not super temporal resolution in Google Earth although because they buy their images they do provide a lot of resolution which is which is great. It's great for displaying information. It's great for sharing information. Not always perfect for analysis because of the temporal resolution issues. So how do we overcome this temporal resolution issue? In this video I will alternate between two possible satellite imagery sets looking a little bit at land site but also Sentinel-2 to talk about how we overcome those issues. We can download data from USGS Earth Explorer for Sentinel-2 and for Landsat. Landsat imagery does not have the temporal resolution or the spatial resolution of Sentinel-2. Landsat comes at about 30 meters squared whereas Sentinel-2 comes at 10 meters squared and Sentinel-2 is collected more often than Landsat because it was launched after. It's a more advanced platform. Currently we're on the eighth generation of Landsat so Landsat-8. More satellites will be launched in the near future which will push Landsat and Sentinel farther beyond their current generations but Landsat-8 is pretty much what we have as best for Landsat right now and Sentinel-2. While there are multiple types of Sentinel satellites Sentinel-2 does the best resolution for land-based analysis. So what we have here is the Sentinel hub. The Sentinel hub is linked to in the document. We are looking at a Sentinel so you can change your different settings here. This is a Sentinel level one Sentinel level two and you can see the difference between the two. Both of these are showing different sorts of atmospheric correction. Level two is what we want. It provides atmospheric correction clearly that shows ground details much better. This again should be at 10 meter resolution. In the Sentinel hub EO browser you can choose several different ways of displaying your information. So on the left here you can see that I'm right now based on true color. I can also do a scene classification map. I can do false color, false color. So these are presenting different parts of the scene in different ways. So you might want to experiment with this. You don't need any extra software on your computer to do this. This is all based on combining the different bands from the satellite together. Once you're using, you might have noticed the resolution just deteriorated badly. Once you're using bands which are not two, three, four, eight, you're going to lose a lot of resolution. So you see here in 12 and 11 and four band combination. Sentinel collects more than 10 bands that all collect different things like shortwave infrared, near infrared, so on. You will see a loss of resolution. So here this is a combination to show the normalized difference in vegetation index. So it should be showing areas that have high amounts of vegetation and low amounts of vegetation. One of the most interesting things in terms of human rights monitoring, in this case people wanted to know about flooding possibilities. So the NDVI might help that. In DWI, you can see it's very similar to NDVI. It's using some similar bands here that's showing the waters. That's the normalized difference in water index. And it is sensitive to vegetation but it also clearly shows any water present in the image. How do you search through these results? How do you search for these types of images in order to actually see everything that we're seeing now through these different sorts of presentations of indices? If you're in the EO browser, you can search for different forms of data. Again, Sentinel-2 and particularly L2A, which is the level 2, is the best for this type of monitoring because it gives you 10 mean resolution. You can see now that I'm in the search function that Basson Char has disappeared. It's no longer there. Whereas I go back into visualization, you'll see it's there again. So again, as I mentioned, sometimes the vector graphics may or may not show you the actual geographic features and locations you're looking for because they haven't been transferred from satellite imagery into vector graphics. So we'll talk a little bit about how to deal with that but one of the main ways of dealing with that if you're not dealing with Google Maps is to actually know the latitude and longitude of where you want to be. So here we've selected Sentinel-2, L2A level 2, and I set a maximum cloud coverage at 11% so that no images that had 11% would be included. I put December 1st to December 1st, which is tomorrow 2020, and then I did a search. These are the results. You can see in the EO browser we have maximum cloud coverage. We recently have an image from December 20th. We have one from the 18th. Pretty good temporal coverage, very good spatial resolution, another from the 15th, and it's up to you to decide what works best for your needs. There are many functions within the EO browser that may allow you to avoid completely, actually allow you to avoid completely downloading data and putting them into a other software package. So here now I have the track 45 QZE selected. If I zoom out a bit, you'll see that these are all different Sentinel tracks here. Click visualize, you'll see that appear there. So this is the true color. Now there's many ways of visualizing this type of information and getting this type of information onto your computer. One other option I should show you in terms of visualizing it without having any specialized software is through the EOS.com browser. The EOS.com browser is a little bit more flexible. It also allows you to download commercial data. You can set areas of interest. So here I've set my area of interest as Basant Char. And basically you're seeing some of the best right now, 10 meter resolution, but the best free imagery of a location right as we look here. There's all sorts of different ways you can change the settings. So let's first look at the areas of interest. If you set up an account with EOS.com, land viewer, you will be able to establish your area of interest and always come back to it easily. So here you see the lat long and decimal degrees. That's why it was important to take that information earlier. And here you see the result of a search. The search which is done through their interface of the browser allows you to take up to 10 free scenes per day. A scene can be you can either have a mosaic or a scene. A scene is one image. A mosaic is multiple satellite images. So if I go into the search settings I can take passive sensors. Passive sensors include things like Sentinel-2. Sentinel-2, level 2. I have selected here. We could also select one of many others. We don't have to have all these in our outcomes. As I mentioned, Landsat-8 is the most up-to-date recent one for if you wouldn't want to see human rights monitoring but still is around 30 meters. You can use one of the bands in Landsat-8 to do what's called a panchromatic correction and get it down to 15 meters. So there's some ways to sort of compare and combine in your analysis. Landsat-8 and Sentinel-2 but that's a bit advanced for what we're looking at today. So you can adjust cloudiness. You can adjust sun elevation. Let's take our cloudiness again down to around 11. You can adjust the dates. It's like a December 1st to December 31st. And we're looking at passive sensors. You can also look at high resolution imagery for an analytics and high resolution for you only. It turns out that in this case for Pasanchar there is no high resolution imagery on this platform because it is a newly formed island over the last 20 or 30 years. Which appears maybe to be growing but is a very unstable in terms of land masses. So once we've got that all set up we should have several different results over here. You see that we have Sentinel-2 from December 18th. If I change the cloudiness we'll get more Sentinel-2s from the 13th to 3rd. But again it's important to make sure that we're getting a good image not one that is included by clouds as that will affect how we do our analysis. So the image I've loaded here is from the 18th to December of 2019. You can see it has 0% cloudiness, 42% angle and its track number is 46 q c k. We're looking at level 2 data which has been atmospherically corrected so we can do a lot with it. One of the nice things about this interface is you can actually do a comparison slider. So on our left now is is the image I originally loaded and if I wanted to load a new one for the right I would click down here and load a second image. This sort of comparison allows you to quickly look at changes over maybe a small period of time as long as you have two good images to look at. That's all being done by the naked eye of the human whereas I think what's important is to think about certain types of combinations of the bands and what that tells us. So if you do the comparison you can use either the naked eye of the human through this natural color or you can use many of these bands which are developed around Sentinel-2. So this is the burn index so looking at burns we have fire detection band so again pulling out specific areas there may have been fires. We have NDVI looking at the health of plants versus the health of the the unhealthy of urban areas. We have SAVI, we have EVI, so on and so on. You can look through these and learn more about them using the information buttons to understand what they might be showing and how that might be useful for you in your own analysis. You may be wanting to look at a specific period in time to look at where water features were in order to understand where people might have put for example victims into water and that changes over time. If you wanted to specifically look at something like disturbed soil or some sort of specific feature that you need to identify spectrally then you would need to most likely download these files and do analysis on a software application in on your computer. If you're just monitoring things like military build-up this could be useful in the sense that you get some pretty decent resolution using even just the natural color but you could you look at things like the urban areas to see if you can pull out areas that were recently cleared or forestry areas that were recently cleared, healthy vegetation areas that recently cleared that would allow you to be alerted to maybe a specific military build-up over a small window of time. This is the forestation index shows over a period of time areas that have appeared to have lost vegetation. If you need to download a scene there are several options and I'll go through how to get these through USGS Earth Explorer. Before we get to downloading we can look at a couple other things so here's time series analysis you can look at how things change over time this is a normalized difference in vegetation index and water it's a great tool and this is scene downloading so you can actually adjust where you would like to download just the size of it. Right now every pixel here is giving me about 10 meters. You notice does that make the window larger these pixels are they change right so if I make it like big like that it's 20 meters whereas if I go down small I've got about five meters. Now with a free membership to EOS you're only allowed to do the small version of downloads so they're encouraging basically people to get a subscription but if you're only focused on quite small areas you can get five meter resolution of a small area no no problem there's three different options for download the GeoTIF the JPEG and the KMC the JPEG is an image which you could put into a report the KMZ or KMZ you can actually put it into Google Earth to help people visualize exactly where in the world this is happening. In GeoTIF you can use within a GIS program to visualize a specific location and maybe even do some analyses. Another internet interface which allows access to great data is the Sentinel hub so Sentinel hub is focused on data from the Sentinel satellites which are part of the Copernicus launch and there are many different ways you can do searches here it's got a couple other things in it like Landsat and MODIS but it allows you to segregate by the level two and the maximum cloud coverage we get our results and then you can visualize it with on the map once you visualize something you can put it into true color false color and so on so these two different platforms the Sentinel hub and EOS.com their viewers are great ways of getting access to Sentinel data and Landsat data but they I think that the EOS.com one it appears a little bit more powerful for to me you can actually adjust what bands are being used within your image if you need to you can look at specific indices and crop your image as well you can look at some information about fires and and so on now the based on what I know the power of Sentinel hub is that you can actually go in and do custom rendering of the bands now you can do that in EOS as well but here's a a view of a normalized burned area or burn ratio where people are looking at recently burned areas this obviously would be interesting if you're open to looking for or if you're searching for recently burnt villages looking for areas that have experienced fire and possible population movement in tandem so both of these are useful resources but they and again this is not this one is a wildfire map it's not in Basant Char but both these are useful resources they are not the most detailed way of getting that information but they provide quick free and easy ways without a lot of technical training to look at images and try to get more out of one of the things you should know about classifying images and interpreting them the more you know about the ground the better it helps you interpret objects better that knowledge of the ground impacts your accuracy assessment there are scientific models and algorithms for accuracy assessment those models are based on spectral signatures in order to do a spectrum of signature you would be identifying certain areas as forest certain areas as urban as you've seen here they've already built in several ways of doing that just through the through their interface