Landsat 8: Estimating Land Surface Temperature Using ArcGIS

hi there Johnny we're kind of doing some remote sensing using ArcGIS today this video is a response to a couple of questions from my viewers with regards to how to estimate the land surface temperature using the Landsat 8 data so what we're going to be using today is the thermal band and NDVI for that study area I do have my in DVI calculated already if you do not know how to calculate NDVI you can watch a video of my title atmospheric correction and boundary Shannon and I have my band 11 which corresponds to the thermal band and Banten which also corresponds to the thermal band in different wavelengths and the first thing I did was I went to the USGS website actually to learn how to convert my digital numbers to radians which is the very first step to go and secondly convert my radiance to satellite brightness temperature using this formula so the K 2 and the K 1 the coefficient factors they're given in the metadata power so when you download your Landsat data you will have an MTL file which is the metadata file and every conversion factor number you need will be in that file and this is my hair and this formula gives you the brightness temperature that's been received by the satellite at the time that your image was taken so this is not really the temperature on Brown this is just the temperature at satellite so we're going to go two steps further after calculating this to derive the actual ground surface temperature it's going to be a fairly long video so suggest you go grab a coffee maybe so after doing this by the way this is gonna give us decrease in Kelvin degrees Kelvin and I want the result to be degree Celsius so I'm going to use a conversion factor which is two seventy two point one five to cover it from degrees Kelvin to degrees Celsius and after we do this we're going to go on to calculate the land surface temperature and this is the formula for that by the way there are several ways of calculating or estimating land surface temperature there are different algorithms to do that like the single window the split window and this is the one I'm using here which is I think it's the single window I think the formula says bheegi bheegi here stands for the outside light temperature which is what we are going to be calculating from this formula and divided by one which is a constant plus W here stands for the wavelength of immediate radiance at 11.5 wavelength which corresponds to ban ten in the Landsat 8 data and we're going to multiply that by the earth satellite temperature we're going to calculate and divided by P piece all of this story here blah blah and this is what the result had to fourteen thousand three hundred and eighty multiplied by the lock number of e here's the emissivity land surface emissivity people do calculate this using the land cover classification NDVI and I'm using the NDVI which is this one so this is the formula to calculate the lands of this emissivity and the formula is the square of NDVI minus n DV I mean divided by NDVI marks - NDP I mean and it's going to give me proportion of vegetation and we're gonna put that into this formula which is finally gonna give me the land surface emissivity which is gonna be then required for this unit here to calculate the land surface temperature so let's get started so back to our GIS most of the operations is going to be done using raster calculator my first step is to convert my digital numbers to radians for the thermal band so the first one is ml the other here stands for the multiplicative bond it's in the metadata file and the cue card is there troll digital number of the thermal band plus al al stands for the additive band here so I'm just gonna go to my metadata for my metadata the radiance multiplicative plan if you buy antennas this and I'm gonna multiply this by the actual digital number and add this number here 0.1 so back to my rest atop later it's zero point zero zero zero three three four two x like Batman which is the actual digital number plus zero point one and I'm just gonna rename these to bond 10 radians and I'm just gonna okay that and it's gonna give me a new raster layer for the radiance fourth band tan I'm just going to perform the same operation and defined 11 so since I wanted to just basically the same thing I can just go to geoprocessing resolves and it drinks of my previous executions raster calculator and I'm just gonna replace by the time we found 11 this factors are basically the same if you look at them in the metadata power and I'm just going to rename this to point eleven irradiance and I'm gonna okay that and once it's done I'm done with my first step which is conversion of digital numbers to radians and I can now go straight up to my second step which is converts a new radiance to not set light brightness temperature so K 2 divided by an odd number of K 1 divided by my radiance plus one and I'm gonna subtracted by two seven two point one five which is the conversion factor from Kelvin to Celsius so I'm just gonna look for the k1 indicator in my metadata file so this is my K - what about 10k - for biotin is this one I'm just going to copy and paste if you be careful while copying and pasting so you don't paste in the wrong thing end up my last tab later I came to divide it by the log number of k1 by k1 is this / my Valentine radians plus 1 I'm just gonna subtract 273.15 so I'm converting the decrease Kelvin to degrees Celsius I'm just gonna rename days to find sand SAP jump for 5/10 of like temperature and I'm going to okay that and hopefully I should get a new raster file with the actual temperature has been received at the satellite during the time of the image so my temperature ranges from twenty three point five to thirty nine point one I'm just going to repeat the same process for my band eleven so geoprocessing results and brings up my results again and my k1 and k2 they've changed for the bands and in band eleven so I'm just gonna do a new copy and paste the game so copy might take you five and eleven and replace it in the formula and I'm gonna double my K 1 what K 11 for my band 11 sorry and replace it in the formula and change the font and radiance to bomb 11 rating and this conversion factor remains the same and I'm just gonna rename it to band 11 satellite temperature I'm gonna okay that and here I now have the two satellite brightness temperature for both Banton and 511 some people just walk with them differently what I like to do is combine both of them in averaging so I use the average of the two bands and what I use for that is the cell statistics you just search for it it comes up and I'm just gonna drag in both of my satellite temperatures for banter and level and then every name its satellite 1011 and there are several overlay statistics here I'm gonna stick with the mean because I want to get the average and I'm just gonna do that and the other this is my satellite temperature data and I can go ahead and walk with these in several Anna Modlin and do all sort of spots with it my data ranges from twenty three point zero to thirty seven point nine three degrees Celsius what we're gonna do now is go a bit further to calculate the land surface emissivity and finally calculate the actual ground surface temperature for this image at the time that the satellite took this image so back to my formula again what we now don't have in this formula is the e which is the loss of its emissivity and this is the formula for it so we're going to be needing the NDVI to do this so I'm just going to go back to Pollock GIS scroll down to my NDVI dizelmine BVI I'm going to bring up the raster calculator so if you look at it is the square so first thing we want to do is bring up the square function and so square of the NDVI - NTV I mean and this is the NDP I mean so it's gonna be minus minus 0.05 which is going to be minus times minus plus so plus 0.05 eight seven three eight four divided by then it'll come to the formula again NDVI max minus M DV I mean which is at the NDVI max is zero point six so it's this so it's a zero point six zero eight seven five nine minus NDP I mean which is this I'm just going to copy and paste it to make sure it is when every name is to proper fetch for proportion of vegetation we're going to need it to calculate the actual land surface emissivity so think of the raster calculator one more time look at my formula copy and paste my formula just paste it in the rest of that later so here where it says TV TV stands for proportion of vegetations I'm going to multiply it by portion of vegetation and finally we're gonna name it l s e which is land surface emissivity and I'm gonna okay that so now I have my all my data I'm gonna need for this final formula which is this and this is gonna give us the land surface temperature so basically this last step I'm just gonna bring up algae is Rusted up later and it's going to be the satellite by a nice temperature and divided by 1 so it's at set my brightness temperature of the Bands hand one time supply temperature divided by 1 plus W which stands for wavelength of immediate radians are eleven point five ascetic corresponds to van ten so the actual content without any correction so van ten x brightness temperature again x by intense at high temperature divided by P which is this body for ten thousand three eighty divided by we see close it and multiplied by the lot number land surface emissivity lot number of lots of his emissivity so it's gonna be one plus one tan x I'm just gonna put in some parentheses here and this should give us the ground surface temperature for the time that this image was taken so I'm just gonna rename it van zan LST and I'm just gonna make sure everything is okay and finally we now have our Banton land surface temperature which is this so basically we are done just gonna give it a new color ramp from hot to cold cold to hot weather so my Banton has a temperature on the ground for the day that this image was taken which is probably trying this one elevation was about 49 degrees and the horizon so it's probably in the morning I really don't know to calculate this kind of stuff so in that's it you can look into the metadata file just to look at the time that the image was taken and all kind of stuff just to be sure of what data you're working with and you can repeat this process for the band 11 and just use the cell statistics that I've arranged to 110 and 111 for the length of this temperature and you can work with that I hope you enjoyed the video if you did comment or like thank you very much

Channel: Usman Buhari

Views: 113,483

Rating: 4.9349594 out of 5

Keywords: Remote Sensing, GIS, Geographic Information System (Industry)

Id: uDQo2a5e7dM

Length: 14min 28sec (868 seconds)

Published: Thu Feb 12 2015