Exploring Sentinel-2 multi-spectral band combinations in SNAP

hello everyone welcome to our next tutorial today we're going to be looking at the Sentinel tool box and how we can use it for downloading and visualizing satellite imagery my name is Sean Levesque from the geospatial ecology and remote sensing lab I'm going to be running this tutorial on my Mac but the great thing about the snap tool box is that it's multi-platform so you can run this either on Windows Linux or Mac and I'm just gonna make it fullscreen this is the snap Ireland snap it's produced by esa that's european space agency we'll start by saying a big thanks to the ESA for not only producing the satellites which provide us with such great data but also for for funding and providing a toolbox to analyze those data streams this program is free software you can redistribute it and/or modify it under the terms of the general public license as published by the Free Software Foundation now we've installed the full snap version which means we have the central 1 toolbox for analyzing Sentinel 1 radar imagery Sentinel tool to toolbox for analyzing multispectral imagery Sentinel 3 is Mars radar set and also pro Bovie now the tools from these different tool boxes are embedded within the menus so we won't be using the tool boxes individually but just be drawing on the two different tools as we need them for today we're going to be focusing on Sentinel two data which you might be familiar already with Landsat type imagery multispectral medium to high resolution imagery Sentinel 2 is more recent and has a very high spatial resolution with great multispectral wavelengths as well so if we want to be able to find and download some Sentinel imagery we don't have to leave the snap environment over on the right hand side is the product library we open that up we see this interface with the little world map can zoom in to an area of interest down in the bottom here you can use the right-click to pan the map back and forth use the scroll wheel on your mouse to zoom in and out its navigate up to the top orientate ourselves for those of you watching from elsewhere in the world we're based in Darwin which is up here for today's exercise I'm going to zoom in to an area right up in the Northern Territory world famous kakadu national park i'm going to draw a little box small box roughly in our area of interest so what this does is define our spatial area of interest now if we have existing data on our computer we can find it within the product library but if we're looking for new data we can access the ESA science hub you'll see that that'll bring up options of their different missions and we can easily access Sentinel one Sentinel two and Sentinel 3 data from this interface for today we interested in Sentinel 2 and we're going to be looking at the level 1 C data importantly a few more things if if we know specifically the products we're looking for we can type in the names acquisition modes and passes if we know the track number we can limit our search by using these variables for today let's start by searching for imagery over the course of one year so we'll search from 1st of January 2017 to December 31 2017 and a few other things we can include is the percentage cloud cover that we are willing to accept obviously in most situations we prefer to have no cloud working up in the tropics that's not possible except in the very middle of the dry season so let's put in a limited range let's go 0 to 5% cloud cover and let's hit the search button that can be found up here I must just mention that you may be prompted for a username and password you do need to register an account with the SCI hub from the ESA so that you can access the data through this interface there's no cost to having an account that's merely just registering username and password with if the European Space Agency's SCI hub now we immediately get some information down here in the status bar we can see retrieving entry information 39 entries in this little status bar shows us our progress what that means is that they've searched the archive for the year of 2017 for images with less than 5% cloud cover and there are 39 scenes that intersect our little box and depending on how many entries you find it can take a while to download and keep an eye on the status bar down there this workspace here will then be populated with the different images and you can navigate through select the ones most appropriate and download them for now we won't wait for that to complete I've downloaded a set of images already which you can find in our Sentinel 2 folder and let's have a look at the file names very long file names the first part telling us Sentinel - that's s - we have s 2 s and s 2 B's bearing in mind that Sentinel 2 is a constellation and as such as actually two satellites which are opposite each other in orbit allowing for a more frequent revisit time this tells us that it's a multi spectral imager and we're dealing with level 1 C da da means it's really useful for us it's not a completely raw product what can sometimes be confusing is that there's two dates in the filename you can recognize this 2017 of five twelve year month and day and looking at cross over here 2017 of five twelve this is see this is pretty consistent throughout that the dates match they might not always match the first date is the date that the image was actually acquired and the second date is the date that it was ingested into the archive so they can sometimes be a difference in that will primarily always be interested in the acquisition date so if we saw it here we have a number of different images obtained at C November got to in August's one in October when in September when in June and when in May so I've selected a range of images throughout the year let's open up the the first scene the earliest scene so that's from 12th of May 2017 and you'll see that these are zipped folders you can unzip them and open the XML file but snap does support just opening the zip and it can read the contents of the zipped file it's quite a nice way of transferring these we click on this little downward arrow it might look slightly different on Windows machines but in general structure is very consistent across different operating systems clicking on this little downward arrow we have four folders metadata vector data bands and masks open up the bands folder and you'll see that this Sentinel two image like all Sentinel two images are comprised of 13 bands see then they first looks like there's only 12 but remember that we have two bands in band eight there's band eight and band eight a then a very similar wavelength so we range from 443 nanometers through to 2119 now if we are every word hope' well we can open up each of these individually which we'll do in a moment but usually what we want to do when we first open up an image is have a look at a true color RGB composite and we can do that just by right clicking on the file name and you'll see we have another quite a few options pop-up so just bear in mind when navigating snap familiarize yourselves with all options under these different headings up here but a lot can be done in context so right-clicking a file gives us the option of opening an RGB image window so let's do that and what's quite nice is we have a number of profiles automatically built in and we here we have a profile preloaded Sentinal to multi spectral imager and natural colors remember that for a natural color composite that's very similar to what our eye would see is like as if we're looking out of an aeroplane at the landscape and this is a true color matching we're putting band for 665 nanometers into the red channel and 3 560 into the green channel band to 490 into the blue channel so let's open that up remember that Sentinel to starters at 10 meters resolution in the visible range so quite big images take a little while to open the whole scene and there we there we have it beautiful images some lovely differentiation in the ocean there we have quite a brown looking landscape a few clouds hard to avoid clouds in the tropics we can zoom in using the scroll wheel on the mouse we also have a zoom tool up here as well as a hand for panning moving around the image coming down to the left here we have an navigation panel click on that we can use this also as a way of navigating this is our keeping context of where we are in the image so if we zoom in a bit closer what's quite nice if we use these magnifying glasses the one with an a takes us to the full area of the image one of the P brings us into the pixel level so the highest resolution of that image we can grab that little box and move it around and use that to explore the image in higher detail and we can see some very nice patterns it's an amazing landscape kakadu national park northern australia while we're having a look around here let's just see what other options we have open an RGB image window to do that again but now we have a couple of other options so if we use this one false color infrared we've now put band 8 which is infrared band into the red Channel band 4 into the green and band 3 into the blue so what we've done is we're making a false color infrared composite let's click OK and this view will be heavily optimized towards exploring photosynthetically active vegetation and that's why we see so much red in the resulting image red here being green vegetation in reality so absorbing and reflecting light differently let's zoom in a bit closer here have a look at some of these areas alongside the rivers see a lot of green you see a lot of very interesting spatial patterns in this landscape let's come right up to the edge possibly some mangrove fringes and some riparian vegetation with savanna woodland in upland areas now while it's great to be able to do that what's very interesting is to be able to view both the infrared and the true color at the same time so at the moment we have these in two tabs up here and we can flick back and forth between them but a more elegant way is to tile the images either horizontally or vertically I'm gonna tell them vertically for now and make sure you've ticked on these two little icons here this one the bottom one synchronizes cursor positions across multiple image windows and this one will tie synchronize the views across multiple image windows and what that means now is that both the upper and lower image are showing exactly the same area and that I can move through them at the same time so last way to navigate is just to grab this little window in the bottom left-hand corner and as you move that slowly around we see the same image in false color infrared and in natural color that's a very nice way of exploring these differences in in how light is absorbed and reflected in different wavelengths what I'm going to do now is close these two scenes and I'd like to look at a few features in isolation so for instance let's start by bringing up band 2 open image window here we have it let's open band four let's go to band five and and eight so now you can see we've got these four images open moving from a narrower wavelength stepping through to longer wavelengths and becomes increasingly apparent that as we move from the shorter wavelength we see a lot of detail in the ocean and the land is very dark that means a lot of absorption of Al and a lot of reflection over the ocean as we come through to the longer wavelengths moving into the near-infrared see the ocean becoming most completely black meaning that infrared wavelength is almost entirely absorbed by water but we see that much brighter coloration over the land because vegetation is strongly reflecting the near-infrared it's a nice way to look at this is to go back to our windows and say tile evenly and let's zoom out to the full extent now we have these four images on our screen and we can see this nice transition moving from the very short wavelengths sort of the blue lights through the greens Reds into a near-infrared so I'd like you to take a few minutes to explore some of these different bands and close the windows we have open and have a look at some of these different combinations remember that in the as we get further along the wavelengths to the longer wavelengths we move into the shortwave infrared and that some of these bands will have a slightly coarser spatial resolution but our sampling much longer wavelengths and I'm just opening up here the first band in the shortwave infrared see the substantially higher and the visible and you can see how pure black ocean and the rivers are as it's completely absorbed by water nothing reflected back you can also experiment a bit with different RGB visualizations so if we were to open this up and try a new combination let's say we wanted to bring in that shortwave infrared into our hard UB composite and make something like this using bands from the shortwave infrared the near-infrared and the visible you'll see we get a little warning here reference to rest as are not of the same size now that's because the raster resolution is not constant across the wavelengths and our visible bands are all at 10 meters resolution but the shortwave and some of the red edge bands are at 20 meter resolution so if we click OK it's gonna give us a little error saying that the image can't be created was we're trying to mix bands with different resolutions what do we need to do is head up to rest are geometric operations resampling and what we'll do is just re sample those 20 meter resolution bands down to the up to the 10 meter resolution of the visible bands so this is a pretty consistent interface for for snap you have your input and output parameter tab and then your actual action tabs over here so let's go input output the name of the product is the one we have already selected target product is automatically named just with resampled on the end we'll leave that for now resampling parameters to find size of resample products what it allows you to do here is to either to find the pixel size the spatial resolution by a band from the source product or by specifying the width and height of the image or by specifying the pixel resolution if we come up to reference band change that to band 2 it's now one of the bands in the in the visible spectrum we know that's at 10 meters we have a variety of different sampling methods we just going to use the defaults for now so just be sure that that's banned too and we'll say run now something that snap will do is inform you that the target product has successfully been created and opened in snap but actual processing of source to target data will only be performed on demand for example if the target product is saved when image view is opened so that means it's putting it on the to-do list but it's only actually going to do it when you really need it so you can just say ok and we see that that resampled image has been added over here so we can say close now let's just quickly look again at our RGB options for this first image so when we say open RGB and week on the profiles you see we only have these three options natural colors false color atmospheric penetration let's have a look at the new image we've just created open RGB image window and let's click on these profiles and you'll see suddenly we have a lot more options that's because because we've resampled all of the Bands to that 10 meter resolution we now have more combinations that we can play with so let's go to the shortwave infrared composite that's the one we're trying to do earlier then 12 8 & 4 and say ok and now that opens up nicely for us allowing us to incorporate the shortwave infrared that much longer wavelength into our image and our analyses play around a bit spend a bit of time having a look through all these different options see some dedicated towards urban detection features agriculture healthy vegetation separation of land and water natural look with atmospheric removal if we take one of the healthy vegetation option we'll see we've got red got band 8 in the red channel we're pulling in that shortwave again into the green and then keeping blue in the blue just have a look at that one this one really highlights differences in in vegetation so actually let's it's close other one leave healthy vegetation open and let's open up the false color infrared that we had earlier so that's sort of the more a quick way of looking at photosynthetically active vegetation if we tell this horizontally rather let's rather go for a vertical tile and zoom in that little island again and what I want to just show you here is that if we're only using the near-infrared so this false color infrared is using the near-infrared and the red and it shows us photosynthetically active vegetation but having a look at this lower image as we as we zoom in here and bring that up you can see that we've got much more detail in the vegetation we can start distinguishing if we look at this top image we wouldn't think that there's much difference between the vegetation say over here on the left-hand side over here when you bring in that shortwave infrared really helps you see more subtle differences in the vegetation and that's also true as we move through the rest of this image so that's all I wanted to get to for this first tutorial on band combinations in the next one we're going to look more closely at spectral responses by investigating spectral response curves so thanks for your attention and hope you found that useful and before we end in just one more thing most forgot if we head back to our product library up in the top right corner you can see that this has now been populated with a whole range of images that intersect at that little window that we drew earlier and what's great is down here on the bottom is a little timeline you can see when these images were collected throughout the year and just scrolling down here you can click on a particular image you get a little scroll window that pops up and you can see what that looks like quite important to keep in mind the file size each of these full scenes is about 700 megabytes and once we click some of these we have the option of if you highlight that you have the option then of downloading it so that's all for today and as I say we'll be following up a new video will be posted each week I hope that was useful and see you next time

Channel: GEARS - Geospatial Ecology and Remote Sensing

Views: 52,379

Keywords: Sentinel-2, remote sensing, multi-spectral, satellite, open-source, ESA

Id: vtlN5MXYGaY

Length: 29min 20sec (1760 seconds)

Published: Tue Mar 13 2018