Estimation of Land Surface Temperature with Landsat and ASTER

you hi I'm Luca congedo and you're watching fringe is to remote sense this tutorial is about the estimation of the land surface temperature starting from a land cover classification in calculating surface emissivity so in this tutorial we are going to use a Landsat image we can download the Landsat images so in the SCP menu select download images download Landsat in this interface enter the user and password of the service in this case however we are going to use a Landsat images acquired in 2016 which is available from the Amazon service which doesn't require a password so here enter the search area that Left coordinates 2 for longitude 49 for latitude a lower right corner 2.5 longitude and 48.8 for latitude here we can set the satellite so we are searching a Landsat 8 satellite but we could also search for other Landsat satellites and we can set the acquisition date so here type 2015 927 because the last image that we are going to use a was acquired at 27th of September 2015 then click this button fine and after a few seconds we have the image list the year in the image list select this image can click this button to display the image preview here we have the image preview sometimes the color are not correctly displayed so just click this button here QJ s to display the image correctly with a proper stretching and in the interface click download options saying download options we're going to select the bands that we are downloading so uncheck the one we leave these bands checked we uncheck been date panchromatic ban benign we leave Benton which is the thermal band we uncheck band 11 and band QA we are also going to set these options here in particular we are going to use the process images when this checkbox for process image is checked after the download of the lancet band will be converted to reflectance using these settings so we are going to check here apply this one atmospheric correction and we uncheck create band set because we are going to create the band set later Landsat band the 10 which is the thermal infrared will be converted to brightness temperature so now in the Lancer download we can click this button run and select an output where we are going to save our Landsat images and when the download is finished we have here in a lot integer yes The Lancet bans automatically corrected and converted to reflectance so now we are going to clip the last image to our study area which is the area of Paris in France so in the SCP menu select clip multiple rosters click this button to refresh the layer list and select all the bands loaded in QJ s so we are going to clip these Lancet images using the coordinates click this button and with a left click and right click in the map we are going to set the clip area you can see here the coordinates however we are going to set manually also the coordinates here so for the upper left and lower right corners enter these values as described in the tutorial and now we can click run and select the output when the process is finished we have all these clip bands loading huge AES of these bands starting with the clip we can select and remove all these lengths of the images because we are going to use only these rosters that were equipped now we are going to create a link our classification so select the bands the clip bands that will be our input for the land cover classification so we're going to set the band set here we are going to use only the visible and infrared band so remove the band tan and set the center wavelength selecting Landsat 8 in the quick world and settings so this is the input of the land cover classification here you can see that band set is defined as input image so if we now select an RGB 4 3 2 color composite we can see here this color composite created on the fly we can see your vegetation red and the urban area in blue in the center of the image so now we are going to create the training input so click this button here and define the training file so here we are going to create several rice for creating a link our classification using the SCP tools so for instance we can create here roy for the built-up area particularly we are going to use this for land cover classes water the top vegetation under soil so after the collection of several and Roy's and of course the calculation of the respective the spectral signatures can create a preview in particular we are going to use the maximum likelihood algorithm and select the macro class ad for diversification when we are satisfied of the result we can click this button run and select the output destination for the classification raster click Save so after the calculation we have here the land cover classification of the study area according to these four land cover classes and now we are going to reclassify the land cover classes to the emissivity values so in the ACP menu select was processing and rectification tool and this tool click this button to refresh the layer list and select the classification raster has input next click this button to add the four values for rectification values and enter as old value the macro class ID so one two three and four a new value we are going to set the emissivity values so here these are the old values according to the macro class ID here as you can see these are the old values hang the new values we're going to enter dem civet e values according to the land cover class so 0.98 for water 0.94 for the top 0.98 for vegetation and 0.93 for best soil next and check use code from signatories and click this button run to select the output of the emissivity raster so set the name for instance emissivity and click Save after a few seconds then see the raster will be calculated as you can see this is the rectification of the our land cover classification now we are going to calculate the surface temperature using the emissivity raster so click this button and open the Ben calc tool of the plugin so we are going to use this expression here where surface temperature is a function of the brightness temperature and emissivity so here in the Ben calc expression click the name of the Benton raster which is the thermal infrared band of the Landsat 8 image divided by open parenthesis 1 plus open another parenthesis 10.8 which is the center wavelength of the band 10 of the Landsat Ben 10 times the length Ben 10 divided by fourteen thousand 388 close parenthesis times a logarithm of the emissivity you and close the parentheses so when this expression is green we can click run and select the output destination of the surface temperature raster and click Save after the calculation we have here the surface temperature calculated in Kelvin so we can also convert the surface temperature from Kelvin to Celsius so in the bank out click this button to refresh the layer list and select the surface temperature minus 273.15 and click run so we're going to calculate this new raster where temperatures is calculated in Celsius so here we can define a symbology for our surface temperature raster because if I hear the intervals and here we have the surface temperature you can see here the lowest temperatures are over the vegetation light areas while the highest temperatures are over built up in urban areas okay so we have calculated the surface temperature of last image we can do the same using a pastor image which is now freely available so open the hustler download tool and enter your the username and password of the NASA account which is required in this case so pick this button then in the map left click and right click for defining the search area then the search parameters enter the acquisition date from and to enter 2008 24 because we are going to use an image acquired on the 24th of August of 2000 then click this button find find images and after few seconds we have your the image which was acquired in 2000 so we're going to download this image select this image and click this button to display the image preview so here we can see the Aster image now in the interface of the plug-in select pre-processing Astor because we are going to set the parameters for the automatic conversion after the download so in particular check apply this 1 atmospheric correction we can leave this time I checked create band set because we are going to use disbands tanning the download Astor tab click run and select the output the destination we can see that all these bands were automatically loaded in the band set but now we are going to clip this band because as you can see here these bands have a little shift one from the other so here you can see not all the bands are perfectly aligned and we need to clip all the bands into the same surface so we can use the tool clip multiple raster here so here in this tool refresh the layer list and select the downloaded and converted the Aster bands here and then we can use this option here use temporary dry for clipping so in the tool bar click this button to define ROI and draw a ROI which is slightly smaller than the image area so that all the pixels will be aligned then in the clip multiple rosters to click run and select the output destination after the process we have here of the eclipse bands as you can see here loaded in QGIS and all these bands are now aligned and ready for the classification so now we're going to use this clipped bands as input for the land cover classification of the aster image so we are we now clear this old band set and select the aster bands that we have clipped so here we check every aster band but we're not going to use the thermal infrared bands so select these bands from 1 to 9 and set the center wavelength of the bands so here we can see in the color composite of the faster image we can see here that the color composite 3 to 1 represents the false-color composite with the near-infrared which is the band 3 next we create the training input file so for instance trying aster and click Save so now as we did for the Landsat image we're going to create several roy's so now we should delete the old spectral signatures of the Landsat image and we are going to create several rise for the same macro classes here so for instance one ROI for water using the original tool here we can set the map class the class ID as you can see the macro class information is automatically loaded and after the creation of several rows here of course the same macro classes we are going to create a land cover classification using the macro cassadee and the maximum likelihood algorithm we can see here a preview of the classification so when the preview is fine we can create the classification raster in this case enter the output name so as we did for Landsat we have now the land cover classification we are now going to reclassify the land cover classes to the emissivity values we have here the same length cover classes in the same emissivity values that we use for Landsat as you can see here these are the values for the macro classes and now we can click run and select the output raster so for instance emissivity pasture in click Save after a calculation we have your name CVT raster so now we have the emissivity raster and we are going to calculate the surface temperature so click this button to refresh the layer list and now we are going to enter the expression for the conversion of the brightness temperature to the surface temperature so we are going to use the bend 13 so double click this layer divided by open parenthesis 1 plus open parenthesis 10 point 6 which is the center wavelength times the band 13 which is the term our infrared band converted the two brightness temperature divided by fourteen thousand 388 close parentheses times the logarithm of the emissivity and cause parentheses so now the discretion is ring we can click run and select the output destination which is the surface temperature faster click Save so now we have the surface temperature we can as we did for the Lancet we can convert the temperature from Kelvin to Celsius so click this button the ban calc to refresh the banlist and I put surface temperature minus 273.15 and select the output destination so now we have calculated the surface temperature in Celsius we can define a symbology you can set for instance the symbology here with eleven classes so we can see here the values the temperature values of disaster image again we can see that the highest temperatures are over built up and urban areas you can see here the lowest temperatures are over vegetation and water bodies of course the accuracy of estimation relies on the values of emissivity that we have defined here so for better results we would need field measurements of the emissivity values for the different land cover classes we can for instance compare the surface temperature of the Landsat image with the surface temperature of the aster image you can see here the differences in the range of temperatures the differences are caused by the different years of acquisition but also the different seasons and then color change so well done we have calculated the surface temperature during the next few weeks I'm going to upload other new tutorials and please join the Facebook group and the Google+ community thank you for watching you

Channel: Luca Congedo

Views: 30,311

Rating: 4.9597316 out of 5

Keywords: QGIS, Supervised Classification, Land Cover, Tutorial, Landsat, ASTER

Id: 7W4IwlvPLbQ

Length: 26min 33sec (1593 seconds)

Published: Sun Sep 18 2016