 Welcome to tutorial 12 on hydrologic modeling using microwave remote sensing. So till now throughout each of the tutorial sessions we have been learning about how to access and how to work with different satellite products mainly we dealt with precipitation, soil moisture, water levels from ultimetry and also we have been individually addressing how to work with each of these variables. So in this particular tutorial we shall see how the individual puzzles fit in with respect to hydrologic modeling. So firstly let us try to understand about hydrologic models as in what they are and what do they give us as output and what is their relevance in hydrology. And let us try to understand this through the water cycle defining the underlying hydrological processes, modeling them, running the model and validation of output. So what you see here is a small caricature of the different processes which are part of the hydrologic cycle. Let us try to understand them in detail. So water cycle basically it refers to the movement of water between ice, oceans, atmosphere and freshwater. And we know that it consists of several processes like we have precipitation that is condensed atmospheric water falling to earth, precipitation. We have evaporation that is transition of liquid to gas. We have transpiration that is movement of water within and out of plants into the atmosphere that is transpiration. We have infiltration that is water entering the soil from the ground surface infiltration. We have runoff flow of water from the earth surface that is runoff. Then there is interflow that is flow of water within the soil layers interflow. And then we have something known as routing that is movement of water downstream. Just to summarize each of the components individually now with respect to precipitation we have already seen that it can take many forms and it can be measured either at a point using gauges or it can be estimated over an area using Doppler weather radars or satellite bond sensors. So typically a hydrological model is fed with precipitation which is gridded at regular intervals throughout the length of simulation. And what you see here is the precipitation data set from IMARC. Now to get gridded precipitation data we could either grid the rain gauge data or grid the data from weather radars or satellite sensors. And we even have multi-satellite precipitation products that are available at every say 3-hourly intervals. Again not all the precipitation that falls onto the surface of the earth not all of them reach the ground. Some may evaporate back into the atmosphere, some may get intercepted by the canopy of plants. And hydrological model usually deal with these in different manner of which say one solution is to make the intercepted rain part of the modeled evapotranspiration process. So evapotranspiration basically is a combination of evaporation and transpiration and it is a general terminology used for phase conversion of a liquid into a gas wherein the liquid is water. So as I mentioned earlier transpiration is the transfer of liquid water from plants to the atmosphere as a vapor. And typically evapotranspiration abbreviated as ET. You know it is hard to measure directly as it varies drastically across smaller scales because it also depends on the species of plants, their age, their health, the sun angle, the temperature, humidity, wind, land cover. So all in all ET is not easy to measure directly but then it can be measured experimentally but then it is again expensive and subjected to error as well. So usually we calculate it indirectly. For example if water in equal to water out then the change in amount of water I am going to term it as s. The change in amount of water must equal the precipitation P minus ET that is evapotranspiration minus G that is ground water minus Q that is stream flow. So of these we can measure easily the Q, the G, P and S which means we can solve for ET. So this is an indirect means by which we estimate ET but there are flux towers that can be installed which give us more accurate information about ET. Moving on now let us come to potential evapotranspiration abbreviated as PET. So PET is typically expressed as a water depth. Now the amount of water, this amount of water must always be removed from the model. So if there is enough precipitation available we use that, else we use other sources. Now the steps to deal with PET are as follows. Shown here is a small caricature wherein you can see a bucket or a cylinder with precipitation and then we have potential evapotranspiration, surface water and ground water. So assume the PET takes water from precipitation cylinder until the PET cylinder is full. Say it is not full, the PET cylinder is not full, then it takes water from rest of the model, rest of the model that is surface and ground water until it is full. Now once PET is full it means the remaining precipitation can flow to the rest of the model. So what are we trying to do here? We are trying to understand a hydrological process namely a water cycle through the eyes of a hydrological model. So let me reiterate for potential evapotranspiration through this small drawing. So PET takes water from precipitation cylinder until the PET cylinder is full, say it is not full, then it takes water from rest of the model which can be surface water or ground water and once PET is full, the remaining precipitation is available to flow to rest of the model, which means at this point our PET requirement has been satisfied and then the precipitation is free to reach the land surface. Now moving on, so of the precipitation that reaches the land surface, some of this will drain into the soil and some other portion will act as runoff. Now to understand runoff imagine a walkway, say a slab of concrete or asphalt or any surface that is impervious to water. If water cannot be easily absorbed by the surface, it will contribute to runoff. Now just to compare, consider a pervious surface like soil can be slightly tricky here because soil can absorb water but up to a certain extent, up to a certain point, after that the water starts becoming runoff. So here I have defined runoff as water which does not percolate or infiltrate into the soil but flows over the land surface. Now infiltration is the part of water that percolates as the name suggests which infiltrates into the land but then how do we decide how much is going to contribute to runoff and how much is going to contribute to infiltration. So it depends on the properties of soil and in the case of hydrological model of course on its settings. So moving on, till now we have covered the few different processes which are part of water cycle such as precipitation, evaporation, transpiration, infiltration and runoff through this small diagram which outlines each of these processes. Two more are remaining that is interflow and routing. So let us try to understand them. Now as I mentioned earlier from the portion of precipitation that reaches the soil, variable infiltration curve tells us how much is excess rainfall and how much is infiltration. Now excess of rain can materialize as overland runoff and interflow. For example when part of precipitation reaching soil I am going to call it as P soil. So when part of precipitation reaching soil is less than infiltration rate of soil all of P soil becomes interflow and then there is no overland runoff. So we have part of precipitation reaching soil and then we have something known as an infiltration rate. So if P soil is less than infiltration rate all of P soil contributes to interflow. There is no overland runoff. And higher the hydraulic conductivity of soil more the P soil becomes interflow and less available as overland runoff. So interflow and overland runoff can be routed. So routing basically means we are moving downstream, moving it downstream through separate model processes. So both interflow as well as overland runoff can be routed downstream through separate model processes. So just to reiterate of the water entering the model. So now I am going to use the term model instead of land because we are trying to understand how a hydrological model understands a hydrological process taking water cycle as example. So of the water entering the model some contributes to evapotranspiration, rest reaches soil and contributes to infiltration, excess rainfall again can be overland runoff or interflow. What next? So to account for say overland runoff, overland flow which is flowing towards me. I can add it to P soil which means it is going to be like precipitation falling which can be evaporated, transpired, infiltrated, converted to interflow or overland runoff. So just for the sake of better understanding two cylinders are shown here. Assume the color in blue is precipitation from upstream overland runoff and whatever you see in green is the new precipitation from forcing at downstream. Okay? Now each type of flow shall be routed until it reaches a river channel and then it becomes open channel flow. For overland runoff, routing is a function of roughness of land. So for flow of soil it is dependent on soil saturated hydraulic conductivity for interflow. So the small terminology is to help us get an overview, get an overall picture of what a hydrological model does. So in the previous slide overland flow is accounted as additional precipitation when we are not in a river channel. So assume that it is a river channel then what happens next? We have to account for it by adding upstream water to overland runoff. So here what you see in blue is the old infiltration from upstream interflow and here what you see in green is the new infiltration from variable infiltration capacity process at the downstream. Again you know in the same example we can assume that now there is also interflow. Okay? That is water under the surface which is flowing towards semi. We have to account for that as well by adding it to the infiltration water at new downstream point. Okay? So here blue is the overland runoff from upstream and green is the overland runoff newly generated at downstream. Now in real world we know that water travels both horizontally as well as vertically through land and we try our best to choose a hydrological model that realistically or closely represents the real situation as best as possible. And upstream runoff as was discussed in last 2 to 3 slides can be treated only one or other way not both. But infiltration modification process tends to occur whether it is a river channel or not. Now let us try to put the pieces together. So computers think in terms of grades and cells which means the area of interest that is the region of interest over which you are running the hydrological model gets divided typically into cells to model and as users we get to select the size and the number of cells. So here what have I done? The earlier diagram we have been seeing I have divided it into cells just to give you a feel that in a hydrological model computers think in terms of grades and cells and ultimately water cycle is turned into a series of model cells as shown here. Now in each of these processes there are factors like hydraulic conductivity and roughness etc and we often adjust these within a model and this is known as calibrating the model calibration. So calibration influences heavily the quality of simulation or output from a model. So what is calibration? It is a process by which you adjust the parameters. Now if the parameters can be adjusted for each cell independently we call it as a distributed model and if this adjusting parameters if it can be performed in a few spots over a river basin we call something as a semi-distributed model. So you know let me try to explain through this simple flow chart that tries to summarize the different types of hydrologic model. We have stochastic model, deterministic model and then we have models based on process. Deterministic model themselves can be either lumped or semi-distributed or distributed and based on process we have empirical model, conceptual model and physical model. Now with this flow chart in mind let us try to understand each. So if the parameters that we discussed earlier if they can be adjusted independently we have something known as a distributed model. Now if adjusting parameters can be performed in a few spots over a river basin we have a semi-distributed model and say we have just one constant set of parameters for the whole area just a single constant set of parameters then we have something known as a lumped model. So lumped model the parameters basically do not vary spatially within the basin and the response is evaluated typically at one outlet. It is not applicable for event based processes but then it is easy to use at it as it requires minimal data. For example the SCN based models they are lumped models. Now as far as semi-distributed models are considered the parameters are partially allowed to vary in space and that is done by dividing the basin into a number of smaller sub basins. They are of two types that is kinematic wave theory models like HEC HMS probability distributed models wherein probability distributions of input parameters are considered across the basin. For example SWAT is a semi-distributed model we will look at SWAT shortly and then we have deterministic hydrologic models. So coming on to distributed models. So in distributed models the parameters are fully allowed to vary in space at a resolution which is user defined. Let me reiterate in distributed models the parameters which we discussed earlier they are allowed to vary completely which means this requires a very large amount of data but then the spatial resolution can be user defined. Examples are Mike Levenchy model etc. Now we also have physically based models, physically based deterministic models which help us gain better understanding or clarity of hydrologic phenomena and it is based on the laws of physics say conservation of mass, momentum, energy, continuity equation, the equation of motion or equation of energy. So one or more of these are used in physical model development. I shall not be going through the detailed equations right now but we will try to understand it through the NOAA land surface model. We were trying to understand the different processes which are part of the water cycle and how a hydrological model understands this and then we also had a quick overview of the different types of hydrologic models. So as next part what we will do is we will try to understand more about as an example the SWAT model as well as the underlying processes or equations that are embedded in the NOAA land surface model. Let me hope that you found this section informative. Thank you.