 Hello everyone, welcome to the next lecture in the course. We are discussing about the applications of remote sensing in water resource management where we discussed about the different ways of estimating ET that is evapotranspiration and also we started discussing about soil moisture. So, we discussed what soil moisture is like the different depths of soil moisture measurement, how it will play a role and optical and thermal methods of estimating soil moisture. Today, we are going to cover how microwave remote sensing is useful in soil moisture estimation. While discussing microwave remote sensing both active and passive, I told that microwave signals react to the electrical and physical properties of the features on the earth surface. That means, whenever water is added to the soil, then it will change the electrical property of the soil which will be picked up by the microwave signals. Both active as well as passive microwave can be used. We have seen some examples while dealing with those particular topics itself. So, here we will see like the basic principles behind the estimation of soil moisture from passive and active microwave remote sensing. First, we will start discussing about the passive microwave radiometry for soil moisture estimation. So, we have already seen that passive microwave radiometers measures or observes brightness temperature that is what we will get from the data. So, the brightness temperature depends on the actual temperature of the object and the emissivity of the object. So, that Tb is equal to emissivity into T in microwave wavelengths that is like a direct relationship. So, whenever a soil is wet, the emissivity will change and actually it will go down. For dry soils in microwave wavelengths, emissivity will be higher. When the soil gets wet, the emissivity will go down, due to which the brightness temperature of wet soils will be lower than the brightness temperature of dry soils. Observing this particular difference is helpful in retrieval of soil moisture estimation from passive microwave radiometry. So, the retrieval of soil moisture is kind a two-step process. So, first, the brightness temperature observed by passive microwave radiometers has to be inverted to get the dielectric constant of the surface. That is based on the dielectric constant like dielectric constant is a property of the object which relates to its like electrical nature. So, the first relationship is to from Tb or brightness temperature scientists will try to retrieve the dielectric constant through what is known as using radiative transfer models. That is if this is like the basic land conditions including the dielectric constant, how the brightness temperature will be. So, the radiative transfer models will be able to simulate this. So, they will be trying to use this brightness temperature to retrieve to retrieve the dielectric constant. And once it is done, then using physical relationships, physical or empirical relationships, people will try to connect this dielectric constant with the soil moisture. So, this is done using what are known as dielectric mixing models. So, first from the brightness temperature we get the dielectric constant and from this dielectric constant we go to soil moisture. It is kind of a two step process. So, apart from soil moisture, the brightness temperature is also influenced by vegetation, surface roughness, atmospheric and cosmic background emission. So, basically when we get brightness temperature, these effects the atmosphere and cosmic background emission effects will be removed or at least reduced to the great extent by the data providers. So, the brightness temperature what we get after processing by the data providing agencies, we can assume it to be free from these effects. But still we have to account for the effects of vegetation and surface roughness while we are trying to retrieve soil moisture. So, how vegetation will influence we have already seen this. So, presence of vegetation above the soil will attenuate the signals from the soil basically. So, whenever a microwave radiometer sees a land surface containing vegetation and soil together, then essentially the microwave emission from the surface is going to be attenuated as it passes through the canopy. And we have also seen that longer wavelengths or shorter frequencies are capable of penetrating the canopy to a better extent than shorter wavelengths. And actually studies have proven that L band is like kind of suitable for soil moisture estimation that is why we have two dedicated missions smalls and smalls for soil moisture estimation operating in the L band. So, this vegetation effect is typically characterized as vegetation optical depth. So, that is you can think it off in terms of like water content present in kind of a the volume present above that particular pixel. See there is like a pixel. So, whatever present over there like vegetation like whatever vegetation present over there it will have water content in different parts of its like the parts of plant rights whether it is shrunks, stems or leaves vegetation will present. Even in the case of like recent rainfall the leaves on the top will contain rain droplets that is again like water present within the in the vegetation system. All these things combined will affect the signals from the soil. So, the vegetation optical depth typically characterizes the vegetation water content. So, if you have like a 2D pixel imagine it kind of like a 3D volume. So, whatever the water water contained within that particular volume will actually affect the signals from the soil. So, it will be modeled or it will be obtained through some other ancillary measurements in order to remove the effect of vegetation. So, the basic equation what we call the radiometric transfer equation, radiative transfer equation is the brightness temperature observed by the satellite sensor majorly comprises of 4 different components. The first thing is temperature of the land surface that is the brightness temperature of the land surface attenuated by the atmosphere. This is component number 1, the downwelling atmospheric emission that is if there is like a land surface some part of like the particular land surface will receive emission from the atmosphere that is component number 2. Similarly, we have also seen the cosmic emission that is the emission from the outer sky may influence their measurements. So, that is component number 3 and component number 4 is upwelling atmospheric emission. So, the upwelling atmospheric emission is what is going towards the sensor directly from the atmosphere. So, among the brightness temperature components, components 2, 3, 4 are actually like atmospheric and outer space components which are kind of unwanted things to us. So, once we remove it then essentially we will be left with the brightness temperature from the land surface which we can use further in order to reduce soil moisture. And this brightness temperature from the land surface again comprises of different different components. So, even the land surface has several components within it, there can be like soil, there can be like vegetation standing over the soil or some other features. If we come if we assume that the land surface is comprised of vegetation and soil then the brightness temperature can be written as can be split as the first component is temperature of the soil which contains the signal of soil moisture. So, T s is the thermodynamic temperature of soil and epsilon is the emissivity. Then emission from the canopy because canopy also has its own temperature that will do some emission. Then the emission from the canopy towards the soil that gets reflected that is third part. So, in the three parts we have one the direct emission from the soil that is attenuated by the canopy it has to pass through the canopy if something is present. Then the direct emission from the canopy and the third part is the emission from the canopy in the downward direction reflected by the soil. So, among these things again if you look at it we have this parameter what is known as like the attenuation parameter by vegetation, then we have what is known as like the single scattering albedo and so on. So, there are like several variables involved, but for the sake of simplicity we will not discuss all these things in detail here. But just as an overview the brightness temperature of the land surface observed by a passive microwave radiometer again comprises this particular three components out of which only one component is of direct interest towards. So, the emission from the canopy component has to be removed from this and that is why like scientist will always work towards like this VOD like vegetation optical depth as one of the parameters which contains information about vegetation. So, the basic radiative transfer equation what we have just seen comprise of three methods can be solved in two ways either in forward modeling way or in inverse modeling way. In the forward modeling, so the soil moisture is assumed as given a particular value so that the brightness temperature will be simulated and it will be matched with the observed brightness temperature that is due to some effects of soil moisture and the land surface will have a particular brightness temperature right. So, this particular brightness temperature will be simulated by feeding the model with soil moisture and this will be simulated and whatever is observed by the satellite will be compared with this and the errors will be minimized by playing with the parameter. So, this kind of modeling is called forward modeling. Assume initial value of soil moisture and try to retrieve brightness temperature and match it. This is one way. Other way is satellite has already observed soil moisture sorry brightness temperature. So, the using that particular brightness temperature use the radiative transfer model in the reverse fashion that is now I know what is the emission from the land surface. So, what causes that emission? So, come in the reverse direction. So, that is called inverse modeling. So, radiative transfer model actually works like this okay this is the land surface property this will be the emission. If we do this that is forward modeling but if we observe the emission and come towards the land surface property that is inverse. So, soil moisture can be modeled in both the ways. Normally that brightness temperature will be fed into radiative transfer models to estimate soil moisture that is how normally the satellite base retrievals will work normally. So, again we need to account for vegetation parameters as well as surface toughness parameters. So, again there are like different classes of algorithms exist or based on the sensor characteristics. If a sensor has say a mono configuration, mono configuration means a single instance angle, single frequency or like a fixed polarization everything is like fixed without any change. Then the algorithms that are developed to retrieve soil moisture has to depend upon other ancillary data sets in order to estimate vegetation and surface toughness properties. Say like you can search in the internet for the what are known as like the ATBDs algorithm theoretical basis documents of say SMAP satellites, SMAP SM retrieval measurement or SMAS SM retrieval measurement. So, these are like publicly available documents. When you go through them documents it will have like all the technical information about what other ancillary data the scientists use how important they are all this information you can get. But if the sensors configuration is very simple without much options available for us then essentially the vegetation information or the surface toughness information has to be obtained independently. So, some ancillary data sets has to be developed. But if the sensor has multiple configuration say like a dual frequency or a dual polarization then the algorithms can be suitably modified in order to retrieve soil moisture, VOD and our surface toughness simultaneously. So, normally if you take like the SMAP SMAP retrieval algorithms we have both the single channel algorithm and dual channel algorithm. So, in the single channel algorithm the vegetation of properties or the VOD will be estimated using Ndva information available from optical data sets whereas in dual channel algorithm the vegetation optical depth will be retrieved simultaneously along with soil moisture because of the way the algorithm uses the data from the satellite. So, based on the satellite configuration and the data available to us there are again variety of algorithms available to us. We will not see the algorithms in detail but just like a generic overview is what we are getting from this particular lecture. So, this is with respect to passive. So, the basic principle is from the brightness temperature observed separating the unwanted effects like first removal of atmospheric effects then removal of canopy effects and surface toughness effects will lead us to the retrieval of soil moisture. This is like the basic principle and using this basic principle only algorithms are being developed and operational data products are also available for soil moisture estimation like we discussed about like several data sources. So, the SMAP data soil moisture data is available from what is known as the NSIDC National Snow and Ice Data Center which contains soil moisture data obtained from SMAP satellite. So, which is like freely developed it is available at 36 kilometer resolution 9 kilometer resolution at multiple temporal frequencies different levels and so on. So, interested users can download the data is again available in the HDF format which is again we just briefly saw how it is and use it for various applications. So, next we move on to the active microwave for soil moisture. So, similar concept addition of water to the soil will increase its back scattering coefficient. So, the soils or wet soils will appear brighter in active microwave images. But the problem with active microwave remote sensing is in comparison with passive microwave active microwave remote sensing is highly influenced by surface toughness parameters and vegetation parameters. So, they have to be perfectly modeled or ground measurements has to be taken care of in order to retrieve soil moisture from active remote sensing. So, in active remote sensing both the sensor configuration that is which frequency it is observing at what look angle it is collecting the data, what polarization it is collecting the data. So, all these kind of sensor configuration and the surface characteristics especially the presence of surface roughness and vegetation characteristics will influence soil moisture estimation. So, the advantage that active microwave remote sensing has over passive microwave is its spatial resolution. Say we have already seen that passive microwave radiometers will have very coarse spatial resolution in the order of few kilometers and the soil moisture observing missions has spatial resolution close to 30, 40 kilometers. So, which each pixel size will be roughly 30 kilometers or 40 kilometers or at best the disaggregated product itself will be available at say 9 kilometers and so on. Which is kind of extremely coarse for various applications for regional or local level applications. But active remote sensing can provide data in the order of hundreds of meters or sometimes like Sentinel normal provides data at 10 meters or 20 meter resolution, Sentinel 1 satellite. So, if we can collect all the information required for soil moisture estimation like surface roughness and all, theoretically it is possible to retrieve soil moisture at 10 meter or 20 meter pixel size which is large improvement when we compare this with the passive microwave radiometry. But as I told the surface roughness effect has to be removed and this surface roughness effect comes into picture to a larger extent when the satellite or the look angle changes like normally the lower angles are preferred rather than comparing with higher angles like look angle and instance angle these two are like relatively used or interchangeably used. So, the instance angle normally lower instance angles are preferred for soil moisture estimation because at higher instance angle surface roughness effect is more pronounced. So, the different classes of algorithms available for the retrieval of soil moisture can be classified as physical models, semi empirical models, empirical models and change reduction methods. So, physical models uses physics based equations relating back scatter coefficient to dielectric constant and then relating dielectric constant to soil moisture. So, these are physical equations. The problem with physical equations is they demand lot of data like if you want to remove surface roughness effect we need to know the surface roughness. So, what is known as like the RMS height we should measure. So, there are like some instruments called like micro profilometer or something which we should take to field, put it in the field and then measure how the topography varies very minutely all these things we have to do we have to calculate the RMS height and feed it or sometimes we may have to measure vegetation parameters carefully or we may have to measure even soil moisture at the time of satellite overpass all these things may be required for physics, physics based models. So, normally physics based models can retrieve high quality soil moisture provided we give all sort of data required by it. So, that is one class of algorithm then semi empirical models. So, they are kind of like a balance between physics equation and what we measure in field really say if we write all the relationships in using like physics based equation like this is the dielectric there are like plenty of equations which I am not giving here, but just like schematically speaking let us say there is an equation relating dielectric constant with the back scattering coefficient. So, if this equation is kind of like highly physical which involves lot of parameters to measure can we model it simply say can we directly measure soil moisture and can we relate back scattering coefficient to soil moisture something like that. So, doing some sort of simplified assumptions and using ground measurements to substitute what is there in the physics based equation that is semi empirical. So, it is a combination of field developed equations maybe vegetation parameters you could have developed simple equations based on NDVI or LAI that may be valid for that particular region which can be used in the physics based equations. So, these kind of models are semi empirical. Then comes empirical, empirical is directly relating the back scattering coefficient with field observed value say you have a region we observe soil moisture at certain plots at the time of satellite overpass and we somehow try to develop a relationship between the soil moisture and the satellite returned back scattering coefficient. So, then using the back scattering coefficient at all other plots and using this relationship developed we can estimate soil moisture. So, the empirical models are extremely simple, but at the same time they have to be applied or they can be applied only over the region where they are like where the relationships are developed and they are like non-transferable and even like nonscalable even if the satellite parameters changes then the equation will change and also like measurements has to be taken at the time of satellite overpass which again may complicate the data requirements. The final class of model is the change direction models. So, the change direction based models will not actually retrieve the true soil moisture value the actual soil moisture value, but they do some sort of like a relative level ok what is like the change between two time intervals. If we have some reference soil moisture with respect to that reference how the soil moisture changed. So, we can always get to this relative change. So, the change direction methods how it works like a simple equation is given here. So, the normally what they are like different different ways which it can be done. So, this is normalized radar backscattering moisture index which works based on the backscattering coefficients observed at time t1 and t2. This equation is developed based on having like a reference dry and wet backscattering coefficient values for one particular pixel like you need to have like a large time series of data say data something looks like this within the time series say this is like the minimum and this is like the maximum and if this is like the in between backscattering coefficient value how this changes. So, this is kind of like relative measures. So, basically change direction methods assumes the vegetation characteristic and the surface toughness characteristic changes in a much slower way in compared to the soil moisture that is you we always compare the backscattering coefficient in time t1 and t2. If we assume the change in backscattering is only due to change in soil moisture then obviously we are assuming that surface toughness parameter has not changed much and vegetation parameters has not changed much. So, these are like again simplified representations of real world. These kind of change direction techniques may be applicable over like crops which has like longer seasonality where the vegetation growth takes place lower and the surface toughness may not change once crops start growing farmer will not try to disturb the surface characteristics. So, under those conditions it can work or like under like a uniform vegetation canopy then these things may work under natural environments this may work. But again there are like there are limitations but still change direction algorithms one of like the most widely used algorithm for soil moisture estimation. So, as I told the different factors that affect soil moisture retrieval from microwave are the sensor characteristics the frequency polarization the instance angle plus the surface characteristics which includes surface toughness vegetation biomass and water content soil texture and topography because I have already discussed especially in active microwave remote sensing the overall backscattering coefficient can change based on the topography whether it is on a flat topography or the topography is phasing outwards or away from the radar and so on all these things will play a major role. So, essentially our aim is to remove all other effects apart from soil moisture for us to retrieve soil moisture from this particular method. So, on an overall like for overall comparison between active and passive microwave methods if we look at it then the passive microwave is actually highly related to soil moisture and hence the results are very much promising especially over like bare soils and also they are not affected by clouds or daytime conditions. They provide high temporal resolution in the order of like once every 2 days, once every 3 days or sometimes model based data comes every 3 hours once. So, passive microwave radiometers are like the best ones at least now to provide soil moisture. But the major disadvantage with them is their core spatial resolution which is available in order of kilometer scale and hence we need to develop some sort of like disaggregation or downscaling algorithms to bring this coarse resolution soil moisture to fine spatial resolution. So, active microwave again goes over this limitation it provides us fine spatial resolution again it is not affected by cloud covered but at the same time it is highly influenced by surface roughness and vegetation amount and coarse temporal resolution. Say active microwave data may be available say central 1 and 2 combined we may get data once every 6 to 8 days that is now but if you look at olden days data may be available once every 20 days and so on. So, the temporal resolution is bit coarser and the influence of surface roughness and vegetation cover is very high when we compare this with passive. So, in order to overcome the limitations of passive scientists are trying to combine optical data like thermal visible and air data with passive microwave data for improving the spatial resolution. And also both active and passive combinedly are being used say the SMAP mission has a SMAP Sentinel combined soil moisture product at 3 kilometer spatial resolution. So, that is combination of active and passive remotes and things. So, this is kind of like a trade off some characteristics of passive microwave will be taken some characteristics of active microwave will be taken. Similarly thermal data is often integrated with passive microwave radiometry data in order to retrieve or in order to disaggregate soil moisture. So, this is also being carried out especially from the context of improving the spatial resolution of soil moisture observations. So, bugging there are like some advantages and disadvantages for these kind of like synergistic methods especially when we discussed about the difference between passive microwave radiometry and thermal infrared we discussed one important thing that is like the depth of penetration. The thermal infrared remote sensing normally senses or what we observe is only from the top millimeter of a surface the skin temperature. So, essentially the soil moisture that we retrieve from it will have only the top surface soil moisture information be it canopy if it is canopy we will see only the canopy if it is bare soil we will see only like the top portion of soil. Whereas microwave especially longer wavelengths will have certain penetration capacity if there is a canopy standing over the soil some sort of penetration would have occurred or if it is bare soil the depth of measurement may be few centimeters at least say 4 or 5 centimeters at least. So, the depth of measurement of soil moisture observed by passive microwave radiometers and thermal infrared sensors are different. So, when we try to combine them we should keep in mind this difference. So, what we are measuring is because I already told you as the depth of measurement changes the soil moisture amount may change or what we actually want for our application will be different from what is actually measured. We may be wanting root zone soil moisture whereas thermal measurements will only provide the skin soil moisture. So, these kind of differences will always come and we should keep in mind when we try to combine different wavelengths or different technologies of remote sensing together for improving the spatial resolution. So, this is like a very broad overview of usage of microwave remote sensing for soil moisture estimation again this particular topic itself is very wide and very broad. So, normally we cover this particular topic in few weeks of lectures when we offer this as a part of course to the students here in IIT Bombay. But since this is like introductory remote sensing course I did not want to divulge more into the technical details and I provided like a broad overview. But in the slides there will be like plenty of references which the interested users can always refer to. So, with this we end this particular lecture. Thank you very much.