 So, hello, we are in the third lecture of fourth module, last lecture we were discussing on how precipitation can be estimated from radar remote sensing and in particular we were trying to understand about satellite bond radars, is not it? And we also discussed about a few indices, few metrics that can be derived from contingency table to help in the comparative assessment of quality of precipitation coming from satellites, alright? So, we are still trying to understand about precipitation and in this regard in today's lecture we will try to understand about Doppler weather radars, okay? So, weather radars they typically operate in X band, C band and S band frequency. Of course, radars do operate in KU bands and such radars are used mostly for studying cloud physics. If you visit the India Meteorological Department website, you can see the radar network of India wherein S band, C band and X band radar locations are visible. Typically, 6 products are made available and uploaded on the IMD departmental website which are updated at 10 minute interval. And if you see, typically at the coastal locations S band radars are preferred at the coastal stations. There is a reason for it because of the low attenuation. You know the coastal regions you may be aware that they are battered by cyclones which tend to produce heavy rainfall which means if radars of higher operating frequencies are installed in these coastal regions, the radar rays are going to be obscured, obstructed by heavy precipitation and hence effective coverage range of radar will be reduced. Now, I know that a few terminologies that I am discussing now, it may sound a little new to you. We will be covering it as part of today's lecture. So, just to give you a preview of what we are trying to analyze. Further, the different products that are available from IMD weather radar are shown here, okay? You can see that PPI plots that is plan position indicator plots are represented for Z that is radar reflectivity, radial velocity V moving ahead. We also see the plots for surface rainfall intensity, precipitation accumulation and maximum Z. Here Z stands for radar reflectivity. What it is and why is it useful? We will cover it as part of today's lecture. So, you see the observational range of radar, it is limited by the attenuation of radar waves by rain and by the earth curvature effect. Typically, whether radars can detect systems up to a range of 500 kilometers, but then for velocity measurements the effective range is 250 kilometers, okay? So, let us try to understand the working principle first of a Doppler weather radar. So, in the case of DWRs, in phase coherent electromagnetic waves are transmitted by a radar transmitter through a directional antenna in a focused manner. So, by now I am assuming that you have a better understanding of what is coherence, what function does a radar transmitter do, what is directional antenna and what are the interference pattern that can be caused, okay? So, let me reiterate. In phase coherent electromagnetic waves are getting transmitted by radar transmitter through a directional antenna in a focused manner. So, a part of this transmitted energy gets absorbed by the atmosphere, okay? And a part of it travels further through the atmosphere and a fraction of this transmitted energy is getting scattered back by the targets. And the amount of received power depends upon the radar parameters like the transmitted power, radar wavelength, horizontal and vertical beam width, scattering cross section of the target, atmospheric characteristics, etc. So, echoes that are coming from clouds, okay? They also depend upon something known as a drop size distribution and also the physical state of precipitation, you know, whether it is rain drops or snowfall or hail storms, etc. So, the amount of return power gives us information about the intensity of weather systems. And the azimuth and elevation of antenna gives information about the location and height of cloud systems. Remember, always in radar the travel time of electromagnetic waves they are used to estimate range or distance from the radar, alright? So, now what we will do is we will try to understand a few terminologies that I had initially introduced, okay? Starting with radar reflectivity. It is denoted by capital letter Z. It is nothing but the volume sum of back scattering cross section. So, what is back scattering cross section? We had discussed as part of module 2, radar reflectivity, volume sum of back scattering cross section. You see, there are different relationships of radar reflectivity and rainfall rate which has been established through measurements of drop size distributions. And in here, the drop size distributions have also been modeled by exponential functions. And several ZR relationships have been obtained for various regions as well as for various storm types, okay? So, whenever we try to analyze Doppler weather radar data, we look out for radar reflectivity because it has a relationship with the rainfall rate. And empirically, these relationships have been estimated for various regions and for various storm types, alright? Moving on, the slide in front of you shows frequency shift which is related to motion of target, okay? Minus 2 into radial velocity by wavelength gives the relationship. Again, we had subtly covered it as part of earlier lecture. Just for refreshing your memories, you know, objects with negative radial velocity are travelling towards the observer. Objects with negative radial velocity are travelling towards the observer whereas those with positive radial velocity are moving away from the observer, okay? Fine? You see, hydrometeors are the name given to precipitation sized particles in the atmosphere, okay? Hydrometeors. The round trip propagation time between a radar and near boundary of volume of hydrometeors causes a delay, okay? The round trip propagation time between a radar and the near boundary of volume of hydrometeors causes a delay and resolving individual echoes are not that easy or not that possible. Therefore, the weather signal is sampled at discrete range time delays, okay? Range time delays. Given here is tau s, assume it as the time delay of a sampling gate, then the approximate range of scatterers that contribute the most to the sample of weather signal is given by c tau s by 2, okay? So, now I am a little confused because you know, does measurement of one parameter alone help in the estimation of precipitation? Liquid water content and cumulative precipitation amount is that enough? Just one parameter, z alone, is that sufficient to estimate to quantify the precipitation? Typically, if you take any textbook on radar meteorology, you will find that z r relationships are most commonly used to estimate the precipitation rate. And for many years, radar meteorologists have also tried to estimate an empirical relationship that can link r to z rainfall rate to radar reflectivity. But unfortunately, there is no universal relationship because you know, again to give you a little bit of information about the drop size distributions because you know, the measurements of DSD around the globe have been made under different climatic conditions, okay? And here it is also important to mention about calibration, okay, what we discussed in the last lecture. It is quite difficult to calibrate radars to within a decibel and there could be some systematic bias in the radar measured reflectivity. So, whenever you are using an empirical relationship of z and r to estimate rainfall, please be aware that there could be some systematic bias in the z, that is radar measured reflectivity. And you know, some of these errors, some of these systematic biases can be compensated by choosing an appropriate relationship linking r and z, okay? With this background, let me ask you a question, you know, how does a radar distinguish between ice and liquid phases of precipitation? Because especially when we talk about convective storms where liquid water can exist at temperatures colder than 0 degree Celsius and ice can be found at temperatures warmer than 0 degree Celsius. You know, especially in convective storms, when you find both the phases, ice and liquid phases, how does a radar distinguish between both? Because it is equally important to quantify rain, snow, hail rates, okay? And radar no doubt, it has an immense advantage because it can survey vast areas and it can result in millions of measurements in a fraction of minutes and also they are efficient in measuring more than one parameter. Now, the need is for us to accurately characterize the relationships between the density of water that is water density and the cloud dynamics with good spatial resolution as well as to sense the threat of unusual but significant events such as flash floods, hail storm or heavy snowfall, okay? So, with this background, let me introduce you to a few sources of noise in Doppler weather radar data, okay? Remember, noise to you can be information to me, okay? Now, these are the general sources of noise and depending upon application, your noise can be my information. So, we have noises in radar data that can come from biological echoes. For example, there can be a swarm of insects flying or birds and bats flying that also get detected as radar echo, okay? Biological echo, we call it as. We can have echoes that come from aircrafts at higher elevations. We can have something known as a sea clutter that is the echo reflected from the water body in the presence of waves, sea clutter. We can have refraction, unusual rates of refraction in the atmosphere that can cause anomalous propagation of a radar beam and of course, the ground clutter that is whenever there is an increase in high rise buildings, the echoes get reflected from targets of earth such as trees and these high rise buildings. They cause the ground clutter. So, quality control algorithms form a very important part of pre-processing Doppler weather radar data. You know, before it can be utilized to estimate precipitation, it is very essential that we clean the data, which we call it as quality control of radar data. So, just to give a visuals to whatever we discussed just now, whenever we have clouds with hydrometeors that is precipitation sized particles, normal propagation is shown in front of you but sometimes there is abnormal propagation wherein you know I am exaggerating the abnormal propagation here. Just to hint at the idea, just to drive the idea to you that there is abnormal propagation possible due to significant variations in the refractive index of the atmosphere because of which you know the radio waves they tend to hit the surface of the ground or the sea. Now, with this background, let me show you how a DWR data looks like. If you try to open the file of Doppler weather radar data from India, say you pick one station and you get the file from IMD and you open what you see as attributes is shown in front of you. So, you have different attributes that are tabulated here and there is something known as a short range scan elevation angle and long range scan elevation angle. See, in the case of radars in India for the example data that I am discussing, they collect information at different elevation angles. So, say the first scan 360 degree scan happens at 0.2 elevation degree. So, the radar scans at 0.2 degree, it slightly increases its elevation, it scans at 1 degree, slightly increases elevation scans at 2 degree and so on. So, there are multiple elevation angles at which 360 degree scan is happening for a Doppler weather radar data given here are the pulse width. So, by now I assume that you know exactly what is pulse width. The temporal resolution is 10 minutes, S band frequency is used again. This is not generic to all the DWRs. I am just mentioning about a radar data, sample radar data file if you open the attributes what you will see inside. Shown here is the horizontal polarization. So, now we have the understanding that for Doppler weather radar data, it scans at multiple elevation angles and there is something known as a short range scan and long range scan. The temporal resolution is aware, we are aware of it 10 minutes and the frequency used here is S band. So, one more point that is worth discussing is pre-processing of DWR data. You see before the data can be used for any kind of modeling or analysis, as I said before some amount of cleaning is required and we discussed that it can be to remove the noise in the Doppler weather radar data that is one form of pre-processing. Second is if you look at the diagrams, I showed you PPI plots, is not it towards the earlier slides. So, Doppler weather radar data pre-processing includes something known as a coordinate conversion, thinning of radar data to model resolution and finally quality control. So, let me reiterate pre-processing steps are coordinate conversion, thinning of radar data and quality control. And by looking at the visuals, you can get an idea that DWR data are available in coordinate system wherein each data point is represented in azimuth range and elevation not XYZ, azimuth range and elevation which means I need to transform these data points to Cartesian coordinate system. If you are an enthusiast interested in processing radar data, feel free to implement these in programming environment to pre-process the DWR data especially for coordinate conversion. Right now, I am not going to go into the details of the equation but they are explicitly explained in the slide. So, all in all, the geolocated data can be thinned to model resolution using interpolation techniques and then finally, the radar data can be written in a format like ASCII. It can be written in a format that you are comfortable with for modeling. So, these are some of the sample plots that tell you how radar data looks like. So, towards your left side, you see a cappy plot that is constant altitude plan position indicator, CAPPI, constant altitude plan position indicator and towards your right side, you see the range height indicator wherein X-axis is range and Y-axis is altitude. Remember this, the examples about the attributes and how the data looks like. So, all this is taken from the single station single file in S-band. Again, what you see here is the reflectivity image, reflectivity Z, reflectivity image and shown here is the elevation. You know the data what you see has been captured at an elevation angle as displayed on the screen and towards your right side, you see the velocity image. So, in your tutorials, you will be covering in detail about how to work with radar data and how to extract the attributes and how to create plots. Let me ask you a basic question. So, we do have a radar data that gives highly dense point clouds, highly dense information about weather, but then how do we use it? Can we use it for forecasting? Can we use it to predict that there will be rainfall in say the next one week? So, this is where I would like to introduce something known as a now casting, which stands for very short range forecasting. You know the Doppler weather radar products of reflectivity and radial velocity and spectrum weight, all these are the products that you get from radar. So, they can be used for thunderstorm now cast, especially the NWP that is numerical weather prediction now cast products are uploaded by numerical weather prediction division at New Delhi. And the data used involves the current weather data from automatic weather observing system, the automatic rain gauge data, satellite data and so on. So, you know the couple of my slides are purposefully intended to audience who are interested in research, you know. So, I have tried to mix the contents so that you get a fairly good idea of how Doppler weather radar products are being used in India. So, shown here are a list of operational and research systems. Again, you can explore more if interested and here I am going to mention in particular about swirls. So, swirls analyzes the radar echo in two successive radar scans and then computes the direction and speed of movement of rainfall areas. It obtains information for the distribution of high resolution radar echo motion distribution and these are just to summarize the research systems that have been used in different countries. And we have something known as a warning decision support system which is developed by National Severe Storms Laboratory and Cooperative Institute for Mesoscale Meteorological Studies at Oklahoma and it has been shared with IMD in 2006. So, the WDSS II components, it consists of a suite of algorithms for interrogating the radar data and it gives you the 3D earth and time centered display for visualizing as well as it has an application programming interface library in C++ which supports algorithm developers. Again, if you are interested to use this for your research areas, feel free, the website is given in the slide. Now, in the recent years, there have been increasing trend of extreme precipitation events throughout India, is not it? Heavy rainfall, increasing trend has been observed throughout India. So, the occurrence of these extreme precipitation events can be predicted using short range forecast which are generated for a period of 48 hours, 2 days. So, what I will do is I will quickly discuss a case study using DWR which we conducted. Before that, have a look at the right side. According to Browning 1980, short range forecasting can be performed in many different ways. Remember, short range forecasting 48 hours. Forecast using synoptic scale numerical weather prediction model with model output statistics. Forecast using mesoscale numerical weather prediction models and forecast through extrapolation of Doppler weather radar and satellite observations. So, for this case study, we have picked the heavy rainfall event which occurred in the Chennai city from 30th November to 2nd December 2015 and it occurred due to the propagation of convective systems from the West Pacific Ocean to the Indian Ocean. And the convective systems are guided northwards towards Indian land region by two highs of mid troposphere situated in the east and west of the Indian region and the high to the east was due to the effect of El Nino and that in the west was associated with a global shift of upper tropospheric cross bi-waves. So, what we did was we could obtain the Doppler weather radar data for this event for the city of Chennai from India Meteorological Department. The summary of data used is given here with the plots of reflectivity as well as radial velocity. You can see the location latitude longitude and the site height was 35 meters above mean sea level. The short range scan elevation angles are displayed. You can see the beam width, the temporal resolution and 3 GHz frequency was used. Polarization was horizontal. So, what we did is we tried to use the Doppler weather radar data in data assimilation. The whole purpose was to analyze the impact of background error statistics on Doppler weather radar data assimilation within a 3D bar assimilation system. So, once the reflectivity data after preprocessing was used in a model known as weather research forecast model at a resolution of 10 kilometers and several experimental setups were created by us which gave us a 6 hour accumulated precipitation what you see. Visually the output that we could get from the model was compared with that from GPM. And then finally, we could find that zonal and meridional winds when used as control variables improves the scale of 3D bar radar assimilation system. Now, if you are interested to know more about the mathematics and details of this work feel free to read the article as discussed. Now, as I mentioned earlier this lecture is a blend of fundamentals as well as research as in how do you use the Doppler weather radar data available in India for your research. So, if again if you are more interested feel free to browse the research articles in this area. Let me hope that you could understand a glimpse of the work that is being carried out and I will see you in the next class. Thank you so much.