 Welcome to the NPTEL course on remote sensing and GIS for rule development. This is week 12, lecture 4. This is kind of the last application that we will showcase before we go into the summary and discuss about how the weeks are interlinked and what the future steps for this course could be. As I have always mentioned, this is not just a learning of remote sensing and GIS, but very specifically for rural development. And as the NPTEL reviewers requested, multiple facets of rural development were discussed. We also made sure that students who learn the concept of rural development engage with datasets that are easily available. Therefore, open source data was shared and open source GIS platform was taught. All the exercises were done using open source software. On the same note, we will be looking at the last applications for this NPTEL course, which is creating indicators and dashboards. So, as always you know that remote sensing for rural development has many case studies in terms of how we want to use remote sensing and at what spatial and temporal scale. And we did discuss that multiple indicators are available. One such indicator that we went in depth was NDVI. And also we looked into NDWI. However, we have to understand that not all datasets are and indicators are readily readable by policy makers. So, unlike the other remote sensing and GIS users, this is very different because we want to improve or serve the rural committees. And for that, we need to make sure that those decision makers and policy makers should understand these indicators. So, the indicators can be very specific or applied. There can be one individual indicator like NDVI or a combination of indicators. We looked at this when we looked at the database for indicators. In addition, these indicators can be used to drive models and can be used as a dashboard. So, what is a dashboard? We are using dashboards without calling it dashboards in many instances. All the mobile apps you use have an inbuilt dashboard. The dashboard will have multiple buttons. You are allowed to select and then put in an area of request or an interest and then you see how the output is given. We have looked at the water quality indicator and we mentioned that the model was developed using linear regressions and then calibrated and validated. That's it. We just gave it as an equation. So, how can an equation be helpful for a policy maker? They are not going to run these things. They need it as a visualization as a result that pops out and that is what a dashboard does. So, in the dashboard on the behind scenes, the algorithm will be there. The linear regression will be there. All the person has to do is just click on what area of interest and then boom, the dashboard comes with results. So, this is similar to the Google Earth Engine Sentinel Hub that we were using wherein instead of downloading the data and doing the data for indicators, all of them are already there on a dashboard. You just click some buttons and then the results are populated. While this has been a very extensive task in the past, now it's almost very easy to run these dashboards on open-source systems. In the previous times, you need to have a database server to hold the data for your dashboard and now you can just rent it up. You can just buy a portal and then you can put it for some time during your project, very, very cheap, low-cost or you can also use open-source systems like Google Earth Engine, OS and Mapper trackers where you could just put your data there for a long time and then there's a dashboard that can be used for it. So, let's look at the other benefits. Data may be easily fetched by optimizing the results. So, instead of looking at the generic dataset, you can optimize the results by using different datasets and your own dataset in some instances and then put it into the mapping. Dashboards are easier for decision-making and that is why they could call as a decision support system DSS or a decision support tool DSTs and these are very important because, as I said, the policy makers may not be learning all these techniques and what remote sensing is, but definitely they know what on the ground it means. So, by clicking different buttons and figures, you will be able to get that support through the dashboard. Let's take a look at some examples, especially the ones from my team because I do know, as I said, the constraints and why it evolved and we will discuss. One important indicator that we developed is the remote sensing-based ecological index. It was based on the framework or system dynamics concept of pressure state response framework. We will show you what it means. So, when you have a pressure and this was done to evaluate or support the IWMP Mandrega projects. So, in the Mandrega, there is a scheme where the farmers are kept without migration for 100 days by some nominal wage. So, minimum wage is given so that the farmers don't migrate to urban centers. So, slowly what has happened is that time of the farmer has been used for IWMP programs. What is IWMP? Integrated water management programs and plans. So, these plans and programs are now being used for increasing water storage, soil moisture, etc. But what is the benefit? So, that is the pressure and the state, you see the arrow mark going, the black arrow mark, the state is the ecological status. We wanted to see how that impacts the ecological status. This was a work done by one of my students, Shivanand through his master's project. And you could see that how the system dynamics approach was used and different key indicators and players are going to come up. So, the response is yes, the pressure is the Mandrega projects and then it has some impact. We don't know what, but it has some impact on ecological status. Normally, it's a positive impact because we are letting people work for the nature rather than against the nature. So, the response could be change in production, crop area, soil moisture, etc. And then that can also come back as a pressure on the system. A pressure is positive and negative also. And we wanted to see how that keeps the cycle going on and on and on. So, this framework was used for the Mandrega IWMP projects to showcase better healthier vegetation, increased soil moisture and crop production through the better use of Mandrega. So, if this is correct, then all the states can adopt Mandrega for better management practices rather than just paying them and not following up on their time. So, this can actually create ownership for the program and also make them work for the nature. So, this creates an improved land surface ecology. Ecology is the living organisms at the land surface and mostly it constitutes the soil living organisms and also the plants. Equal environmental changes can be looked at as other characteristics of land surfaces because if the ecological is good, activity is good, it can impact positively the land surface as well. And then these are theory so we can use pretty old ones. And then we have the ecological indicators of these characteristics include moisture, greenness and dryness by zoo. So, these indicators are moisture, greenness and dryness but how you establish these indicators is also key for which we will be using remote sensing data mapped using aggregated remote sensing index. As I said, we will be using these index in the following part. So, let's discuss how the RSEI model was developed. So, RSEI stands for remote sensing based ecological index. So, remote sensing is the key and the ecological index is done for a rural setting. So, that's where the rural part comes back into picture. Yeah. So, what happens is RSEI creates 15 remote sensing indices. So, the first literacy review was done to analyze how many very sensitive parts are there for this model and it was found out that there are at most 15 remote sensing indices and that the RSEI is a function of moisture, greenness and dryness and heat. So, this was as found by the previous paper. So, while this is a function, how do you account for moisture? So, moisture we knew that in the previous exercise we did NDWI. So, same way we could do NDWI for the moisture. For the greenness we can have the NDVI and dryness and heat could be coming from climate indicators. So, on that note, there could be 15 remote sensing indices representing ecological indicators of land surface. So, all of this can be put as a function to create the RSEI and then again the greenest moisture dryness heat to assess changes in characteristics of the land. So, this is very important and there are multiple indicators. So, based on these functions of themes moisture, greenness, dryness, we have multiple indicators and these are the 15 indicators that were selected from Literature Review as IWSI, MSPSI, NDWI, MAVI. So, this NDWI we have already done. Then we have NDWI, EVI, the vegetation index, OSAVI, NDWI, FVCI, TCR, IVCI and then we have a dryness plus heat together as one theme because both of them could be functions of temperature such as NDDI, NMDI, NBDI and dryI. But the issue here is what we did extra modeling effort is all indicators may not have equal weightages. So, if you see this equation here, if you see this equation here, you have a function and the function says it is a moisture, a function of moisture, greenness and dryness and heat. However, we do know that moisture and greenness and dryness heat may not have the same impact on RSEI. So, there should be some variable in front, a parameter, a normalizing function coefficient, we will say, in front of moisture to show that it is weight more or less depending on the area and the issue. So, on this note, we were looking at multiple, multiple schemes and looking at multiple indicators. Then using the PACA approach, which is the principal component analysis. So, to be honest, you can start with more indicators, you do not need 15, you can have 30, 25, etc. You can just put a number and then for your particular area of interest, you can run what is the most valuable or sensitive indicator and that indicator will be having a higher weightage. So, for example, my moisture, all my moisture indicators are having high weightage except SI, WSI. So, then you can put SI, WSI as weighted 0.1, where the others are 111. So, you see how the value of SI, WSI will go down if you put it in the same function. So, that is what putting weightages will do for your dataset. And then what will happen is, we will be looking at the database of how many water conservation activities were going on, drought proofing activities and the renovation of water bodies. These are all the MG NREGA work under IWMP. So, there are a lot of work going on, but we do not know how it is impactful. So, there is a dashboard that we created under the Rudra Lab, where we could see that these are multiple indicators that can be helpful for looking at the impact on the work done through MG NREGA and then IWMP. So, this has been published as a page paper as index-based impact monitoring of water infrastructures and climate change mitigation projects, a case study of MG NREGA IWMP projects in Maharashtra. So, we have selected Maharashtra because while the student was working in Maharashtra, we were able to get a lot of data that is very important for this project. So, just not getting data from remote sensing is key. We also needed to get and evaluate these schemes. So, these schemes are present, but we have to make sure that they actually were done properly or if it is just a number of let us say 7286, we need to be careful if that actually has some potential on the ground. So, as I said, this paper was published in Frontiers in Water and let us look at the methodology. So, the methodology says that first we wanted to look at the satellite data. Let us look at the left side first which is the crop production and productivity data. As I said, we wanted to see if the MNREGA projects had an impact on the crop productivity. Why? Because the crop productivity is kind of a measure of the ecological status. Only when the soil and the micronutrients, the organisms are healthy, there will be a healthy crop production. So, it is kind of working backwards up and you will see if we have multiple themes for this in terms of finding which root is better to attain crop production. So, we assume that MNREGA work will definitely impact the crop production and on that hypothesis, this methodology has been done. So, we have to collect the crop production and productivity data from 2008 to 14. So, for example, in my Dahu paper that I showcased, we did not have that data for that because the farmers, as I said, are very, very small-scale farming. They did not document all these things. But here in Maharashtra, it is a very progressive area where a lot of agriculture is going on. So, they always have good data. And more importantly, it could be the sugarcane which is very, very important for the state of Maharashtra and the yield that they get. So, looking at this, so first they take the crop productivity data and then they do some correlation analysis to find some relationship with the satellite data. But before that, we need to look at how do you get the satellite data. So, Landsat 7 was used because that has the data range from 2007 to 2014 time frame. And then some data pre-processing and post-processing was done. Remote sensing indices were then extracted out to support the ecological status assessments. And all the 15 indicators are there. So, all the 15 are driven by the Landsat data because there's multiple bands and the bands are enough to take this data out. So, that is what the data we would be taking at. And then we had the PCA. As I said, the PCA part comes up where we had the weightage for each determinants and indicators separately assessed. For example, if you can if you can model the crop productivity by using seven indicators, not 15 indicators, why would you use 15? So, that is one of the other reasons we wanted to use PCA so that to see that if five indicators are very powerful and then three are very less powerful, then we can negotiate the weightages and or neglect the weightages and then put zero, which means it cancels out. Zero is also weightage. So, that PCA exercise was done. Then the RSI estimation was done to determine lands surface ecological conditions, which decides the crop productivity, which improves the crop productivity or impacts the crop productivity and then goes to correlation analysis. So, you get the left side, which is just the crop data productivity from 2008 to 2014, whereas the lands are data one year before we always take for pre-conditions. So, we have 2007 to 2014. Remember that the crop productivity depends on the previous year's land. So, for example, 2008 crop productivity is taken. That should depend on 2007 status of water, land, surface ecological status, etc. So, that is what was taken and then quickly all these were analyzed for correlations and then comparative accuracy, assessment to determine the best index of percentage of all the ecological of land surface, which impacts the crop production. So, as I said, Mandrega, through Mandrega there is a lot of activity done and the activity can impact crop production. So, we took the crop production data then we established the indicators for Mandrega work, take the indicators and put weightages on the indicators and only those with high weightages were used. So, at the end of the day, what would happen is you have a temporal RSEI assessment on the study area with using Sentinel-2. Why would Sentinel-2? Now we know that which bands are needed for the indicators and now we know also that it has some correlation with the crop productivity. So, after 2014 it is smart to use the new data set with the same wavelengths or bands. So, the number of the band would differ because Sentinel-2A has high spatial resolution and temporal resolution. Landsat was at 30 meters 16 days whereas Sentinel is at 10 meters 6 days. So, we have better spatial temporal resolutions and also it has multiple bands which are sometimes higher than Landsat depends on what you measure. So, in that case, it is better to use the same algorithm for a particular study site and then use the updated satellite source. But as is the impact Sentinel-2A was used by created assets that are created during the Mandrega, IWMP, etc. So, the assets were already showcased in this part. So, on the top we have the water conservation assets, drought proofing assets and then rejuvenation of water bodies, this kind of your tanks and then malas, canons, etc. This gives you all the indicators like a database we always have and then some papers that have used it or devised these indicators are on the right side. So, what this index has used, you can definitely look into it. And the RSEI was done for the entire state of Maharashtra. So, before that the number you see under the district is the number of IWMP schemes or projects that were completed from 2007 to 2020 and then now it's compared to RSEI. So, if the RSEI is higher, you could see that highlighting applied green and controlled on brown sites in yellow markings. If you see the RSEI is pretty high, then it is green in color because ecological status is improving whereas brown color indicates a very low RSEI. So, what you could see here is Ahmed Nagar and Solapur are very similar in geological and environmental conditions, which is very important to have. You cannot take two study areas very different. It has to have some commonality or most commonality. So, the rainfall, temperature, the slope, the gradient, all are the same between Ahmed Nagar and Solapur. However, there has been more IWMP projects in Ahmed Nagar. So, look at it. It's almost 10 times. Solapur has 2,800 schemes, projects running, whereas Ahmed Nagar has 20,028 projects. Running, completed, etc., etc. So, even while running the project or establishing the project, there is a considerable improvement in the ecological status. For example, you're building a check dam, a series of check dams or a cascade dam. So, you build one by one. However, while you build one, the impact is still coming through the picture. That is what I mean. So, the higher color, the green color indicates a higher RSEI, whereas the brown indicates a lower RSEI. And you could see that that beautifully correlates to the number of schemes that are present. And that is almost similar stories for across. However, we have to make sure that the geological conditions are the same. Otherwise, you're comparing apples and oranges. You need to compare apples and apples. So, that is what we have done here. Otherwise, for example, I could have compared Ahmed Nagar to Bandar or Gondia or even Garchiroli, which are in different colors, not green in RSEI. Why did we do this? Because we cannot have this control there because the rainfall is the same in here. Whereas the rainfall is much drier, maybe on this side. And that cannot be a ethical comparison. It has to be same, the area, the product. So, when you do remote sensing, please make sure that you cannot show that one area is getting more rainfall. And then the crops are growing well, whereas the other areas are getting less rainfall and crops growing well. And that could be because the soil condition is good, the groundwater potential is good, all those things. So, please make sure that both these two areas that you compare are of the same nature. And as here, it's the same nature. The only thing different is the IWMP numbers. It could be because of the budgets that were allocated the population of farmers or other externalities because Mandrega money is directly linked to the number of farmers. So, we have established a fact that this RSEI can be used as a tool for establishing the benefits of Mandrega work. And these dashboard that Shiva created could always get updated easily on the open source mapping dashboard software. So, this is an open source dashboard. He doesn't have to pay for it. All you just do is VMAP kind of thing where you, my map, you put all these values in and you keep updating and then the data set gets updated on the map. There's no payment. However, there's only a particular storage you can use. It's like a Gmail where you can use a particular storage afterwards you have to pay. Let's go to another dashboard that another student had created through the GIS and work. So, this was looking at malnourishing indicators in rural India, especially the different blocks. And what you could see that if you put it as numbers and tables, it doesn't look that great. But if you put it as a image with blocks colored differently based on the percentage, let's say here, wasting is very high as red. And then 10 to 15 percent is high prevalence, 5 to 10 percent is medium, and then less than 5 percent is good, which is green. I will not get into the malnourishing indicators of wasting, stunting, and then anemic. But we will just look at why this could happen. So, here you could see malnourishing indicators as wasting and stunting, and stunting could be for your particular age. There is a growth chart. You're not growing well. And then there is also you are underweight. So, there is another indicator which is underweight. And these all have some costualities. We don't know what costualities, but when you make a map and you see these clusters coming out, then that is an indicator of some problem there. For example, there is some problem in this side, and then there is some problem here on the coastal regions in the southern tip of India, of Kerala. And so these indicators can be used definitely. However, as I said, dashboards can be used. This was done in our programming. Again, it is very simple to use. You cannot make very high commercials yet with open source. Why? Because you do need a big storage space. With the available storage space in your Google Drive, you can connect it to our programming, our dashboards, and then you can do it. So, see how beautifully a dataset has been plotted. Your GIS map is there, but then it is static. Whereas your dashboard is dynamic. You can see that there is a slider. If you move the slider, then you have a difference of combination of malnutrition and stunting in the state of Andhra Pradesh district, Adilabad. And that can actually create better understanding of the data. Also, another map has been done for, so these are the three indicators as stunting, wasting, and underweight. And you could see that you have another map for stunting. So the stunting map may not be the same for the wasting. You can see stunting is more prevalent along the Ganges basin, very, very strange. With such a fertile land, why there is stunting? Maybe there is a particular scheme that the government should work on to improve the stunting issues in that part of India. To complete, I will show the land use land cover classification done by another student. We have already discussed this, how different AI ML packages were used for classifying the land into sugarcane grapes and other aspects. But I didn't stop him there. I asked also him to make a dashboard so that the policymaker can quickly look at it. So when you give a land use land cover map, it's just a paper map or a report or an image on your email. But it doesn't help you to fully look into some options or comparisons. And maybe this is not done for the Sentinel hub for this area. So it is better always to have one for yourself. And that is what this student had done to look into the area of one yield predicting. So I've opened this. As I said, it is in Google Earth Engine apps. You can have an account there and then see how the data works. So what this student has done is on this side, you have the year for classification, which is 2016, and then the yield. So we want to see the classification and how much yield comes. Because if you have grapes, 10 acres and 10 acres of sugarcane, they need not give the same yield every year, because the acreage is different. How much yield comes is different. The yield depends on many factors, climate change, water, labor, correct time, labor, pesticides, fertilizers, disease attacks, all these are there. So that is why we did both. So for your classification map, we wanted to see your yield map. And then the previous year, the upcoming year from the 2016-17 was 2017-18, we have that. So you can see that how this map readily captures it. So on the left, the 2016-17 map shows clearly that it's not growing well. The yield is still less and or the cropping pattern has changed. It is not the same. Okay, so you can zoom in to Balgaon. I'll just give it a minute too. So this is running behind in real time, because all the data was downloaded in this Google Earth Engine or given links. For example, if you want this to be run on Sentinel-2A data, then you just write a code on the dashboard and then link it to it so that it pulls the data from Sentinel and then maps it. So you cannot go too much out because this data was done for the Maharashtra region. So we can just focus on the Maharashtra region sangly to look at it. So you can also change. So what is the dashboard? This is what the dashboard does. You can change the years. As I said, he went to 2019-20. So we have that here. You can just choose the one part. I'm just going to show you, so let's say 2019-20 and then 2019-20. We can search multiple parts in the same map or just one part so that you can look at the difference. So now I just put just for your sake all of them the same so that you would see that both the left and right are actually the same maps and there's no change much. So it is pulling the data from a database. It is not made because if you just make it, then these two will now crash because if you put the same date. So we can put the previous year. Let's make sure that both are the same and then the yield and the classification is done. So more on this, the paper has been shared and we can definitely look at these models more in detail and upgrade them. So these papers were done in Cambridge. A lot of professors were working on it. So I said, okay, this is very good and you should collect data. So he went to the field, collected the data on the spectral signatures and then he ran the codes for the spectral signatures which classify the land but he didn't stop there. As I said, I asked him to do a dashboard and now this can be showcased to multiple people to see if this is worthwhile in the time or to make a decision. So for example, now you can say that along the river there has been good benchmarking so that it's not fully populated along the rivers. There is some space and different crops are growing. The white land reflects the urbanization and then the crops are green and red, depending on the crop type. So let's get back to our presentation where we have most time to complete. So with this, I would like to conclude today's lecture on the fourth lecture on using of indicators and dashboards. Now you could see that initially just the bands are not enough. You should convert them to indicators and then also that we saw that indicators are not enough. It has to be in a dashboard so that readily you can visualize what are the issues in these districts. And then look very specifically for precise management practices. Just similar to precision agriculture or precision surgery. If I need to operate here, in those days they will cut the entire thing and then go in and operate. But now precision is there. You just apply medicine here. You just apply surgery if needed. You can just put medicine inside there in a capsule. So these kinds of things are coming up in big time and that is where dashboards also fall in the same place. With this, I will stop today's lecture. I will see you in the next lecture, which is the final lecture for this NPTEL course. Thank you.