 Welcome to the NPTEL course on remote sensing and GIS for rural development. This is week 8 lecture 5. In this week, we have been looking at land use land cover, defining what is land use and land cover. We initially start with land cover as a definition of a layer on top of the earth and then how that land cover is used is called land use. We looked at multiple data types that can be used for understanding the land use land cover and more importantly the LULC change. In today's lecture, we will look at some more data sources. Last lecture series, we also looked at data from ISRO Buwon website. We spent considerable time going into each aspect in the data and the dashboard. We made creative maps, understood the data, strengths, limitations and challenges and now we will be looking at other data. As I indicated, it is good to use more Indian space research organizations data, ISRO's data, NADAC, etc., but when it comes to application, sometimes we need the best data available. We will be showcasing methodologies using open source systems. QGIS is one where open source data is widely used. What we will do is we will look into some more data that can be used for understanding LULC globally and for India. So as I explained the last class and previous lecture series, Google Earth is very, very powerful tool to understand the change in land use land cover. We looked at some locations along the Ganges River in Birely, tributaries and we found out that the land use has changed tremendously. Land cover has also been altered, where more agriculture has been ongoing and a lot of erosion and deforestation has occurred. This is just a qualitative analysis, but can be updated once we have more data in the picture. So what we will do now is, since we have looked at Google Earth, let us see on the fly how the analysis is done. Once you open Google Earth Pro, you can zoom in to a particular location of interest and then demarcate a land use land cover type. What happens here is you will be looking at the particular land use land cover type which is of interest to you and which you want to look at. In terms of let us say, you would like to see agriculture, especially for rural development or water bodies management etc. So what happens is when you work on these type of systems, you have one time snap of LULC and you need to expand it to multiple time series to do a LULC change. We will go to the steps and do a small on the fly analysis. Change years and time, as I said we can drag and drop the time series and then see how the land use land cover type changes. But demarcate an area, demarcate an area, demarcate an LULC type. Assess change in the area. You can look at changes in the area because of human interventions, natural phenomena such as floods, cyclones and also earthquakes. But most of the time it will be through human interference, anthropogenic stresses. Then what happens is you can create statistics. Again statistics can be created by using another tableau form. So where you collect data, you put it and then you say how much the area has changed. Person change calculation, everything else can be checked. So let's do a small exercise of this in Google Earth Probe. So I am going to share the Google Earth screen. So we can look at, let's look at the plantation across. So this is along the western guards. And this is a government plantation that plants rubber, okay? So initially the land use land cover would have been a forest because it is on the western guards boundary. But then slowly development has happened. And development has happened in terms of converting that into rubber trees. So as I said, we can zoom in to a particular area. So you can see here these lines, rows, etc clearly indicate that these are not natural. Because natural trees do not grow in rows and columns. It just randomly grows and then you understand that these row column plantation is a clear plantation. So we set the tilt, okay? So you can see that this is the corporation, the factory, and these could be the trees, right? Okay, here we have all the plantations happening. And as I said, let's demarket an area, okay? At least over the last 50 years, let's see what has happened, okay? And then style and color, you can make it opaque, okay? So you now have a red line of interest area of interest. As I said, let's click on the time analysis. So now you know we can estimate the area here by running some calculations along this. You can say that, okay, my path, I'll just retrace this path to just get the approximate area. Again, since it's an area change, you can quickly do this assessment, okay? So you have the perimeter as 500 meters. And then polygon, we can see how that becomes clear and then we'll do an area, right? So the area is around, around 500 meters, perimeter area is around 14,000 square meters, hectares, let's say 1.44 hectares, okay? This is the land we are going to look at. So 1.44 hectares, you take a note and then now we'll go back in time. As I said, let's now clearly say that this is a plantation, rubber plantation and the earliest you can go is 1985, which may not be clear enough because of the resolution, but it's fine. You can go to 2006, 2006 you do not see plantations. It is a pure forest, a conservation forest, conservation area because you don't see why I could say that very prominently is the growth of the tree and the coloring is different. In a plantation, it will be the same. Why? Because they plant on the same days. So the colors of the leaves and the height will be the same because you're watering, you're taking care of the trees together. It's not like you plant one and then after one year you plant another tree, right? And then you can slowly see that this land is being cleared, okay? So initially it was forest. So let's take data note as in 2006 it was forest, 100%. So 1.44 hectares was forest. Now we say 30% is forest, 70%. So this qualitative I'm saying, but you can also measure it. How will you measure it? You do another plot. Go back to polygon and say that of the 144 hectares, this is not. So 0.67 hectares, okay? So almost 50%, 40% is not into, is the remaining forest. So all this is not a forest now. So we will have to remove it. So now the percent change is starting to happen. And then you go on to the next year, random year. You have some more forest, some more clavings, some more plantations, okay? Still no forest in 2016. So there's a forest cover increase, forest cover loss, natural phenomena is happening. And then you see boom, these plantations, you can see that every row and column is planted. So this is where a forest which could have given some livelihood options to rural development has been taken and converted to a plantation. However, this also gives occupation and livelihood. But what I'm trying to say here is you need to look at it from an angle off. Quantifying it on land use land cover. Again, here you have a lot of these land that has been covered, cleared. And then only some forest has been kept. So 10%, 20%, now slowly you'll see that every part of the land is going to be used for the plantation. So most of it is used for plantations. There is a house that comes up or a factory to process these rubber. And then some cloud cover you cannot use and then more and more. So the area gets increased. Can you see here? So in 2021, you can see that the entire area is increasing. And you see, as I said, the row cultivation is there. The height of the plant is the same throughout. And then you have this same growth, same growth. And now the leaf color, the leaf shape is the same, which means it is a uniform, homogeneous forest. And that doesn't happen. You can see here, this is not homogeneous. You can see there is dark trees, dark leaves, light leaves. And then here you have all the same color and shape. Here the shapes are different, the crown is different. So those are the clear demarcations of different land use land cover. So this is good. You take, as I said, now this is 100% plantation. So initially it was 100% forest. Now it is 100% plantation. And this is how you put document change. Now let's move on. You have clear statistics. So it is good for your class or research project. Let's move on to the presentation. So this is the on-the-fly analysis. You take an area. You quickly put a polygon. You quickly put an area estimate. And that is all is needed. Area change estimate. For one particular LULC. So here the dominant particular LULC we found is forest, which has been converted to a rubber plantation. Now we look at a USGS global crop land data, which is very, very comprehensive data set. I'll click this and we'll open the new slide. So what happens is now you have, when I click that link, this will open up. It is a Google Earth Engine Supported Dashboard. So here you can say search places, let's say India and zooms down to India, beautiful. You can pick a particular region. You can do these maps just basically as your Google Earth Engine map or Google Earth Pro map. But here they have added data, which is a global crop land data and some products. You can see all the products here on the right hand side panel. These are the very, very new ones. Very high resolution products, 30 meter resolutions. And these are different because it is not just giving you a crop land or not, like the ones we saw in Bhuvan agriculture or not, barren or not. This is going to give you a irrigated area product. Why is this very important? Irrigation means application of water. So there is two major types of farming. One is the karif farming, which is based on the monsoon water. So which means rainfall happens, the water comes to the field, plants grow. You don't apply energy, time, money to supply water. So all these are saved for the farmer. However, after the monsoon season, which lasts mostly four months, there is still need of water. There is some reminiscence water or soil moisture residue, maybe one or two months. So let's say five to six months, you have good water supply already in the soil. Not much irrigation is needed. But for cash crops and crops that grow using a lot of water, even during the rainfall, it's not enough. And thanks to climate change and other factors, this is a growing issue in rural regions. So what do we do is we need to support the plan using other resources of water, canal irrigation, groundwater irrigation, tubal irrigation, and then you also have lift irrigation, multiple types of irrigation where you take water from a water resource body, surface water, groundwater, and then you apply it to the field. So there is a source procurement, transportation, application, pumping, energy costs are a lot of costs involved. So to understand the profit, the net profit in an agricultural area, it is very important to understand the irrigated, non-irrigated crop types. So here what you could see is the land use, it's kind of a land use, land cover map, but all the major things are kept and more diversified. For example, water, ocean is kept, non-crop lands is kept, which is barren. So if you look at the Boevan's classification, you will have barren, wasteland, wetland, all those stuff, forest, et cetera. Here, none of this comes because it is focusing on only the crop area, non-crop area, and there is some human settlement development. So let's go to Tamil Nadu, it's a global map. So you can go anywhere in terms of the coverage. And you can see that it is kind of a land use, land cover map for Chennai. Chennai has a lot of organizations, so you can see that urban systems have grown. But more importantly, it is also a highly intensive agricultural state. You can see that a lot of agriculture is happening, less compared to the other regions. So here, Kerala, you'll see less. Almost entire Tamil Nadu is covered with agricultural, and then same as that in, along the Vaisangas, not much. Same as that in Maharashtra and Karnataka, Andhra, you have some split. And in Odisha, that is where less, less areas, then going up north, you have less areas in Rajasthan, which is blank. The Hilly regions also are very less. Same in the northeastern regions. Okay, so coming back, in the Tamil Nadu region, you could see that there is a lot of, of the right hand side, southern right hand side, you see a lot of irrigation happening. And these irrigation are happening along the coastal regions, along the regions where the major cities are present like Chennai, Puducherry, et cetera. So how is this sustainable is the question, because all these water bodies are facing high water stress, groundwater resources are facing stress because of these irrigations. Then you have the rain fed crop lands. Sometimes you have rain fed overlapping the irrigated area, which means you have an area where rainfall crops are used, but then after the rainfall, it still is being used for irrigation. Let's say rice. Rice could be used as a curry crop where a lot of rainfall is taken up for growing. Then after the harvest is in the same land, it can be used for irrigated crops, like legumes, like ground soybeans, ground nuts, vegetables like carrot, onions, potatoes, all these things can be grown in this area. So you have the major two types, irrigated crop lands and rain crop lands. So for India, you could see that the central regions are mostly rain fed, and that is the concern also because if there's a big climate change impact like droughts and floods, there is no much water available for agriculture. And so there's a need for building climate change resilient crops. Whereas here you see some in the western side, some regions are rain fed, they like Coimbatore and stuff, but they get more rainfall also because of some rainfall coming from the western guards and stuff. So this is your rainfall irrigated areas in yellow, and then your irrigated crop lands in green, and then this slider will give you an opacity. So same thing if you want to reduce the opacity to see the background and show how this crop is working, you can play with the slider. So most important is in this, you have a ready-made Google Earth engine plugged in with NDVI. So let's say I click this enable NDVI and I'm going to click this map here. It's going to generate a graph of the current scenario May 22, September 2022, and this is December. Until December, the NDVI value is available. The left hand side, the units may be scaled. For now you could say that is it increasing or decreasing along the baseline? So the baseline is this. You can ask if the data is increasing or decreasing. I'm just going to make it big. And you could see that the NDVI value goes down from February 22 to December 22, a little bit down and then goes up almost above the level in February. So Jan Feb, there is a good winter rainfall, maybe in that region. And so there is crops picking up. So, but you can also go to this area and say, okay, these are all irrigated areas. Let's click on a map. And then you have very high NDVI compared to the previous ones. So from here to 0 to 7, whereas here to 0 to 8, and most of the values are in the peak, you have an August rainfall capturing high NDVI value. So NDVI, when it's high, it is higher vegetation is happening. When it is low or minus one, it is less vegetation and minus one equals to water. So you could see that here, there is not much differences because there is two rainfall seasons, one in December, one in August. And so there is a good application of irrigation water. We'll also click on the water body just for sake. And as I said, it goes below minus because negative is for water bodies. And you could see that the chart generates by itself. Okay. So this is how the system works here. You could, I'm just going to... So they use modest data to calculate the NDVI. The NDVI is a indicator of vegetation cover. If the value is positive and high, it ranges normally from minus one to plus one. The range in this particular dashboard is off a little bit or maybe they scaled it, which is fine. But normal range is minus one to plus one, whereas minus one relates to water bodies and barren land is from minus one to zero. So the negative values are mostly for barren land and other aspects, whereas the positives are green cover. So as the crop is healthy and growing, it attains full growth and goes to plus one value of NDVI. So that both of them we can see. So this is one product lands that derived a rainfall and irrigated product at 30 meter resolution. So all these are, each pixel is 30 meters. Let's take it out and then we'll go to the next one, which is global crop lands extent product at 30 meter resolution. So basically this is getting populated. This is total crop lands. Okay. In the initial one, it is crop land plus water body plus barren. So you could see that some overlap is happening. What I am going to do is, I'm going to reduce this cover opacity, so then we can also see this guy. And you can play up and down with this to see that. All these same spots are done. So basically crop land is together. It is merging the irrigated crop land and the rainfall crop land, just to show an image of where is the majority crop land happening. Okay. So the layer on the top always has higher precedence. So for example, if I keep it high, you can see the yellow marks and the brown marks coming, which is not part of the crop land down. So crop land is only crops, but for some reason it is also picking up the Western Guards, which is not crop land. So these are the things that they did it uniformly, but they should understand that Western Guards is not a crop land. It is forest. It is a conservation forest, they should not. So there are data, which is good because this is done for global. So you can see that globally it has been done. And you see it beautifully populated for the entire Indian region. But then when you zoom in, you've got to be careful about using it widely. Okay. So I'm going to remove these two products. Now you have these 250 meters product, which is bigger in resolution. Let's do Australia and then you have multiple years. Okay, let's do 2015. And what they're going to do is, they're going to show Australia's land, use land cover, crop land, rainfall, season one, season two, and then the crop lands fallow. All these are done. So you could see that there is tremendous fallow land in Australia. The central region is very, very dry. And that is one of the reasons they are high importers of food crops. Food produce. Okay. So let's take this off also. Then we have the 1000 meter products. All those are the products we have, the global GCEA multistating crop land mask. And then it is basically a 1000 meter product of land use land cover, minor fragments, very minor fragments, irrigated, minor, irrigated major, different schemes of irrigation. And then you have the global GCEA dominance, which is the wheat mixed crops and other things. And you could see that we have mostly in the southern region, a lot of rice. So wheat and rice dominant regions are dark blue, light blue are rain fed, wheat, rice, soya beans. So these are the two colors that come in the south. You can now look at where it becomes yellow. So yellow is wheat and barley dominant. So there's not much rice. So India is beautifully divided in terms of food and major staple. More rice is had in the south, south, whereas wheat is had more in the northern regions. So chapatis, rotis are consumed more in the northern part, whereas rice, idlies, chawal is, our southern soil is taken up in the southern regions. Okay. And one more layer which is very important is the human settlement layer. I've clicked it, but it doesn't populate anything yet because maybe the data has not been plugged in, which can wait for some more time. You see it picks up. Yeah, in Chennai it picks up slightly. You see that the brown color is picking up. These are the human settlements for 2015. All these are 2015, which is similar to the Bhuvan's data product of 2015. We will do a hands-on course on doing a very, very current land use land cover classification in week nine first lecture. Okay. So before we finish, I also wanted to showcase the other dashboard data sets that are available. So these kind of dashboards come up often. So how do you know what is the protocol? What is the data source, et cetera? You can click download data. It will open this page, which I've already opened for you. But let it go through. Okay. So once you hit the data download page, it will give you information about the download data. The 30 meters says it is a 30 meter resolution irrigated versus grain crops, et cetera, et cetera. So which one you want to download? You can just download at a global scale for the year 2015. But this is highly, highly validated. I'll show you how it has been validated. And then it is only for visualization, et cetera, et cetera. So this part of .org is for visualization. So you have all these data that can be downloaded and mapped into GIS for further analysis and experimentation. Then you also have the information you can click and know about what are the data sets, contacts, and documents, different maps that have been already made. So let's say web map products. So this is a Google Earth Engine that we already used. And then the left-hand side will populate. So this is the visual dashboard. And then you can also look at area maps, percentage to total pro plan. So these are maps that have been made for different regions. And then some numbers are there for 2015 year. So it gives you cropland by continent, how much acreage, 152 million hectares, 8.1% of the global. So you can see that Asia, Pan Asia has the highest contribution to agriculture. Cropland is 33%. Almost one-third of the planet derives its food from Asia, Asian regions. Whereas the next highest would be the European continent, 25.5% and Russia. And then you have China also contributing to the Asian parts. These are maps that have been made already. How would I get there? We just went to data products and then maps and the cropland. And the final maps are there. Interactive cropland maps are also there. Where you put interact in terms of what areas you want. You want to have a percentage. So I can click the percentage. It says what is the percentage, 9.6% of the total land area percentage. But again, the boundary should be very carefully used. So since this is a USGS website, they use different boundaries maybe. But one thing which could be of use is if you go to the products and then say area maps and then the cropland percentage, you can see that the boundaries have been updated. So there are some updates needed about these maps. Then the other interesting part is the information. You can see other documents contact me, download data, etc. In the download data, we saw the links to be knowing about the data, the metadata. For example, read me and it talks about the data, use a guide about the data. So very, very recent, see how it was released. It was released only in January 2023. So just two months ago, this data has been come up and it has been widely used now. So this is the part where you went into data download and then address the things. In the data, you can also look at the reference data which is very, very important. What is reference data? So these are the ground points which I always ask you to collect from the ground. You collect points and then you supply it to the coloring scheme so that the color green reflects a turf or a tree or a plant can be demarcated. So green is green for our eyes, but on the computer panel, it may look different. So here you can see that there's a lot of these data sets have been taken from different, different sources, source type street view they have taken and made these maps. So for example, cropland, rapeseed or canola, an area, a green color has been marked and it has been marked during the rainforest season of 2030 using street view and for Thailand area. So that was used for Thailand's color coding, et cetera. Let's Google and find where India was used. You could see that India very less data was used. Ground data is where you go to the ground and collect data. So cropland, rice, rice data for India, the color of the crop was taken by ground estimates in 2017. You could ask me why 2017 was used for 2013. It took time for them to make these data sets and if the crop type doesn't change between 2015 and 2017, you could still use the crop identity. So here they use rice and then they use forest unknown. They don't have any source. Maybe it is a map from a government or a literature review. And then they have ground, ground data for cropland and built up. And then street views use ground data for cropland is also used. So India now and then there is good forest color for ground data. And then so six data sets have been used for the Indian database. Okay, so I think this is good for understanding how it's been done. You could see that these are the data points that were used for demarketing where the area is. You can show that where the cropland data filters and then this will filter out saying only cropland and forest. And you say apply, it will apply to the points. It'll also give you sometimes the points where the data was taken. Let's say India again, I'm clicking on the link. It doesn't open, but normally it does open the link to the data set. So maybe they'll update it, but I'll show you how these data sets does work. So you can definitely use these. Let's click just a normal location of United States. It still does work. And this data set may be used in the future. So don't ask me why it doesn't work now. When it started, maybe it worked. But then as I said, they do work through issues of data and other issues. Pick it up. If you're on ground truth, data you can apply, reset, and then look at where the data was collected. So for example, this data you can click and see if it gives some validation, calibration for your models. Because they have already went and taken the data. And then these are the Indian points. So with this, a good exercise will be following up soon on the the formation of land use land cover using satellite data. You can also do an accuracy map to see where which regions are more accurate. So you can see that producer, accuracy map, user accuracy map, overall accuracy map. You can see that India is okay. This region is having an accuracy of 85% or 83% overall accuracy zones by zones they created because by zone they collected data and mapped the data. Okay. So this is now 88% along the western regions. Whereas this is around 92.4% overall accuracy 78% 83%. So as you move your mouse, you could see that how the accuracy of percentages change. With this, I'll stop here. I'll go back to my initial slide and then conclude the presentation with the need for mapping and data using multiple sources. Whatever is the best, please use it. And so I'll conclude here and meet you in the week night. Thank you.