RUS Webinar: Flood Mapping with Sentinel-1 - HAZA01

hello everybody good afternoon from Rome my name is Miguel Castro and today I will be guiding you through the second route webinar which topic is foot monitoring with Sentinel one day so before starting let's quickly have a look to the objectives of this session what you can expect in this way now you will learn the basics of image processing for flop mapping and for that we will use the route service to download process analyze and visualize the free data acquired by the Sentinel satellites so just before starting also be aware that this webinar is going to be recorded and that you will be able to repeat the exercise by yourself but don't worry about that I will give you more details later on so that you can see exactly how to do that let's first have a have a look at the outline of the for the next hour hour and a half we will first have a look to the study area and some general information as I said what floods we will also cover a little bit the theory about was not sensing and how it is used for this type of situations we will then learn what is routes and what can routes do for your projects with Santa Laila how it can help you and then we will use this service to perform our exercise starting by downloading the images and finishing by producing a final product that can be using that type of situation and finally at the very end we will have some time for Q&A so 20 to 30 minutes more or less not fixed time and so you can send if you have doubts during the webinar just send it to us to the chat tool that you have and doing that session will start to to answer to you so the duration for this webinar would be more or less one hour three minutes so now yes let's start and let's have a look about some general info so among the different natural hazards fault accounts for almost half of the weather-related disasters recorded during the last 20 years the Republic of Malawi declare a state of disaster on January 13th 2015 in districts of the southern part of the country were affected due to heavy rains related to the cyclone Bansi and the tropical storm ketse so throughout December 2014 and January 2015 more than 150 percent of normal rainfall was experienced in the area the flood left unfortunately 276 people dead 170,000 people displaced and some areas completely inaccessible it also caused extensive damage livestock infrastructure and crops in total 64,000 hectares of land were damaged for good evening the human time disaster that happened so how can we use in this situation remote sensing or satellite data well during a flood event precise information is highly required by decision-makers to improve response activities they need the information as soon as possible so that they can manage properly the resources they have to help the population in this context satellite earth observation company can play a key role in order to provide that information but flooding flood events are usually linked to cloudy conditions and this may useless optical remote-sensing which relies in the back scatter light that comes from the Sun and this this type of radiation cannot penetrate clouds but there is a solution for that and that is synthetic or synthetic aperture radar also known as SAR as it does not depend on soil elimination or on weather conditions sorry mod sensing is quite relevant then to monitor emergencies such as floods so for this exercise we are going to be using Sieben Sentinel one sharp data provided by the Copernicus program The Sentinel satellites are included as I said in the space component of the Copernicus program of the European Union and European Space Agency and once completed this program will be formed by six constellations of two satellites each with a range of technologies from SAR to multispectral imaging and but not least the data are freely available to any which is the user and that is a key feature in this new generation of earth observation satellites so let's now move into the route service and let's learn a little bit more about this product so first of all routes stands for research and user support for Sentinel curve products it is an initiative funded by the European Commission and managed by the European Space Agency with the objective to promote the uptake of Copernicus Sentinel data and support R&D activities the service provides a free and open scalable platform in a powerful computing environment hosting a suit of open-source tool boxes pre-installed on virtual machines which allow you to handle and process the data derived from the sending of satellites so what does that mean in other words what Bruce offers you a cloud computing solution that is a virtual machine with the necessary storage and processing capacity to deal with large amount of data captured by the Sentinel satellites in addition to all that ruse also provides you specialized user helpdesk to support your mode sensing activities with Sentinel data and a dedicated training program so we have a couple of websites where we have all the information about this project so let's have a look to those websites and let's learn a little bit more about the details of this program so let's go first we have the goose training per photon where all the information about our training activities such as this webinar is shown and then we have the Rouge Copernicus that you put on where you can subscribe to the service and perform your request for beta machines etc but don't worry I will show you that right now so let's go first to the restraining bottom so here we are this is the home page we have some general information about the project have the project an intro video so news about events if we go for example to the news tab you can check HERE the our activity what we are trying to do so it's a good way to stay up to date and don't lose any any activity or chance to to receive this type of support those have or the training tab where you can see the upcoming events right now we don't have anyone planned but don't worry we will have new sessions person something very important is this past training sessions if we could here we can access the list of events that already took place for example the fab mapping sensor one last Tuesday also another way now which tip detection and this is very important because if at some point what you learn something and you didn't manage to subscribe for this way now you can come here click on the webinar and you will see the recorded video of that women are some instructions to BPD exercise by yourself within a course so very very relevant webpage the other one it's a ruse Copernicus W and this is the main the home page so here we can see for example again some news about the project also the conferences we attend the events we organized you can also see some basic info about Bruce service such as what is exactly the rules who can use it and how does whose work etc but something relevant is the routes offer and more in detail unity computing environment so if we go here and as I said before the bit on machines of who's come with a list of pre-installed software so here you can have a look to this list and you can know exactly what you can expect once you receive your virtual machine so for example we have observation software such as now we have also QGIS some processing libraries very well known such as denial or fail in books and also somewhat development tools such as R or Python and a very long list so you can you can have a look here to tell of this something also important to highlight is the ICT resources so when you apply for a rental machine you we will ask you and you will see that later on for details about your project so what type of finances are you going to do what type of data and storage capacity you need and this is because based on that you will receive a different virtual machine so based on your needs you will be a level a B or C and as you can see this table you will have a different setting for processing and storage capacity so here you can have all those details great so let's let's start to interact with the root service and so the first thing we have to do is to register in this in the project and for that we go to the login tab in this right upper corner the website and we click login so here we have two options we can either register for the service or again of course let's first register and let me show you how you can do that so very easy we just need to fill in the information here so let's do it very quickly [Music] okay so that's basically all you need to have to do all the things you have to specify of course the cuticle and once you are done with that we just need to click on register so after the registration you will receive a confirmation email to activate your account and in that way you are already a user of those and you can start to request the services we provide so in my case I'm not going to click register because I already have an account but that would be the process if you are a new a new user so let's go back to to the portal and let's now login to show you the webpage so we go to login again and we click login we specify our credentials and now we are we are running the website so first thing to note is that there is a new dedicated tab called your room service so it is here when you can interact with with roots and start your project so you have three options your profile well you have all the information you provided during the registration process you also have your training in case you are invited to an event and you can here insert the training code but the relevant part is your dashboard so let's have a look so once you click on your dashboard this is what you can see so so far you can see that I have two projects this is because I have a rainy request a virtual machine but in your situation if you register for the first time to lose you will not have course any project yet so you have to apply for that and so let's let's think that we don't have any project here and what we should do is to request a new user service so we click in this tab here and now requests of information starts so as I said before we need you to specify your your projects and the type of finances you are going to do that you're the virtual machine that will be assigned to you meets your needs let's go through this request to have a look how we can ask that virtual machine so first thing well some years of experience if we have already worked with Copernicus data or you have already processed that and here something very important so as I said at the very beginning you will be able to repeat this webinar and how to do that is like that you register in ruse and you request virtual machine and in the first step we are asking you to introduce a training code that when you code I will provide it to you at the end of this webinar and by specifying it here we will know already that you want to repeat this exercise and we will give you the virtual machine with the necessary data and a step-by-step guide so that you can follow the same exercise I am going to perform in few minutes so if we click here and we write for example the code we have two options and we can say that we want to use the virtual machine only to repeat this exercise or that we want to use the virtual machine to repeat the exercise but also to keep working with our own products so let's use this second option for example let's go to next so next thing to do is to specify a little bit more about what we are going to do so first thing to say is the somatic area so for today we are going to select this one then the type of operation we are expecting so let's say for example algorithm development then we have to specify our preference so the route service also provide an option so that you don't have to download the data so you can just send us the details about the data you need and then once you have your virtual machine everything will be stored there so you can skip that part and focus on your analysis but you can also of course say ok I want to know my own data and then we'll just click self downloading next thing will be some details about your activities and what you are expecting to do so well you can write whatever you want but just remember the more details we have the better we can understand your needs and here's some name for the project so let's say flooding for today let's go to the next step so the last step is to specify the data and now yes so for example for today we are going to use 701 and we are going to use the grd product you can also specify sentinel 2 and so on next thing will be to define our region of interest so for example for today we are focusing in Malawi the south of Malawi we can also specify the lat/long negatives longing comments or chronic cognates in case you know them and of course also upload a file with a study area finally define if we are going to perform multi temporal analysis or not now in case for today's at this size we are performing a meter table and Isis and we just specify the dates we will start using an image at the very end of November 2014 and the last image will be March 2015 so we specified that the period and then some more X details about what we want to do ok that will be all next thing to do review the defense we have specified and if we are we with that we just need to check the terms and conditions so here you can have a link to exactly the details about how whose work and what is our one for example personal data privacy etc and if you do agree submit the request and that's all in that way you have sent the request for your virtual machine and within a few days the roots team will come up to you they will communicate that your machine has been approved and it then will give you their source to login and then you can start to work with boots so let's have a look so in my case I'm not going to submit a request because I already have virtual machines so I will skip that but that will be the process so let's go back to your dashboard and now yes afternoon we have sent a request and after it has been approved we have here our projects in my case the name is restraining one so we have couple of different options we can the most important we can access the vector machine we can also freeze it in case we are not going to be working for a while we can even close the service before and something also relevant is that we can ask for a complimentary support so in case you know you have doubts about sending data in case you don't know how for example Sentinel one can help in your project or you have some taking of questions about the data you can come here click and submit a request so you explain us your doubt and then a team of remote sensing experts will come back to you with some help so as I said the rules also offers this service apart from the red phone machine so let's access our rental machine so that we can start our exercise so we click here in access my data machine and there we go we have these logging tapes so as I explained to you before once your request has been approved they will send you the credentials for this virtual machine so in an email you will receive the information and you just need to copy this data okay so we put the information here and we lock it ok there we go ok so we are into the filter machine here we are in our virtual machine and this is a vector machine based on Linux so this is a typical Linux environment so let me show you a little bit what you can expect so we have as in all the computers a file manager where we can store our our data we also have a dedicated internet browser in this case it's a 5x5 folks so well internet connection and we have as I said before a predefined list of source so here you can you can see some examples so we have snap QGIS or photo box with monteverde are etc of course in the beautiful machine you have full admin rights and that means that we can install whatever you want either open source software or commercial software of course for commercial options you will need to have your own license but you can do it as far as the subways let's say available for Linux you can do it and something also important to mention is your computer can interact with virtual machine that is you can upload files to the people machine you can download them from the rental machine to your computer and how to do that it's very easy we just need to press ctrl shift and then this menu appears you can change here some settings about the input and about the mouse and some other settings about the written machine but the relevant part is devices so we click here we can access the folders of the vector machine and for example if we go to a specific path of our interest we can download or upload files how to do that what to download we just need to select the file and double click and then a regular download will start and we have already about the final computer to upload files once we are in the in the path we want to stop that uploaded file we just pick upload file and then select for example PDF or or some data and then press open ok so that would be offer the intro to this virtual machine and let's now yes start our exercise so for our application today we are going to use for Sentinel 1 images one before the flooded land in December 2014 and three after the event in January February and March 2015 in that way we will analyze how the felt evolve during the time and how the water layer changes and for the analysis will use the either snaps over the central one tool box and finally note visualization we will be using QGIS so let's start and let's do it by downloading the images for that we open the browser and we go to the Copernicus open access hub so as I mentioned at the beginning central data is free for registered users so in case you don't have an account go to sign up fill in the information and you your account will be created in five minutes something very quick to do so if you already have the account we go directly to login and we specify our credentials ok so once we are in we change to pan mode so that we can navigate to the map and we go to our area of interest and today this is in the south of Africa and in the south of Milan so we zoom in a little bit ok so here is our study area next thing to do is to define that region for that we can either use box or polygon but ok today I'm going to use box I will just define the volume or the box yeah ready so next thing let's specify the data we want so we want to download Sentinel 1 data we want to do it from the end of December 2014 so December 2014 29 and the last image will be at the end of March 2015 so let's put this great so Center 1 data we want round range detected products that is GOP and because we know that from the orbits of the satellite and because we want to have the same geometry for all the images let's specify the relative orbit number which is 6 in our case okay that's all let's now search for the images and we do that by clicking on the search button great so here we have the list of products you can also see in the map the footprint of those images they're quite large you can expand the list to have a better look of what is available and for example some metadata about demons such as the mission the assassinate or the size of the file you can download so now we just need to identify the images we want so we want the one we want this one and of Mars so that is 23 of March we also want one at the end of February that is 27 of February the one before the flood the 29 of this December and then one right after the flood the 22 of generate so to download the images we just need to click on the download button here so you can see if I move to another project this how do you you can also have a look to the details of the image by clicking these I this will give you our metadata about the instrument the product etc so let's start the download process just backing it here so this pop-up window appears we say we simplify we click OK so the donald process is limited to two images each time so i'm going to start the process for two of them but i will not wait since I have already know the images for this exercise but the process will be to wait until the two images are download and then start again to others of course you can ultimate eyes that in other ways but ok for today we are living it like that so we have already our images Donald in the virtual machine now we can start our analysis after that let's open snap is not so forth it goes so first and if we're starting in case you are not familiar with snap the is the first lady up that you can see we have here the product Explorer section well at least it will appear as soon with as soon as we start to work with with the images we also have some options in the lower left corner such as quick ups of the image you will see there once we open it core manipulation to change the colors the volume to locate the product in the globe etc we also have the main area where the image will be displayed and of course we have dedicated tools for rust data optical images and right there images as the one we are using today so it's open the images and for that we can go to file open project or we can directly click on the open product icon so let's do like that we need to navigate to a path where we have store our images this is the one in my case and let's select all the images so in sync with something on one data there is no need to unzip the folder that you know load from the company goes have snap can work with that structure so we can select directly the zip folders and click open great so as soon as you do that here in the program Explorer we see the list of images and I mention you we have here the location in the world view and they all have the same geometry as you can see so in the product name you can see that the date appears after those key information so we have here 2015 o 227 so that's in it's for February and so on so let's open the image from January the one that that that is after the info and for that we expand the product we can click here in this little arrow and we see several folders so those folders contain metadata of the image big loops etc but the relevant one so far is the bands folder it is here when we have the image so we select bands and we expanded and we have both amplitude and intensity so let's double click on amplitude to Hobart and let's do the same for the image in December 2015 so that we see also how was the situation before the flood so with the same we expand the product we go to the band's folder and we open double-click on an tidal TV great so the images will start to they will they will appear they are a little bit big so it might take a while not that much but I think this is something that then you have to know something on the ice they are they're quite large so it takes a couple of few seconds ok so there we go here we have our image of January so we see their regular appearance of a sorry image white to black we have a very big dark region here in the upper right corner that is a lake but we want to focus in our flooded area and for that lets first synchronize the two images so that we can see it at the same time for that we go to windows tile horizontally and now you see that if I move around one of the images and for that I select the pan mode both views are synchronized and I want to zoom in into the flooded area and for that I use the linking I select the area and that's all so we can see here the difference between the image of January which is in the left side and the image of December which is in the right side so we see very big differences those dark regions are flooded area this is working and this is the information we want to extract from the center one image we understand that that layer that mask that water mask so that we can use it in our response that emergency great so let's do don't do that so as I said before we are going to perform the same analysis for the four images so we are going to precursors them in the same way and for that they are going to take advantage of the batch processing option of snap it allows this option to define our processing chain by creating a graph and then apply it to the set of eight so let's do that and let's close first of all those images and let's go to tools graph builder so here we are going to start to add operators and those operators will perform the pre-processing for us so we have first of all a read and write option and in between we are going to place all the steps so we are going to to start with a subset so we do not want to work with the whole image because it's pretty big we want to focus on our area and in that way by reducing the spatial extent of the image we are also improving processing time of further analysis so to include the sub celebrator we right click on the graph builder we go to add raster geometric subset that would be the first thing to do our next step will be applied orbit finally so we are doing a pan or B file because the orbit state vector is provided in the metadata of a certain product such a central one are generally not accurate and can be refined with the precise orbit files which are available based to make after the generation of the product the orbit files provide accurate satellite position and velocity information improving the analysis that requires satellite information to be as precise as possible so how to do that we right click again add radar apply orbit file great so our next step now will be thermal noise removal thermal noise inside imagery is the background energy that is generated by the receiver itself excuse the radar reflectivity towards higher values and hampers the position of radar effectivity estimates level one products such as the one we are using today provide a noise lookup table point measurement data set and that can be used to remove the noise from the product so it's another type of calibration of the image and for that again we go to right-click radar volumetric thermal noise removal so we place it here great so then our next step will be a basic step in some pre-processing which is called calibration so the objective force of our calibration is to provide imagery in which the pixel values can be directly related to the radar back scatter of the scene and calibrated sorry imagery is sufficient for qualitative use but calibrated sorry images are essential to quantitative use of some data so the power of the received radar signal also accounts for factors such as an antenna gain or system loss and this introduces a significant value metric bias in this argument even the metric calibration is done by calculating the Sigma naught coefficient which is also known as normalized rather cross-section courtesy and how to do that again very easy right click go to add radar volumetric titration great next thing in our pre-processing of the sentinel image will be to do speckle filter so speckle noise like feature is a common phenomenon in SAR system also known as the salt-and-pepper effect it confers to start images a granular aspect and random spatial variation and my decrease the utility officer imagined by reducing the ability to detect ground targets and obscuring the recognition of spatial patterns the source of this noise is attributed to random interface between the chorion returns and the principle of speckle filtering is to reduce the variance of the complex back on scattering and improve the estimate of the scattering coefficient okay so how to perform this step in snap again right click add radar speckle filtering specular filter not move the temporal because we are only working with one image great so that's spec of attorney and now we are ready for the last step of our pre-processing historian correction so white my correction well because of the side looking off of our systems every target located on the terrain being observed but the radar would be mapped onto the slant range domain and moreover due to topographical variations I've seen certain images are likely to be affected by geometric distortions sleigh over shadows etc so the terrain correction is the process by which hard data are converted from slant range to ground range geometry and in a defined cartographic system and that's relevant so let's add the operator again by right clicking add radar geometric terrain correction the recreation yes so we'll click and the operator wake up here here it is ok great so this is going to be the our pre-processing of Sentinel one images and now we just need to connect those operators so that the input so that the output of one is used as input for the next one and for that we right-click and we say connect graph ok so once we began we have those red arrows that are connecting the steps and now we can save the draft so for that we click on the Save button here and we go to to the path where we want to save it in our case it's called aux data and let's call it star pre-processing ok so we save it and now we can start our batch processing so we have defined the methodology to apply to the images now let's import the images and apply that methodology for that we are going to close the graph we are going to open the batch processing option of snap for that we go to tools batch processing just below graph build advanced processing and here we have the menu so first thing to do is to add the images for that we have two options either click here and navigate to the path but we have the images or click here when the images that are already in snap with the added so let's use this option and here we have the four images are that are in the product expert let's also refresh these images so that the metadata is displayed and we do that by clicking in the refresh button or icon here so now we see the type of the product appears the acquisition time the track orbit etc okay next thing to do is to load the graph the graph that will be previously saved so we go to load graph and we load the graph that is called sorry processing okay and now we see that in the upper part of batch processing we have all the operators that we have defined before that is great this is what we want so now before running we have to make sure that we are specifying correct for parameters for each operator and it's done starting by subset so we are going to subset the image using pixel coordinates today because we have all the images in the same geometry we are using the relative or the number six in case you don't have the same geometry in case you have images that are not in the same position you can use geographical geographic coordinates break it for today we are using this option and for that let's specify the numbers okay give me a signal be done okay let's now go to apply orbit file we don't need to change any setting here because the software will recognize the image directly and will download the precise information about the satellite position University so we we need that we don't need to change anything the next thing will be thermal noise removal here again no need to change anything the default values are okay the next step would be calibration so calibration as you can see the option of output Sigma 0 Sigma naught band is selected already so leave it as default next spec of filter in this case we are changing the filter we are not using this Sigma but delete and for that we will be using a window size of 7 by 7 ok and finally in terrain correction we just need to specify the map projection that we want so you can leave the rest of the parameters as default and to change the map projection we go to map projection we select here the option and let's specify UTM zone so for that we go we go down and we select UTM automatic so the software will check where the image is located and we'll match it with the appropriate UTM zone in our case if we click OK we see that it is zone 36 south ok great let me just check the subset columns I think I forgot yes ok let me just change those numbers very quickly ok ok and we put the width and the height ok so we are almost done we just need now to check that we are saving our results in an appropriate path and for that we go to the input output parameters tab and we check that the directory is correct in our case it is but just double check in case it is not and this is very important why because we are using the option key source product name so if you do if the output folder is the same as the input folder where the images are stored the products will be over bright over waiting and this is this is very very relevant be aware of that because you can lose your inputs by overwriting them so just check that you are not using the same destination folder okay you know what age we are so we are ready to go for today's exercise I'm not going to perform the analysis for all the images the person doesn't take that much I think it's max five minutes but for convenience I'm going to perform the analysis just on the images on the image of generator and for that I'm going to remove the other one so to remove an image we just click on the image and we select - and okay so butts processing is ideal for to process several images but okay in that way you know how to do it the opposite will be will be the same so let's click on run and let's run the process so it will not take that much although we have a lot of steps because it's not do not save each intermediate step so the only file let's say that will be created your rental machine will be the output of the whole graph and all the steps in between will be virtually self on the software end up going the process can be run much much faster as you can see great so it's done let's close batch processing and now we have already the image of January that has been processed you can see in the world view that the extent of the image number five it's smaller because we we use subset great so let's expand the product and let's go again to the Bands folder and let's open Sigma 0 Sigma naught V V which is the output of the calibration ok we double click and here we have B image ok it has been processed according to our chain and now let's zoom in a little bit to have a closer look so we can identify those dark regions those are the flooded areas after the event and now what we need to do is to create a water mask that we can use in our now in further analysis and how to do that what that's something easy and for today we are going to use as I said an easy but effective method known as by so further areas will have a lower back scatter coefficient than long flooded areas since they tend to create a smooth surface where the star sign up is reflected according to see to Snell's law and that is acting as a specular factor so such reflection returns very little sign of strength back to the satellite resulting in a dark region on the image and taking advantage of that we can classify the image as water not water by setting a threshold to create both classes we select to select a freshman we can use different methodologies there are very advanced ones that you can check on papers and literature today we are selecting that ratio by inspecting the pixel values of all dark areas in the image and then use it to create a water mask and to select that ratio we are also going to use the histogram of the image of the statistics so let's do it first of all let's zoom in a little bit more into a dark region for example here okay and let's now check the pixel values and how to do that it's again very easy we just go right next to the product Explorer tab we have pixel info so if we click here and now we move the mouse over that region we will see that the pixel values are shown in this area here we have Sigma's you'll be me and then we have a number of intensity so check that area if I move my mouse we have the pixel value so as you can see it's very low 0.005 okay say 67 very low body if I now go to a non dark area for example here this white one I have 0.2 or 0.14 I mean it changes but but we see already that there is a big difference in piece of bodies okay so that might be a little bit Brandon to select the freshen just by going over dark region so let's improve I am in this part and for that we are going to upload some polygons that will overlay on top of the dark regions and once we have those polygons we will the statistics of the pixel values below those polygons to have a better idea of which threshold we can say so let's upload those polygons it's not you can create the polygons by yourself we have here this option and you can create your polygon it's very easy okay today I'm I have a red nose polygons created so I will use them so for that we select the product number five and we go to vector import ESRI shape file we navigate to the path where we have the polygons in this case Oksana and we select the file go to polygon it's a shape file we click open and we select no in this popup menu let me go we have some polygons over the image make sure that you have the same projection both in the save file and the image just to avoid ever mind up cause a problem but it's always worth to be on the safe side so you see I have a couple of polygons here two others over here and one more here so now it's second is statistical except of the big ones that are below those polygons and for that we go to analysis statistics great now we need to select use region of interest we select this and we select that our region of interest is the shapefile that we have upload that is the word polygon shapefile we selected and don't forget to refresh because if not no so this no statistics will be shown great so now we have here two instagrams about the image we also have a table with the mean max Sigma median etc so in that way we can better characterize our dark regions and we can better choose fresher so today I'm going to use the maximum pixel value over bit of the pixels below my polygons and that is so now sorry so today I'm going to I'm going to check the mean value and window channel so Max and me and Beijing Dan I can decide my treasure so let's also have a look as I explained you before to the histogram of the image if we go the lower left corner and we select color manipulation we can see the histogram of the image first it's not very visible but don't worry we can stretch it for that we select stretch horizontally great and here we have our image our history I'm sorry so if we move a little bit we can see that by playing with the core display we can we can we can check that our threshold is working properly to differentiate between water areas and other areas so in that we can also be sure that the value that we are selecting is correct okay so for today's exercise I'm using a value of 0.02 and I'm going to create a condition based on that and how to do that well we got the number 2 the product number 5 we are going to create our photo mask based on this input and we right click on the product we select the ban math and we set a name for the new layer that we are going to create I'm gonna call it water mask ok we uncheck virtual because we want to save this layer we don't want to be we don't want to let it to be virtual and now we can define the expression for that we go to edit expression and we select and and we write if Sigma 0 within Sigma naught Vivienne is smaller than our threshold then assign to those pixels the value tube for example and if not let's assign no data to the rest so if your expression is correct you will have a confirmation message here in green ok no errors so we click OK and again we click OK and the pizza ban will be created automatically so as at the you can as you can see we can see still the polygons you can remove them from the layout by going to layer managers in the right side of the software of snap and unchecking vector ok so now that we have this water mask let's compare it to our original image the program 5 and instill that as in the beginning by going to windows tile horizontally great and now let's zoom to see how effective our threshold is okay so let's move a little bit and let's let's remove the polygon so here again the your manager and click vectors yeah that's zoom a little bit more ok so as you can see the white area here is the pixels classified as water and the gray or dark or black area is not data so you can see that it's quite precise we can see that for example even this area where there is no black in the original image it's not classified as water we can also see here the same so you of course you can move around the image and check the quality of them in some areas but ok we can already have an idea of how accurate our first holding our ionisation has been of course the nib analyzation is a let's say trial and error process you can improve it you can change the threshold a little bit increase it decrease it and this will definitely affect the water must have you're creating so that's something to play with and you can check that and how this influence for example the area the total area that I specify as water ok so we are happy enough with our water mask let's now visualize that in a better way in a better way using CGI yes and let's also visualize the water mask that I have been previously for the images of December February and March so as I told you things much processing and even possess those images because I've done it I've done that before so this process I've done it before for the rest of the images now let's go directly to see the results of the four images so the result of the multi temporal analysis for that let's minimize snap and it's open QGIS we double click on the icon and the software will open okay double so you have QGIS let's open then oh sorry I forgot to mention sorry for that once you are in snap and you know you you have your water mask of course you have to export it so that you can open it in QGIS that's something I forgot so once you are here you have to export that result and how do the die it's very easy we just select the photo mask and we go to file export and you can select a lot of different formats for example your teeth you can also in into this plane you can right click go to the bottom mask and select export view as image be sure that you select full scene full resolution that you select the attribute format that you want to use and then you are in the appropriate path for exam so that's processing okay and let's call it for example water mask gently and we say that okay so if file active pressure of this water mass will be created so as I said I've done that already for the image of December and February and March so let's have a look those results in QGIS now yes so we minimize nap and we open in a snap and we open QGIS and now let's open those master files that we have exported from snap so that we go to our our destination folder where we have saved the tips and there we go we have well damask January but also the one from December February and March so let's select all them and click open so here you have the four products there will be displayed right now ok and let's zoom a little bit to see how the further area evolves during the time so let me start by showing you the water mask created from in December so here we can see those blue areas are let's say permanent water bodies or water bodies that were present present at least in December so before the flood let's now see what the water masks in January the one we have done right now and let's see the result so you just put here and we see that the flooded area hasn't I mean the further areas is quite big compared to the original water bodies of December those wet areas are areas classified as watered by our methodology we can see that the affected area is quite a bit we can move around a little bit and see how this is affecting other areas let's now have a look to the image of March let's see how the film evolved in March for that we select go to mask March and we check it ok oh sorry sorry how much February February yes February is after-dinner yeah so we so here we have we can see if every factory is in yellow we can see that some areas are being flooded from the image in January to the image in February and that the total extension of the flood is still increasing a little bit of course not that much as in January a still some areas are being affected and now if which yes image from much let's check it and let's let me show you to your property and for that I'm going to compare it to February only to February and why well because in March we can see that the flooded area is already increasing so we have in yellow February and import from March you can see that the water masks identified by our methodology is not the same it's not expanding is actually doing the other thing so in that way we can see the evolution of this process and we can improve our response our emergency response etc so that's all for the analysis of the data of course of course after that you can still perform all the GIS analysis for example we can use open strip that make OpenStreetMap data to have some base map and contextualize a little bit more our results and for that in QGIS we have an ice cream cone openers plugging we go to web overlay just plug in OpenStreetMap OpenStreetMap and by clicking here this face map will be displayed and that's great because in that way we can locate for example some urban areas you will see it right now you can you can locate if that area is near urban area if there are some roads that are affected some railroads some fields etc and that's great of course you can hold your own fate files you can perform your own analysis it just offers around for example settlements to see how far what's the distance between a flooded area and what could be the risk is that area expands etc so you can perform a lot of analysis so from from here you can start and say more GIS things after you have done these remote sensing analysis of the star center one image so that is all for the exercise let me just highlight a couple of take-home messages that I think are relevant for for today and for that let me just go back to my presentation so in this exercise we have used 701 sub data to monitor a flood event in Malawi for that we have used the veto machines provided by the roost service and those vector machines come with pre-installed open source software ready to use and innovation to them the route service also offers to all registered users a dedicated happiness supported by a team of mod sensing experts to help you in your products with census data so let's imagine you are working with center one for flooded advocated for flood for flood mapping applications and you have some doubts work you can contact us and we'll be happy let me give you some extra information about our training activities we will have another webinar at the beginning of December but this time we are asking you to decide the topic so if you want to participate follow us on Twitter or create an account in case you can have it and vote of a poll that we have paid for this occasion the deadline is the 14th of November we have three options you can go the second choose the one you like the most and of course once the application of for the women are is open register for the webinar now in case you want to repeat this exercise by your own what would be the procedure but go to the roots training sorry who's Copernicus dot-eu registering the service apply for ventilation and as I showed you at the beginning in the request process you have to introduce the code so the code for this training is has a zero one hazard from hazards so you can specify the code a and in that way we know already the same thing all images you need the outside that you need and you will have also a step-by-step guide to do the analysis you of course next week the webinar is going it's been recorded so next week it will be available in our roots training page so you can also use the video as a step-by-step guide ba-lakay in case you prefer some paperwork you have it there so that's one from my side as I said my name is MIA Castro it has been a pleasure for me I really hope that you have done something new today thank you also to the rest of the roosting that has been involved in this well yeah quite a lot of work thank you again hope you have enjoyed this webinar and walk I'll see you in the next one so

Channel: RUS Copernicus Training

Views: 18,790

Rating: 5 out of 5

Keywords: flood mapping, sentinel-1, RUS, webinar

Id: ux4tx87ux94

Length: 61min 0sec (3660 seconds)

Published: Mon Nov 13 2017