 Welcome to today's class. So just a quick preview of what we have been learning so far before we start today's lecture. So till now through module 2, we have been learning about what is synthetic aperture radar, the basics of image formation and then we learnt about the different modes of acquisition of SAR image and few terminologies that will be used throughout this course mainly what is azimuth resolution, what is range resolution and a few terms like swath, nadir, the concept of using chirped pulses and we also learnt how to derive a radar equation after understanding the meaning of wax scatter, sigma, sigma naught and in the last lecture we started our understanding about image defects and in particular the geometric distortion was discussed that is layover for shortening and radar shadow. So we are presently in the 7th lecture of module 2, there is a lot of information contained in the sigma naught imagery and we can either choose to visualize the images or we can break them up into homogeneous groups with similar statistical properties and underlying knowledge about scattering from different targets can help us in analyzing a radar image better. When a satellite is looking down on the earth surface, it is seeing a distributed target and not a discrete target, not a discrete but distributed target which means a footprint of a radar system can contain more than one features, it can contain urban area, a vegetation, water body, so there is a lot of information that is contained in the sigma naught imagery and in this context I want to introduce something to you in terms of scattering. So before I start explaining, I will just play a video so that what I say orally is clear. So we are trying to understand scattering from different targets like vegetation, rocks, background and buildings. So you see the difference here, scattering first from vegetation, then from background and then from buildings. See there is something known as a single bounds, double bounds and multi-scatter. For example, the scattering from buildings, we call it as double-bounds scattering because it is occurring between two surfaces which are at right angle to one another, is not it? It is called as double-bounds scattering because it is bouncing twice as seen in the video here. And then we can have something known as a single-bounds scatter that refers to a simple scattering from a surface, single-bounds scattering. And multi-scatter is what occurs when the radar returns from an extremely thick vegetation, say the Amazon forest, extremely thick vegetation or forest cover. When we have radar returns from such a feature, we call it as multi-scatter. So here the incident wave tends to undergo multiple bounces. So many times it may get scattered within the canopy and hence it is referred to as multi-scatter. So you see the scattering characteristics of different surfaces tend to vary. For example, in the urban areas, we can have very strong backscatter. In the forest areas, we can have intermediate backscatter and again the calm water. You know calm water, it can have very low backscatter. So shown here is a diagram which lets us know the kind of scattering that is occurring from the same feature that is vegetation when there is difference in wavelength. For example, the penetration capacity of microwaves are different. You know for X band there is a different penetration, for C band there is a different penetration, L band has a different penetration. So all in all, the radar backscattering coefficient, it depends upon the radar observation parameters which is nothing but frequency, polarization, horizontal, vertical, HH, HV, we discussed that polarization and it also depends upon the incidence angle that is theta i as seen in part of the previous lectures. In addition, the radar backscattering coefficient is also going to depend upon the surface roughness. As seen in the previous video, remember that different surfaces in a distributed target tend to have different roughness, different dielectric properties. Combined together, the sigma naught imagery gives us valuable information about the target, the target which is causing the backscatter. So in this context, let me reiterate what we discussed about corner reflectors. Remember, in one of the earlier lectures, I also discussed that all targets need not be natural. We can have some man-made targets which are known as corner reflectors shown here are the trihedral and the dihedral corner reflectors which are present in Hyderabad as part of National Remote Sensing Center ISRO. So the speciality of corner reflectors is that they are mounted according to look angle of the instrument, such that there will be maximum scattering towards the sensor with minimum loss. Let me reiterate, it is mounted in such a way that it is aligned with a look angle of the instrument so that the maximum scattering happens towards the sensor with minimum loss. Now corner reflectors as shown in the screen, they are standard targets which are used for synthetic aperture radar calibration. And as mentioned earlier, they are oriented with respect to the bore sight of SAR antenna for each pass. Of course, for precise orientation, we may need to use magnetic compass or digital level box, inclinometers, etc. So just let me play a video so that whatever I say verbally is clear to you visually. A microwave calibration validation site set up with the deployment of corner reflectors at Imgios National Remote Sensing Center, the function of corner reflectors is that it appears as a bright spot in a synthetic aperture radar image because there is maximum scattering happening in the look angle of the radar system. Now the site as such it is free of vegetation and it is a very important factor for absolute radiometric calibration. The site is chosen in such a way that there is low backscatter and there is a strong backscatter only coming from the corner reflectors which is why it tends to appear as a bright spot in a synthetic aperture radar image. Now with this background, let me move to the second kind of distortion in SAR image known as radiometric distortion. So by now, we have acquired the understanding that a radar system measures the ratio of power, ratio of what? The power of transmitted pulse and that of the return echo received and that this ratio is called as backscatter. Radiometric distortions are significantly caused in the case of mountainous terrains or complex terrains. We say that a significant cause of radiometric distortion in synthetic aperture radar images is when there is interaction of radar signals with the mountains, say the Himalayas, complex terrains and the radiometric terrain correction can be performed using a digital elevation model or DEM. We will see what is DEM in detail as part of upcoming lectures but for now understand that it is a digital representation of elevation of a terrain. The three-dimensional elevation information of a terrain can be represented in two dimensions using a raster we call it as digital elevation model. Now remember that owing to varying topography, the actual ground area that is contributing to each pixel is not known. We do not know that and to know that we need topographic information. So the radiometric correction to be precise, the radiometric terrain correction is performed to restore the radiometric uniformity to the image. So the whole aim of radiometric distortion correction is to remove the artifacts caused by terrains to correct them. Now in this regard please remember that we can have radar images from different sensors and if we need to inter compare these images captured from different sensors we need something known as radiometric calibration, radiometric calibration. And this is performed to provide imagery in which a pixel value can be related to radar backscatter of the scene. Let me re-itrate, radiometric calibration is performed to provide us with imagery in which the pixel values can be related to the radar backscatter of the scene. So the whole aim is after this the pixel values of the SAR images they truly represent the radar backscatter of the reflecting surface. Now the actual backscatter provided by a target or a group of targets is obtained after radiometric calibration and it is in this regard that it becomes important for us to know the meaning of backscatter coefficients which are nothing but sigma naught, beta naught, gamma. We saw this as part of the earlier lecture but it helps to you know represent this again and again so that it gets registered in our memory. So what are they? They are backscattering coefficients. We call them as sigma naught which is nothing but normalized backscatter coefficient and then we have beta naught which is known as radar brightness and then we have gamma which is known as backscattering coefficient normalized by cosine of the incidence angle. And we also saw what is the area that is being considered in the calculation of each of these backscattering coefficients, is not it? And we have also seen the relationship between the different backscatter coefficients. Why are they important? Because for radiometric calibration we tend to use the backscatter coefficients we will see how it is done shortly. So for more understanding let me try to introduce Terrasar X. It is a German SAR satellite mission for scientific and commercial applications which is managed by the German Aerospace Agency, DLR. So the primary objective of Terrasar X is to acquire two dimensional SAR observation in the strip map scan SAR and spotlight mode. Remember we have already seen the different acquisition modes of a SAR that is cancer strip map and spotlight. So now I am assuming that you are familiar with these terms. Once again the primary objective of Terrasar X is to acquire two dimensional SAR observation in the strip map scan SAR and spotlight mode. Now the satellite as such it is at an altitude of 514 kilometers and it has an active radar antenna to produce images with a resolution of 1 meter and Terrasar X has been operational from the year 2008. Terrasar X has a 5 meter long body and a hexagonal cross section. So now let us look at the backscatter coefficients of Terrasar X. Now to get more information about the backscatter coefficients I would urge you to visit this site. Just to summarize what is written here, here dn is nothing but the digital number. Digital number here i and q are the real and imaginary parts of the complex number. Beta naught is represented here and ks is nothing but the calibration factor and theta i is the local incidence angle and this can be considered as a look angle for a flat terrain. So shown here are the representation for beta naught. D or dn it refers to the digital number and then we have sigma naught and then we have sigma naught here represented in decibels. Now for different satellites the backscatter coefficients can be different, can vary, it can vary. So just to approve my point let me take you to the ISRO website wherein we will talk about RiceSat 1 satellite. It is a state of the art microwave remote sensing satellite that carries a SAR payload at sea band. The orbit has an altitude of 536 kilometers and RiceSat 1 was launched in year 2012. So if you visit the ISRO website you can see sample images. It shows parts of Bangalore. You can see prominently the layover effect wherein the tower is shown as tilted. We saw what is layover. You can clearly see the layover effect here and shown here is the Haurabridge and then again the metro station in Bangalore. So the whole point of showing these images is to make you understand that what all are the distortions that you see in a synthetic aperture radar image. Now coming on to the backscatter coefficients for RiceSat 1. I can represent the radar backscatter coefficient in decibel using the relationship that you see in the screen wherein dn refers to the digital number, p refers to the pixel. So it is digital number for a pixel p and then we have something known as a calibration constant in decibels. Theta i it is the incidence angle. So as I am representing it for a pixel I am going to use theta i p and then theta i c is the incidence angle at the scene center. So again to reiterate the whole purpose of showing you these relationships is for you to understand that we can have sigma naught relationship, sigma naught values that differ for different satellite. So we have seen how sigma naught is obtained for Terrasar X. We have seen how sigma naught is obtained for RiceSat 1. See the whole you know point is that it is highly challenging to interpret information from a synthetic aperture radar image because visually you cannot see features as clearly as you would see in an image captured in the visible or infrared region. A SAR resolution cell it can contain a large number of targets or scatterers whose return echoes are coherently summed up to obtain the phase and the brightness of the resolution cell which means if there are a large number of scatterers the resolution cell will tend to show you a brightness value which is much larger than the actual brightness of the object. This causes an effect known as speckle in a SAR image. Speckle we will discuss about what is speckle. Speckle look like in a SAR image an example is shown here. The only feature which you may be able to interpret is the water body. Other than that you can see grainy dots okay which we call as salt and pepper noise that is a characteristic of a radar image we call it as speckle. Radar signals as such you know they tend to hit target at many angles as you saw in an earlier video scattering happens in all the directions okay and radar signal themselves they tend to hit the target at many angles and of course this is dependent upon the incident angle of the transmitted signal the local incidence angle and the looks used in creating an image. I have used the term looks we will see about that shortly. Now the returning signal as you saw in the video the returning signal from a target gets subjected to something known as interference let me write it down interference okay. So the return signal from the target gets subjected to interference and how does it happen there is interaction of these return echoes and this interference pattern involves signal adding in phase and out of phase. So what I am trying to discuss is that whenever a microwave pulse hits a target the return echo will have a different amplitude different phase and when we have multiple return echoes from distributed targets they tend to interfere they can constructively or destructively interfere and this interference results in a noise like phenomenon which we call as speckle that you see in your screen and speckle gives the radar image some sort of a grainy appearance okay grainy appearance. Technically textbooks tend to define speckle as noise like characteristic of a SAR image noise like characteristic but then it is also characteristic of other coherent imaging systems like laser, sonar or ultrasound. So essentially let me try to play this video again so that you can visualize what happens when a microwave pulse is hitting a target return echo scattered in all directions a part of the return echo reaches back to the radar system which is what is registered as the return power which is converted to a complex number for each pixel and given to you for analysis. Now till now remember we have not discussed how a elliptical footprint of a radar system can result in a square or rectangular pixel in an image okay that conversion we will discuss shortly but for now I want you to understand that whenever there is scattering happening in all directions the return echoes they tend to interfere constructively or destructively and they cause some sort of a grainy appearance in the resulting radar image which is known as a speckle. Now consider a distributed target such as an agricultural field okay if I ask you whether the field is homogeneous or heterogeneous you will probably say homogeneous isn't it? Now visually an agricultural field may appear homogeneous on the ground say paddy field it looks all the same and it may even have characteristics such as surface roughness which may appear homogeneous for a large area but when a radar system is looking down on the same paddy field the image is not going to be homogeneous the field is not going to be seen or understood in a homogeneous manner why? Because in a radar image the adjacent pixels shall also exhibit a different echo okay adjacent pixels shall also exhibit a different echo as the scatterer that is as the target causing scatter may not be same in two patches of the ground okay the scatterer may not be same in two patches of the ground which means even if there are you know very small variations in the target location and scatter locations that alone is enough to cause interference. Remember every time scatter occurs both the amplitude and phase information is undergoing a change and the final combined effect shall result in a totally different amplitude and phase. Now for clarity let us use the vector representation of scattered waves okay. So as axis I am trying to represent the real part and imaginary part consider one resolution cell of a radar image just one resolution cell and remember all the animated videos that I have shown you as part of this lecture. So the transmitted pulse from a radar system hits the target and it gets scattered in all directions a few get back scattered in the direction of a radar system and if we consider just one resolution cell of a radar image it is going to have a collection of return echoes collection of return echoes from different individual targets from different individual scatterers isn't it. Which means the total return signal is going to be a coherent addition look at the figure shown the total return signal that is registered by a radar system is it going to be a coherent addition. Now remember a very small random change in the individually scattered wave can also cause a large change in the total resultant wave and this is what results and what is known as speckle. Now how to understand speckle imagine you are in a fair you know in a gathering where a lot of people are there and then you do have loudspeakers in every nook and corner of the areas and sort of a mailer. So consider that both the loudspeakers are emitting the same audio we get to experience an interference pattern remember you as a listener if you are moving your head the pattern of constructive and destructive interference becomes all the more apparent or even if the loudspeakers which are kept at two locations even if they are moved apart this also shall result in a pattern of constructive and destructive interference. So here we are talking about sound waves but I want you to visualize this using microwaves and this example is you will find it commonly provided in many high school textbooks. Anyways now that we know what is speckle and what is the reason for speckle what do we do about it how should we reduce it. So this portion will be part of the next lecture so what you see in front of you towards the left side is a synthetic aperture radar with speckle and what you see towards the right side is the same image but think about whether the speckle is getting reduced here is there any difference between both images or both look same okay this will take up as part of next lecture. So just to summarize what you have been understanding through this lecture in today's lecture we understood the meaning of radiometric resolution and we saw the backscatter coefficients of Teras Rx and Rhysat1 after which we understood the concept of speckle an inherent property of radar imagery. So let me hope that you found this section useful and I will meet you shortly in the next class thank you.