 Hello everyone, welcome to the final lecture of this course remote sensing principles and applications. So, here we are going to discuss about and discuss further about like applications of remote sensing and water resource management. So, we already discussed retrieval of river per transpiration and soil moisture. We will briefly discuss mapping and monitoring of open water bodies and the limitations of remote sensing in water resource management. So, this topic, the mapping of surface water bodies is one of like the vital topics for not only for water resource management, but also for environmental monitoring. Because with crappy urbanization and encroachments happening, we are many places are reporting loss of water bodies like number of lakes are shrinking or the aerial expansion of water bodies are shrinking and so on. So, monitoring this is kind of like really vital. So, surface water refers to what is present on the top surface like rivers, lakes, wetland and ocean, but normally oceans will be excluded when people try to monitor surface water bodies because people our concentration will be on mapping and monitoring the whatever is present within land like inland water bodies what we call or the wetlands basically. So, these surface water bodies are highly dynamic in nature like the lakes, aerial extension may change, the depth of water will change, everything will happen based on the availability of rainfall, water inflow to the lakes and so on. So, these are highly dynamic and because of the importance several studies have been carried out for the mapping of surface water bodies both at global scales and at local scales. So, we will discuss optical remote sensing for surface water mapping rather than discussing in detail about all the thing I will just provide like a brief overview. So, optical remote sensing uses the large change in reflectance between land surface and water bodies for mapping it. Say we already know whenever like a water body is present, the reflectance of a water body in NIR is very low. So, we will try to use this equation the NIR, NIR as well as SWIR bands we will try to use this difference and take it to our advantage for mapping water bodies. So, if all conditions are ideal like we got like a perfect image without any sort of like sun glint. So, sun glint is whenever like sun is directly overhead a water body we may get like a bright spot on the image. If both the look angle of zenith angle of sun and the sensor is more or less the same then a specular reflection kind of thing will happen which may actually affect the reflectance that we measure it will appear like a very bright spot we call it as like sun glint. If those things have if those things are absent if the image is like perfectly clear without any sort of turbidity and all then even a simple density slicing of infrared image will help us in delineating surface water bodies. So, what exactly density slicing is like we know land like land features has much higher reflectance in the infrared portion either NIR or SWIR in compared to water bodies. If we plot like a histogram, a histogram is nothing but like the distribution the frequency distribution of the DN value say or surface reflectance whatever let us say we have surface reflectance this is like the different bins let us say this is 0 to 0.1, 0.1 to 0.2 and so on. So, we have divided into several bins. In each bin we will count the number of pixels having a certain range let us say water bodies may be falling somewhere here and then land may fall somewhere here. So, it can be clearly delineated water bodies will have lower reflectance let us say this is in NIR band and land features will have higher reflectance. So, if we put a threshold if the reflectance is less than a certain value then classify the lower reflectance values as water other values as non-water bodies that is all. So, water and non-water it is kind of like a binary map. This is like the simplest way in which we can use infrared images for mapping water bodies. This is under ideal conditions, but this may not always work and that is why people have developed spectral indices such as NDWA normalize difference water index relating green and NIR or green and SWIR this is NDWA or MNDWA modified normalize difference water index. So, using this relationship normally water has higher reflectance in green and lower reflectance in NIR and SWIR using this characteristics people have developed indices for identifying open water bodies. Again these indices have proven successful for identifying water bodies normally they will have values like positive values for water bodies and negative values for non-water bodies. So, but the threshold can vary based on region to region and the quality of data we are using. So, these are sort of indices if we compute it they will directly tell us okay this is water body or not. The only problem using MNDWA is if you recall it has same definition as the tough NDSI normalize difference no index. So, if we want to map open water bodies under areas or regions which also has snow let us say we have like one image some the image has some sort of like high mountains containing snow then there are like lower valley portions which has some lakes. If we apply this MNDWI over that portion since the definition is more or less equal to NDSI normalize difference snow index then both snow pixels and water pixels will be highlighted because of their reflectance characteristics. Hence just using MNDWA may not be suitable for such regions we have to bring other information say even using NDVI there or NIR band data may prove worthy in order to delineate snow cover and water bodies identified by this. So, many studies have used such simple indices to a good successful extent for identifying water bodies. However, one of the major difficulty in using such indices is identifying kind of like a threshold. Threshold means say less than this value non-water body more than this value water body. So, identifying that particular threshold is of paramount importance and any errors in identifying that threshold may actually lead us to miscalculating the extent of surface water bodies. So, this is kind of like a manual process and in order to overcome this several researchers have improved these indices or developed new indices from these reflectance values and this can be seen in this particular slide which shows different indices present on the year in which they were developed. Similarly, what are all the satellites that were launched from which these indices can be obtained. So, the top part of this line gives the satellites or the sensors which provides the data for the different indices and whatever the indices we need are all given here. And people have even used a combination of NDVI and NDWI. This is also been used in order to retrieve open water bodies. So, there are different ways and different thresholding methods have been used. But only one problem is no single spectral index has been found to perform well globally and at all times. So, essentially the studies are regional and global studies have been carried out using some sort of like machine learning or AI based tools. Like even I told you about like Google Earth Engine platform like there are like few studies which mapped global water bodies using like this Google Earth Engine using the algorithms present there. So, this field is actually continuously evolving where people are trying to use the upcoming computing technologies as well as a new satellite data for retrieval of or for mapping open water bodies. So, the one of the references here given here will tell you like a brief overview or like a broad overview about the subject and then the global surface water map which is again like published as a paper. This has been done using like Google Earth Engine platform if I remember correct. So, these are some studies which highlights the different ways in which we can map open surface water bodies. So, finally the issues in remote sensing for water resource management like there are like again plenty of applications which we are not covering for the want of time and also for the want of background information that we need to discuss before we go on to the real applications part that itself will be will became like one separate subject application of remote sensing in water resource management is actually like a separate subject that we offer here in IIT Bombay. So, which we are like highly condensing and providing in terms of like few lectures. So, we are kind of like wrapping it, but before that we will see like the issues of remote sensing in water resource management. The first thing is the quality of data. How well the satellite data matches with the variable that we want? Say reflectance we measure from optical data, temperature we measure from thermal infrared all these things. How well it measures with our variable of interest? Say if you want to measure precipitation, what wavelength to use and how well those wavelength matches all these things is like if you observe like or if you read like the review paper given here in this particular slide it will tell the different variables and the different satellites available to us. So, from which you can understand retrieval of different variables actually requires different types of sensors, passive microwave, active microwave, thermal optical and also like specific characteristics of the sensors in order to improve the data collection by the satellites along with the variable which we need to measure. So, that is one thing. Second thing is resolution both spatial and spectral sorry spatial and temporal. How frequently the data is collected? Say soil moisture changes very frequently, can we get it every day or every or twice a day and so on or at which resolution? Are we getting at field scale or are we getting at like tens of kilometer of resolution? That is again an issue which people are always trying to work. Some data may be very good in quality, but the resolution may not be optimal. The passive microwave retrieves soil moisture or some data will be of very high resolution, but it may not be of good quality due to other errors say due to some nature the thermal infrared might not have got the EET properly or the active microwave remote sensing might not have got the soil moisture information properly due to the missing information in for other variables all these things might have happened. And sampling, sampling is how often the sensors sample the land surface again like related to temporal resolution, but again if the sampling is frequent we may get frequent data or even sometimes what will happen is if you want like thermal data sets if the satellite may sample frequently may be modus samples the earth almost twice a day 2 data points we may get or if we combine 2 modus sensors we may get 4 data points in a day. But the problem is are those 4 data points are free of cloud? If cloud were present over our steady region then all those 4 points will not be of useful to us. So, the sampling as well as the availability of data influences the variable retrieval and our applications. Then the legacy of data how long the time series of data we have say for some applications like dot monitoring we need to have like a longer time series. But some variables are actually being measured only recently say soil moisture from passive remote sensing like this L-band radiometers have developed only from the year like 2009 after the launch of SMOS before that there were other satellites that were in thing that were in orbit. But they were not L-band actually L-band is like the most suitable but they were in other bands. So, there is always be some sort of mismatch when you try to compare the soil moisture from some other bands with what is retrieved from SMOS or SMAP. So, the data length is actually kind of like a problem we may be requiring at least 10 years of data but data may be available once only for 5 years and so on. Then latency say we want data observed today but are we going to get the data immediately after satellite observation or do we need to wait for another week or another month and so on like the time period between satellite observation over the region and the actual product release for that particular day that is called latency. So, for most of the applications we may be requiring data very soon like after 2 to 3 days of satellite overpass or for some sort of a real-life events like floods or droughts the latency time should be very low within a few hours of satellite observation we may need to have data. So, again this latency plays a major role. So, these are some of the things we need to keep in mind while using remote sensing data sets for water resource management. All these issues if we see they are all related with the sensor characteristics. Sensor characteristics means the spatial spectral temporal and radiometric nature of sensors plus their orbital characteristics all these things will influence the type of data we collect, the sampling that we do and also like the launch of sensors and the availability of sensors will determine the legacy and so on. And also only some variables can be directly observed most of the variables in water resource management cannot be directly observed we need to do some sort of retrieval process using some sort of modeling like both evapotranspiration and soil moisture that we have seen are not directly observable. We need to use retrieval algorithms there can be plenty of uncertainties within the algorithm themselves the data may be of good quality but the retrieval algorithms may not be performing well. So, again our applications will suffer. So, all these so this is not like a simple task even though like we know this can be used we need to be really careful in selecting a suitable sensor for our applications like one very good example is like if you look at like the overpass time of soil moisture dedicated missions like smores or smat they will be overpassing in sun synchronous orbit around like 6 am or 6 pm local time not at like during midday or early it will be like early morning and then evening that is because of two reasons one is to reduce what is known as a Faraday rotation effect and also other way other thing is during like early morning hours we have already seen that will be what is known as a thermal crossover like we have discussed this while we discuss the topic of thermal infrared remote sensing like during like pre dawn early morning hours the temperature of different objects on the earth surface will be very close to each other and hence thermally they will not be differentiable at that time these satellites will try to overpass because our aim is to get emissivity. So, if all the objects are at same temperature then the brightness temperature observed is will vary only based on emissivity of the objects right let us say you have like a big region say you have 2 pixels say both of them will let us assume they are at almost similar temperature but this gets a brightness temperature Tb1 this gets a brightness temperature Tb2. If all the way if the temperature of them is very close then the difference in brightness temperature can be directly attributed to emissivities of the pixels say this can be dry having a high emissivity this can be wet having a lower emissivity. So, these things can happen actually. So, in order to get this optimal condition for getting emissivity we do this but if we want to actually get temperature for evapotranspiration mapping we cannot use this particular data when all the objects are at same temperature we need to get proper temperature of the surface or different features present over there. So, there are like different things we have to keep in mind say these satellites are dedicated for the application they are orbital characteristics are chosen suitably but if you want to use some other satellite like in olden days 2000s we had AMSRE it was useful in retrieving soil moisture but it will overpass at afternoon time where the temperature will be highly differing there will be some sort of errors the wavelengths are not will not match with L band frequency and so on. So, all these things play a major role in the quality of the data that we get and also the applications we do. So, with this we end this particular topic applications of remote sensing in water resource management and also applications of remote sensing are plenty there can be different fields like geology, urban studies, environmental monitoring, water quality it is almost like we can keep on counting the applications and it is close to impossible to cover all the applications. As I already told you when we want to discuss applications definitely we need to discuss like the background domain information before we go on to that which will be which will require several courses in order for us to cover it. So, the aim of this course is just to provide you a feel for the different ways in which remote sensing data sets can be used. I broadly touched different data sets like optical, how it is used for land use land cover mapping, thermal data we discussed about its use in land evapotranspiration mapping using surface energy balance equation, passive and active microwave for soil moisture and so on. And LIDAR I just briefly mentioned their needs in like cryospheric monitoring and vegetation monitoring, topographic mapping and so on. So, it is kind of like a extremely broad way of applications, but with the principles that we have learnt in this course and you can having a domain knowledge in your own field like some of the students taking this course maybe from water resource background, some maybe from earth science, some maybe from ecological background and so on. So, combining this remote sensing principles along with your own domain knowledge, I am very sure you can definitely come up with your own applications. There are like plenty of study material available in the internet like scientific literature is now increasing rapidly and it is only up to us to choose the best in order for us to develop our own knowledge and get practice. So, with this we end this particular lecture as well as this particular course. I hope the course like the students would have liked the course the contents the contents of the course would have been useful to you and I wish you all wish you all the very best for all the students who are taking the course. And thank you very much for your learning as well as like interest in this particular topic. Thank you and wish you all the best.