Landuse classification of Sentinel 2A, L-8 & L-7 image in ArcGIS | A complete guide on LULC

hello everybody in this video we are going to learn how to classify different satellite images particularly sentinel to landsat 8 and let's hit seven image the reason behind these different landsat images is because people often get con confused when it's especially beginners when they move from one satellite image to another image that's why to clarify we are using three satellite images and in this video you'll also learn how to increase spatial resolution of lens 7 and landsat 8 image now take a look at this image here the top image shows landsat set landsat 7 image which have special resolution of 30 meter when the below one is 15 meter the way we do that is called pan sharpen and we will see that in the later part of this video how can i do that let's begin our process so at first to have a satellite image you absolutely obviously will have to download one so to download one head over to usgs website this is earthexplorerusgs.gov after that click here wired feature and country select your country or if you don't select it it doesn't matter name your location hit enter all right okay so this is your study area now here's set your date from which date date interval actually range now i [Music] want image from 2019 in between 2019 and 20. okay now remember to select cloud cover and i would prefer to give it 0 to 8 percent after that data sets now as i want a landsat image for 2019 so go to landsat then landsat collection one level one and i would like to download a landsat 8 image so select here and okay good result here you have different landsat images so select one from here before you download see the style if this tag covers your study area and you have to also see that inspect that this image is fresh and it match your desired date so it seems pretty good one no cloud so you can download it and make sure that before download you must be logged in if you don't have an account here make one and after that download it i have already downloaded it so after download they will provide you a zip file actually you will download a zip file so move that zip file into a folder and extract the zip file and then back to our arcgis interface let's begin our classification process with sentence 2 image so before we do that let's say something about sentinel 2 so you may have a better intuition about it centennial2 was launched in 2014 on april and it contains a total of 12 bands you can see here band 1 to 12. and notice here this resolution 60 meters 10 to 10 meter 20 meter and we will use only bands which have 10 meter resolution and why that so because if you use all of these bands to make a composite band then your image resolution will degrade to 60 meter but if you use only 10 meter resolution then it will be 10 meter and it will significantly affect your classification process let's see how now back to your arcgis i am showing you how okay that's it let's see this image it's in this composite band i all i have used only band two three four and this eight and let's see it natural color three two one if i zoom over here see it's pretty clear you can easily identify which area contains which object like this is urban area this is agricultural land and let's say it's resolution of this image under source tab you can see this cell size see it's 10 by 10 that means this image resolution is 10 meter and if you use all of those bands let's see how it starts out i have saved it yes that's it this one you see over here we can you can barely identify which area is actually a iron area this area is but it's not clear in this image you can use various band combinations like eight four three but it's still it's a little complex compared to the earlier one and let's see the resolution of this image see it's 60 by 60. so it will be it would be a wise decision to use only band 2 3 4 and eight so that's it remove it and now start from the beginning go to add data now navigate to the folder while you have stored your downloaded file satellite sentinel to [Music] okay so we'll only drag band to hold ctrl key and click three for edit that okay fine you don't have to worry about it because my study area is not in this region so it's not a matter not a problem okay now what we will do we will mix these four bands combine them and it's called now go to here search then write composite band hit enter here composite bands and data management tools okay input raster input your bands one by one four eight give me the location name okay now yes our composite band is successfully done so once your composite band is ready then delete this bands hold shift key click because we don't need them anymore now let's see it four three two it's something like what we see in line set 8 image when we have a composite 5 3 5 4 3 that's it see it's pretty clear now you can you also use natural color for three to one it's natural color natural color means actually how you see a satellite image from space or land surface from space you see over here it's a stadium okay so now i don't need this whole tile i'm just going to extract it from my ship for my area of into this so now i'm going to add my studio here it's my shape file that's it now go to search option here and write extract by mask and under special analyst click here again put raster it's our extra composite band then provide a steady area leave it in there okay now let's see our extracted file oh it doesn't matter there there it is so it's a good practice to export your data so what i did right click here go to data then export data here you can specify your location i'm just going to leave it as it is save it okay fine now we don't need them anymore now you set that image is ready so i am going to natural color image see over here this is called bear land bare soil this is and it's a river this white thing besides river is actually sediments particularly sand deposits okay now let's see this this is actually agricultural field it will be easier for you to identify if i change it to 4 3 2. see now here this dark red region is actually healthy trees or you can call it forested area forest and this light red color is agricultural croplands agricultural field actually this light red color and beside this light red color this thing is actually a fellow land that is once there was crop but now there is no crop yes now let's change it to four two and one now let's see it now agricultural land will appear in sort of blue color sorry yellowish color you can see over here this yellow color represents croplands agricultural fields this is also agricultural field but right now there is no problem and this bright red color is forest areas actually large trees okay this is everywhere fine so sometimes your image becomes somewhat dark or to increase its brightness but to make it more clear go to windows then image analysis tool now activate your layer here then check this dna too and after that uncheck it in my case it's okay [Music] see it's now a little different somewhat so now you might be wondering that how do i know that this area is a agricultural field a land what is this thing so to know it more clearly and accurately let's move on to google art pro now here you can easily search your location name but i'm going to the file then import i'm going to import my shape file here that it will be easier for me to navigate to trace my location easily so esri okay then navigate to the folder how you have stored your shape file start in here yes okay now field selective field fine save it yes that's it okay this is my study here then right click here go to properties after that style color and make it out filled okay so i said what is this right thing over here i would like to know that fine yeah just in this corner there is it yes maybe this is from this corner beside a diva yes so [Music] i think it's a break failed and you can see over here oh that's it it's a brake field beside that it's actually industrial zone and it's some whatever it bodies see this whatever is over here this this whatever is appeared here and we said this whatever day there they are say agricultural field let's see see over here yellowish color agricultural field or you can say 4 3 2 this light red color is agricultural fit so that's it one another beauty about google earth pro is you can change your date of this image by just click in this historical image tab and this is this image is 25 202 1 okay let's say you are you want to see this image for 2015 drag it and oh there's no image for 2015 one is for 2014. see how was this location in 2014 see in 2020 there's a new brick field but in 2014 there was no new no brake field but still this industrial maybe okay that's it for google earth pro now back to our arcgis now we want to begin our classification so to do that right click here and select this image classification tool and this window will pop up okay then again go to customize then extensions and make sure you have checked this a special analyst tool okay close it otherwise this window will not be activated fine now i am going to start my classification and for convenience and to save our time i am going to classify this whole area as one single bus that is in this area there is a fellow land and agricultural land that is crop crops and fellow land but but i am going to classify both of them as single class okay now go to training sample manager click here fine okay then here you can use different shapes i'm going to select polygon so i'm going to begin by selecting water bodies sorry zoom it then select smarter body know it more accurately yeah there's some problem okay i'm not going to select this water body because there's some mix mix up okay now this is also a water body and now i would like to delete this last sample okay [Music] okay enough whatever it is then select first one and hold shift click last one then here merge training samples select here click here give it a name whatever you can find then i'm going to give a new class this is my agricultural class okay this is also going to be agriculture field so so it's actually when you have a large study here then it's it will take some time and in classification the sample taking is the most tedious part because if you have a large area then you have to take sufficient amount of samples this is also a good culture except it's actually a fellow land but i am going to assign this area also a fellow land but i'm going to assign it as a as agricultural field okay that's it no no that's not it i need to take more samples here this file land i'm going to count it as as agricultural land it brings okay fine now trace that is forest uh um okay are going to provide more samples for forest okay so that's it for forest first now what left oh it's let's classify it here so okay so okay here's there's some built-up areas okay now classify it merge it and save whatever now last term this is sent okay this is send then after that save your signature file give it a name save it once it's saved then go to classification and select maximum likelihood classification this window will pop up here input raster bands okay then you input signature file select location and give it a name signature fine okay let's see how it turns out oh yes it's done so i think yes once you got your n16 make it browse sorry i get green okay forest 25 forest okay and water what is going to be blue that's it so sometimes some areas gets misclassified so you have to do sample you have to sample again again and keep take maximum number of samples and build up area 40 and to give it red but 32 sorry it's send send white send okay now 32 built up area red okay my build up area was actually misclassified i had to provide more samples on it and here in this area water body is a little misclassified so okay fine now over here you can see your specification how your classification was you see the more overlapped different color color patches are the more bad your classification was so here it's fine in band 1.4 but in this case it's get overlapped okay now we're going to properties and then sorry we're going to attribute devil we want to know area for each class so to know that click here add a new field add field name it area type double okay now here right click here go to field calculator yes double click on count before count it then again sorry go to properties and under source tab select copy cell size copy 10 okay and again come here and go to field calculator yes count multiply it by count size two times then close it in brackets and divide it by ten to the power six you want it in kilometer square okay fine our area is being counted calculated okay now i also want different classes so add another field name it to feature feature and text okay now i'm unable to write here so to write here go to editor and start editing now what was your value one from this one where value one was water doubling then after that 16th agriculture land agricultural then 25 it was forest and then we'll tap and finally send okay you don't need it okay so far so fine okay now okay that's it for sentinel 2 now we are going to see how lancer data works here you can see different bands of landsat 8 it contains a total of 11 bands and we will use one two seven band for making our composite band and lenses eight was launched in 2013 maybe okay here you can see different color combinations 432 being natural color color infrared 543s shortwave agricultural six five two that's it then back to our arcgis so i am not going to use the same things i did before i have already saved landsat 8 data landsat 8 composite file this is my landsat composite file now go to three four three two 42 its natural color image for landsat 8 now you can see over here it's 42 yes this is the urban area user immunization happened here and if you change it change it to seven five and three now it looks like this and in this color combination whatever bodies are spectacular it's very clear to see whatever it is let's move on to our previous study area now here you see this is our study area and sentence 2 was more clear than landsat 8 let's see it's special resolution it's 30 by 30 defenses 30 meter resolution okay that's it now you can see what the bird is very clearly this band combination is good for water bodies you can move to five four three it's actually very famous one and it's easy to identify forested areas and croplands in this main combination see over here the dark red color indicates forested area and this light red color indicates cropland these are agricultural plant see this is water this is built up area so i think it would be a wise decision to use sentinel to image rather than landsat 8 if you are using a class if you classify a image from 2014 to 2020 in between edit then it would be a good decision to use sentinel 2 i think so now back to our natural color image 432 let's see barren soil over here you can see this is a barren soil this is also a bad side if we zoom over here you can also see here metal soil this is that original nicest yes that's it this is a bad sign now we are going to see how to pen sharpen or increase resolution of this landsat 8 image so to do that go to the folder where you have stored your landsat 8 image and we will only add band 8 that is pan chromatic band right now go to windows originalizes and click your activate your composite and pancreatic band both of them after that here click this pen sharpen tool there you go now see your pancreatic band is ready now four three two natural image zoom over here let's see you can see over here it increased from 30 meter resolution to 15 meter resolution and here it is bare soil mainland this is bare land let's see is it really increased the resolution good properties and see source say effectively now let's use the swipe to okay let's see see there it was before now so that's it ben sharp and do it will increase your let's see here yes it quite successfully increased our special resolution it will help you to identify different objects more easily well now that brings us to the final step now we are going to classify or see how to use landsat 7 image so go to transit 7 data and select band 1 2 3 4 5 6. seven okay we'll use only seven bands then add it i'm not going to edit because i have just done it so then go to composite composite bands compositive composite seven bands so i'm not going to do that i have already done it all right okay for lenses seven that's it so for landsat 7 natural color image is 3 2 and 1. you can see this is the image of 2001. it's not that clear and that organization was not that so sorry let's see this table here the difference between lens 7 and 8 is in landsat 8 first band is ultra blue and in 97 there is no ultra blue band that's the difference then in landsat 8 number 2 is blue then green red and same in lens 7. till band-aid in landsat 8 number 8 is pan-chromatic band and in lens seven it also contains pancreatic band in last in the last event here now take a look at this at this picture here you can see that for which color combinations landsat 7 and landsat 8 will give same colored image see for landsat 85 for three and lands seven it is four three two that is red color image colored in from it or false color image okay back to our arcgis zoom to the layer now it's master area back to 432 let's see this is the urban area in 2001 it was not that much here it's and you can see over here this is agricultural land this is agricultural land if you change it to seven four two you can see over here little green patches here are actually referring to agricultural land so now you can go to the sample manager and again as we did before you can connect sample and run the classification so that's it for now i think you found this video really helpful and you learned learned something new so if you have any questions or suggestions let me know in the down in the comment section below i'd love to hear from you and until next time i hope i'll see you in another video

Channel: Data Analytics.m

Views: 2,041

Rating: 5 out of 5

Keywords: Arcgis, classification, landuse classification, sentinel 2, landsat 8, landsat 7, LULC, lulc, land use classification with arcgis, Supervised classification, landuse classification in arcgis, landcover, landcover classification, land-use clasification

Id: Kz_nsb2n26A

Length: 58min 49sec (3529 seconds)

Published: Wed Nov 25 2020