 Hello everyone, welcome to the next lecture in the course remote sensing principles and applications. In the last lecture we started discussing about the concept of spectral indices and in this lecture we are going to look at some of the few commonly used spectral indices and their applications. So I told you what a spectral index is in the last class. So just as a quick recap a spectral index is a single numerical value obtained by combined the spectral reflectance values in more than 2 or more than 2 bands for a certain feature. And the main aim of creating such spectral indices are first thing to highlight some important biophysical property about the object of our interest and also to reduce or to completely remove some other unwanted effects such as effects due to topography, atmosphere correction, soil background effect etc. And also in the last class I told you about earliest developed spectral index known as the ratio index and also some of the limitation associated with this. So today we will start looking at the other spectral indices that exist for monitoring different features on the earth surface. The major application of remote sensing even from its earliest days was to monitor vegetation. Monitoring how healthy vegetation is or monitoring whether the vegetation is undergoing any sort of stress is one of the primary application for which remote sensing used still even today because monitoring vegetation has lot of implications starting from understanding the global carbon cycle, global vegetation cycle or if you monitor crops then it will help us to ensure food security, take precautionary measures and so on. So vegetation in a single domain encompasses lot of different species starting from large dins, rainforests to crops that we use for our daily food needs. Since vegetation is one of the most important feature on the earth surface and the primary application of remote sensing is to monitor vegetation, the other indices or the earlier indices that were developed primarily concentrated on understanding the healthiness of vegetation. One such commonly used and most useful spectral index or vegetation index is normalized difference vegetation index NDVI. So the equation for NDVI looks something like this. In order to obtain NDVI, we need to have the spectral reflectance in NIR band and red band. So same two combinations we used for creating the ratio spectral index. But instead of just creating a ratio of rho NIR by rho red, here we are devising a slightly different way of equations. So it is on the numerator we have the difference in reflectance NIR minus red divided by NIR plus red. So this is like a slightly modified representation of the ratio index. So what it helps? So we will look at like the name normalized difference vegetation index. It is a vegetation index primarily developed to monitor vegetation that we can understand. Difference we are taking a difference of spectral reflectance in two bands NIR and red. So that is why it is called difference. Then normalized here normalized means we are changing the reflectance values and bringing them to one single range that is the physical range of NDVI will vary from minus 1 to plus 1. It is now bounded on both the sides. On the lower end it is minus 1, on the upper end it is plus 1. So by normalizing here we mean that whatever be the reflectance value in the red NIR band, we are modifying them and bounding them such that the final result will be one single number within this particular range. So that is the reason for the name normalized difference vegetation index normalized. We are normalizing the reflectance with respect to the total value of NIR plus red reflectance. So whatever be the difference we are normalizing it with respect to the sum of reflectance. On the numerator we have the difference, on the denominator we have the sum of reflectances. So the term on the numerator is getting normalized and the range is becoming bounded. So that is why it is called normalized. Since we are taking difference in two bands, we are calling it difference and then this particular index is primarily developed for monitoring vegetation. So normalized difference vegetation index and again the application is more to monitor the healthiness of vegetation, what we call vegetation health or vegetation vigor. So the same concept applies here to similar to that of ratio index that is for a healthy vegetation NIR will be having very high reflectance, red will be having very low reflectance and when vegetation starts to undergo any sort of stress red reflectance will first increase and finally after like long prolonged period of stress NIR reflectance also will come down. So for a healthy vegetation NDVI values will be in the positive side also like in the higher range maybe like more than 0.5 or 0.6. Say if you take one pixel in a remote sensing image, if the pixel is full of healthy vegetation the NDVI of the pixel maybe more than say 0.5, 0.6 even sometimes for some sensors like modus NDVI can be like 0.8 or something. So, higher the value in the positive domain indicates healthy vegetation. Similarly for land surfaces whatever be the soil urban settlements or vegetation etc. In general for land reflectance in NIR tends to be slightly higher than reflectance in red band for most of the commonly occurring earth surface features and hence for normal surfaces NDVI value will be positive more than 1. Only for surfaces for which NIR will be reflectance in NIR will be less than reflectance in red say water bodies. For water bodies clear deep water bodies the reflectance in red band will be slightly higher than the reflectance in red NIR band. Similarly for snow reflectance in red band will be higher than reflectance in NIR band. For such features NIR will be negative. So, negative NDVI most likely will indicate water bodies most likely and also lower values of positive NDVI that is say between 0 to 0.2 or 0.25 typically indicates bare and dry soil or even sometimes wet soil also but bare soil in general. For soil based on the moisture content within the soil the values of NDVI may be changing from 0 to 0.2 or 0.25 and also pixels contaminated with cloud covered that is let us say one single pixel is here some feature is there let us say some portion of the pixel is covered by cloud. So, this is pixel number one there is another pixel two some other pixel. So, a single cloud is kind of distributed across four different pixels. So, due to the presence of cloud here the reflectance as I said the reflectance between land surface here and the cloud will be totally different and for such mixed pixels the NDVI value will be either negative or in the range of 0 to 0.2 on the positive side. So, by looking at the NDVI number or the value of NDVI we will be in a position to tell at least to some extent the condition of vegetation in that particular pixel or that particular area and NDVI is as I said before is one of the most widely used spectral index to monitor vegetation and due to its high correlation with vegetation health NDVI has been used in several applications related to vegetation monitoring that is starting from understanding vegetation cycle seasonal cycle of vegetation getting crop yield monitoring vegetation stress etc. For various applications NDVI is being used and there are like thousands of published journal articles are available which will tell us the use of NDVI and because NDVI is one of the one of the index with very wide application most of the remote sensing agencies produce NDVI values or NDVI images on an operational basis. Say if we take example of modus sensor from modus sensor we will be getting daily values of NDVI eight-day aggregates of NDVI monthly values of NDVI at different spatial resolutions etc. So, these products are produced operationally that is first they will take the satellite DN convert into radiance do some sort of atmospheric correction and then finally calculate NDVI and give it to us. So, as a end user we can just directly download the data and use it. Similarly, even if you take like Indian sensors NDVI products are commonly available for operational on operational basis for us to download both from satellites in geostationary orbit as well as in near polar orbits. So, it is one of the most widely used and widely produced product also in remote sensing. Then the next index what we are going to see is known as the normalized difference moisture index or otherwise known as normalized difference water index. So, the NDVI that we have seen is highly correlated with vegetation health like it is related to NAR and red reflectance and while we discussed the spectral reflectance property of vegetation we came to know that the reflectance in visible bands green red and blue is highly controlled by chlorophyll absorption. So, any change in chlorophyll activity or the amount of chlorophyll present in the vegetation will change the red reflectance which will in turn change the NDVI value. So, NDVI is highly correlated with the chlorophyll amount of chlorophyll and how healthy it is because we know that for a change in NAR to occur the cells internal structure itself should be modified or destroyed then only reflectance NAR will drastically come down. So, that will take a long time but first thing whenever like vegetation ought to go some sort of stress after like say few days of stress occurrence maybe the chlorophyll activity may decrease or the amount of chlorophyll may decrease photosynthesis will go down all these things will happen which will eventually increase the red reflectance right we have seen it. So, most likely any change in NDVI value will first tell us some information about the chlorophyll activity present in the leaf but let us say some water stress is occurring today. So, for this water stress to translate into a change in the red reflectance or a change in the chlorophyll activity it may take at least some time maybe in order of few days for this water stress to have to be persistent and then only there will be a change in the chlorophyll activity. So, there is always like some sort of time lag between stress occurrence and its effect on the chlorophyll concentration within the leaf. But let us say we want to monitor the water stress in vegetation immediately and we know that shortwave infrared band SWIR is one of the is one of the band where any change in leaf water content will be immediately reflected as change in reflectance higher the water content lower will be the reflectance. So, taking this particular clue in the year 1996 GAO developed a new index called NDWA normalized difference water index or also otherwise known as NDMI normalized difference moisture index the equation for this particular index looks something lies NDWA is equal to NIR minus SWIR divided by NIR plus SWIR sorry this should be NIR there is a small mistake in the slide. So, NIR minus SWIR divided by NIR plus SWIR. So, here we are replacing the red band with the reflectance in SWIR band width. So, here we are replacing the NIR band sorry the red band used in NDVI with the SWIR band. So, what will be the use of it? So, again we know that for a healthy leaf and that is with high water content we call that leaf as turgid. So, if the leaf is healthy and turgid its NIR reflectance will be pretty high its SWIR reflectance will be pretty low and when we combine these two as a difference ratio then the NDWI value will be high. Similarly, as the water content in the leaf decreases then the reflectance in SWIR will increase leading to a decrease in NDWI. So, this suggests for a healthy and turgid vegetation the NDWI value will be quite high and as the water content in the leaf decreases the reflectance in SWIR will increase leading to a decrease in NDWI. This also will vary between minus 1 to plus 1 because again we are bounding it and higher values of NDWI signifies healthy as well as vegetation with high turgidity that is with high water content. So, this is a indicator of vegetation water content and to prove that this is in addition to NDWI if we use it we will get more information about leaf water content. So, when this was developed the author who developed the said it is not a replacement for NDWI, but it can act or add along with NDWI to provide more information about the vegetation and it can in compared with NDWI it can quickly respond to vegetation water stress. As soon as vegetation water stress occurs the reflectance in SWIR band will change leading to a change in NDWI. So, it can help to identify water stress pretty quickly in comparison with NDWI. So, even this is not like one of the most widely developed or used product even though it has like lot of potential applications most of the sensors which was launched sometime back they did not have this SWIR band most of them concentrated on visible and NIR bands. Only recently in the last few years almost all the sensors are equipped with SWIR band. So, due to this factor this has not gained this particular index NDMI or NDWI has not gained that popularity, but it has lot of potential in monitoring vegetation and even though it is not produced operationally still we can do it because most of the satellite satellite agencies give us processed surface reflectance value like modus from modus we can get surface reflectance from Landsat we can get surface reflectance. And once having we have the surface reflectance we can just take the reflectance values from the corresponding bands put it in this equation and calculate NDWI and use it for our applications. So, this is again one of the index which is not widely used, but the same time has lot of potential applications in vegetation monitoring. The definition of NDWI normalized difference water indexes prone to some sort of confusion like in the same year 1996 one more definition of NDWA was given. So, that definition read something like this NDWA is equal to green minus NIR by green plus NIR. So, the primary aim of giving this definition is to identify open water bodies like yesterday or in the previous lectures when we discussed about the spectral reflectance nature of water bodies. I told you that NIR band is most suitable for identifying the border between land and water bodies. Water absorbs almost all NIR wavelength land will reflect good amount of NIR wavelength. At the same time green will be slightly reflected by water bodies which means green reflectance will be higher NIR reflectance will be lower. Taking this particular feature or particular characteristic make features develop this particular index for mapping or identifying open water bodies in a remote sensing image. So, this is also unfortunately has been given the name of NDWA normally is difference water index. Similarly, when we develop the vegetation index you also name that as NDWA. So, this is causing potential confusion even today even like if you look at some papers the same term NDWA may be used in one paper to denote open water body and in one paper to denote vegetation liquid water content. So, it is a cause for potential confusion. So, we always have to ensure caution when we read some terms such as NDWA and we always have to look for the equation for which it is being used. If it contains green and NIR band or a modified definition green and SWIR band if it is there then the primary use is for monitoring or mapping open water bodies. But if it uses NIR and SWIR bands in combination then it is for monitoring vegetation water content. So, this is a place where which potentially we can get confused and we should always ensure caution when we see this term NDWA and when we use the equation. We should be really clear are we going to use this index for mapping or monitoring vegetation water content or are we going to use it for identifying open water bodies. With our applications the definition of NDWA will change that is the reason why in the previous slide I coined two terms NDWA or NDMI. So, in order to avoid confusion some authors in the remote sensing domain prefers to use the term NDMI normalised difference moisture index for the one used for monitoring vegetation water content that is NIR and SWIR combination. So, potentially you can avoid combination by calling that index vegetation index as NDMI. But the original name given by the authors who developed it is NDWA for both the indices for both monitoring water bodies and also for monitoring vegetation liquid water content. So, just one more definition of this NDWA for water bodies MNDWA exist. So, this is combination of green and SWIR bands. So, these two indices combining green and NIR or green and SWIR. So, this particular combination ratio is helpful for identifying open water bodies and this particular index has nothing to do with monitoring vegetation or vegetation water content. We just saw that NDWA is one of the most widely used index and is used in various applications related to vegetation and its monitoring. But is NDWA is only controlled by vegetation chlorophyll activity or will something else will change the values. Two important factors we need to consider when we use NDWA for different applications. The first thing is the soil background effect. When we studied about the spectral effectiveness property of vegetation, I told you that when a remote sensing sensors observes vegetation, it will not only see the leaf, it will also see other non leafy parts of a tree like stems, branches, flowers, etc. And in addition, it will also see the soil and we will get a integrated signal of all these things in a remote sensing image. This will cause a major change in the NDVI value that we calculate from the images. So, a very good example of influence of soil is given in this particular slide. So, here if you see on the x-axis here, percentage of green cover in a particular plot is given. So, what this is what fraction of particular pixel is covered by vegetation that is given by the percentage green cover. On the y-axis we have red reflectance, reflectance in red band. We know that for a very healthy vegetation, reflectance in red will be quite low. But if the pixel has mix of vegetation and soil then the reflectance may be higher because soil reflectance in red band is generally higher than reflectance from vegetation in the same red band. So, when we have a mix of soil and vegetation, then definitely the integrated effect will be seen. So, 0 here, 0 percent green cover means it is like bare soil, 100 percent green cover means the entire pixel is covered with vegetation, we are not able to see any soil in between. If that is the case, you can see like red reflectance is quite high for 0 percent green cover and then it decreases and only after crossing say 60 or 70 percent vegetation cover everything converges. So, the difference between different different fields or different different moisture content. So, each line here corresponds to different crop type or different soil type with different moisture content. So, there are a lot of differences in this. So, you can see just due to differences in soil properties like soil, we have seen that soil has lot more lot other factors will influence soil reflectance including moisture, its chemical composition, etc. etc. So, the difference in soil types and the moisture content will influence the red reflectance and this effect will be quite large when the pixel is composed of both soil and vegetation. That is a safe example, let us take these two curves. Here let us assume like one curve here and one curve here. Here let us assume the major difference between the curves due to the or let us assume the vegetation is exactly one and the same and almost in same condition. But just because of soil effect, background soil effect and the change in moisture content of soil this change in reflectance would have occurred. So, similarly in NAR band if we observe the same thing will happen for healthy vegetation NAR reflectance is extremely high even higher than soil. So, here the trend of this curve is in opposite direction towards red reflectance. So, until we cross that 80 percent or some very high percentage of green covered we are seeing a large effect of soil reflectance here basically. So, these two curves tell us in a mixed pixel. Mixed pixel mean a pixel which contains both vegetation and soil. The influence of soil background will be quite high in changing the NDVA value. So, the basic reflectance of soil the presence of any water in the soil all these things will change reflectance and hence the NDVA value to a significant extent. And this effect is highly pronounced in mixed pixels where soil and vegetation are interwind with each other. Only for when the vegetation cover becomes extremely high in a pixel more than 80 percent in the pixel is vegetation cover then the background effect will reduce we will get signals from pure vegetation. So, removing this soil background effect is one of the primary needs that was identified in the earlier days. So, in the year 1998 Alfredo Huit came up with a new index what is known as a soil adjusted vegetation index SABI. The main aim of developing this particular index is to remove the effect of or at least reduce the effect of soil background effect from vegetation. So, in the original paper that was like a clear explanation and a simple demonstration of how this index being derived from NDVI. It is actually like almost like a modified from NDVI only, but we are not going into the finer details about how this is derived, but the general equation of SAVI is something like this. So, SAVI is equal to very similar to NDVI, but here you have two terms 1 plus L and L where the L takes different values based on vegetation cover. So, L will be 0.25 for high vegetation cover and L will be 1 for low vegetation cover and it will be 0.5 if we do not have any other information or for moderate vegetation cover. So, the introduction of this particular L term was able to reduce or remove the effect of background soil effect in NDVI. So, it is essentially SAVI is nothing but NDVI with reduced soil background effect. So, now we may not know when we want to work at like some unknown remote areas for our applications for our projects, we may not know about the nature of vegetation present there. So, in general the threshold commonly used is 0.5. So, we can like NIR minus red divided by NIR plus red plus 0.5 into 1.5. So, this will give us the information about SAVI which will remove or at least reduce the effect of background soil. Apart from this atmosphere also plays a major role in changing the NDVI. Like when we discussed the effect of atmosphere and spectral reflectance we have seen atmosphere in many different ways can change the reflectance values or radiance reaching the sensor. So, if we do not do proper atmospheric correction while we derive the surface reflectance and then we calculate NDVI then that final derived NDVI will change a lot. So, an example for this is given in this particular slide. So, here on the we have what is known as a histogram plot for NDVI values. So, what this histogram plot says the histogram plot is nothing but say we have NDVI values ranging from minus 1 to plus 1. We divide them into different bins say minus 1 to minus 0.9, minus 0.9 to 0.8 and so on and then 0 to 0.1, 0.1 to 0.2 etc. We divide it into different bins of say 0.1 range. Within the each bin we will count within a given image within a each bin how many pixels are there say 1000 pixels are there between values 0 to 0.1, 500 pixels are there in the range between minus sorry 0 to minus 0.1 and so on. So, here we will plot each bin on the x axis and the count of the pixels in the y axis. We may plot the count or also divide the plot the ratio or the percentage that is the count divided by total pixels in the image. Either we may plot this ratio or we may just plot the count. So, if we plot them like the bin values and the count we will get a figure what is known as a histogram. So, the histogram will tell us about the basic distribution of values how the values are distributed within any given image. So, in image processing of remotely sensed data histogram analysis is one of the most important thing. But since this course does not deal with image processing we are more interested in understanding the physical concepts of remote sensing. We are not going into detail about image processing details. But histogram analysis is widely used. So, in this particular slide we have histogram of Ndva values from the same image. But on the left side we have uncorrected for atmospheric effect that is the reflectance data is obtained without doing atmospheric correction. On the right side we have the reflectance values after correcting the atmosphere after doing atmospheric correction. So, here we can see there is like a considerable change in the Ndva values like here the peak value itself is 0.6, here the peak value is 0.7 and here the maximum frequency occurs in something around like close to 0.4, here it occurs something differently and there is like change in position here all these things will happen. So, in general not doing atmospheric correction will change the value of Ndvi. So, it is always suggested to do or calculate Ndva from properly, atmospherically corrected surface reflectance. We may not have enough data for doing atmospheric action but whatever is possible we should do correct the effects of atmosphere and then calculate surface reflectance and use it for estimating Ndva for our application. Then comes another important and widely used vegetation index called enhanced vegetation index EVI. So, EVI was developed when the sensor modus was launched it was primarily developed for modus sensor. In the year 1999 the first modus sensor was launched to space for with that particular sensor the data from that sensor people developed EVI. So, the EVI is like a combined index we can say like it was designed to reduce the effect of atmosphere to reduce soil background effect and also to reduce the saturation effect 3 effects. One is to reduce atmospheric effect, 2 is to reduce soil background effect and 3 is to reduce saturation effect. So, what exactly is these 2 we will understand atmosphere and soil background. So, what exactly saturation effect let us look at this particular slide. So, here we have on the x axis Ndva value sorry EVI value the definition we will see later EVI value and y axis we have Ndva value. So, we what we are seeing is both of them increase almost simultaneously as Ndva increases EVI increases almost, but after certain range like say after this particular range of say 0.7 Ndva or something or 0.7 Ndva kind of becomes horizontal like if you look in the y axis it is it stops increasing and becomes almost like becomes horizontal. So, only EVI is increasing now Ndva has almost stopped increasing after it reached threshold of something on like 0.7 or 0.8. This is because Ndva is related to chlorophyll content we know that. So, as vegetation becomes more dense and dense Ndva will be keep on increasing, but after certain limit of this chlorophyll or after certain limit of this vegetation growth Ndva will saturate and it will not increase anymore. So, let us say like I in initial classes or before we I told you about a concept known as LAI leaf area index the amount of or the area of leaf content in one square meter area. So, this LAI can vary a lot maybe for 0 for bar soils to sometimes even say 8 or 10 for healthy dense forest. So, after a certain value of this LAI say 5 or 6 values Ndva values will saturate it will not increase, but still we have a large range of Ndva sorry LAI is present to be analyzed say from healthy crop lands sometimes may have LAI close to say 5 or something, but sometime dense forest may have LAI close to 8 or 9 very high amount of leaves in one square meter area which signifies large amount of vegetation is present in that particular range. So, technically speaking the Ndva value for crop field should be less than the Ndva value for the dense forest, but it may not happen. Ndva will saturate and then it will remain more or less constant it may not go beyond certain value of say 0.7, 0.8 even when once the complete canopy closure is obtained. From a crop also can close let us say this is like crop land here like crops are standing let us say the crop leaves became dense completely close the soil, soil is not at all visible. Here let us say LAI is 5 this is like a dense evergreen forest. Complete canopy closure LAI is 8 say the number of leaves present is much higher vegetation activities much higher here in comparison with this crop. For this crop land itself Ndva may become something like 0.75 and if you look here it may be something around say 0.76 within 0.8 not much change even though there is a very high amount of change in amount of leaves present or amount of chlorophyll activity that is happening. In order to this is known as saturation effect after certain range of plus amount of leaves or amount of vegetation content Ndva will saturate it will not increase beyond certain value that is what given in this particular figure on the left side. After crossing a threshold of 0.7 or 0.75 Ndva is almost like constant without increasing. So, in order to remove the saturation effect people design this EVI. So, the EVI has the L term very similar to SAVI it uses reflectance in blue because blue band is highly susceptible to atmospheric scattering and using this blue term will help us to remove the path radiance effect because blue will carry the signals of atmosphere scattering and everything. So, we are subtracting the reflectance from blue in order to remove the path radiance effect. Similarly, we have this gain factor g in order to improve the sensitivity of EVI for high vegetation. So, EVI is kind of like an improved Ndvi which reduces the effect of atmosphere soil background effect and also the saturation. So, if you look EVI for the scroplanded dins rainforest we will see a difference. EVI for dins rainforest will be higher than EVI for the scropland with LA5 and LA8. So, this was the major application of EVI in order to improve the information content of Ndva. Again this is one of the operationally produced product from motor sensor it is operationally produced and from other earths is earth observing sensors also EVI is now being produced consistently and this is also now one of the most widely used vegetation index for monitoring vegetation. So, the main difference of main purpose of EVI is to remove or reduce atmospheric effect, soil background effect and saturation effect. So, just before we wrap up the topic of spectral index there are like lot more other indices we have here primarily concentrated on vegetation indices for monitoring vegetation. But as I said such spectral indices are also developed for understanding various other features such as snow for identifying burnt areas like in forest fire what area got burnt for identifying that for identifying built up areas and spectral indices exist for several of these things. So, now we are going to see spectral index for snow known as Ndsi normalize difference snow index. So, normalize difference snow index takes advantage of the reflectance property of snow in green and SWR band. So, this is very similar to that MNDWI definition we have seen earlier green and SWR, but when applied over snow covered conditions like reflectance and green for snow will be very high in SWR it will be quite low. So, high Ndsi represents high snow covered area and this index is also one of the widely used index for monitoring snow like the operational snow products from modus sensor and various other sensors for mapping snow cover area for mapping snow depth etc. This index is one of the widely used index just by taking the difference in spectral index between snow and other property it is being used. So, as I said there are plenty of other indices represent, but for the want of time we are not going into detail. Interested readers can look at several sources in internet where detailed information about various spectral indices exist that are being used in various applications. So, as a summary in this particular lecture, we have discussed about some of the few commonly used spectral indices with special emphasis on vegetation. We have seen index such as Ndwi, Ndwi, Evi, Savi etc. And also we have seen a spectral index called Ndsi, normalised difference snow index. So, with this we complete the topic of understanding the spectral reflectance properties of materials and also spectral indices. So, from the next lecture onwards we will move head to a new topic of thermal infrared remote sensing. With this we conclude this lecture. Thank you very much.