Good morning, all, and my name is Nelson Mileo. I'm from the University of Lisbon. And my presentation is about a plug-in

that we developed for the symmetric mapping. And first words about what is this work. When in Sunday, when I was coming to Kim range, my daughter asked me, what are you going to do next week?

I'm going to talk with some friends in the conference. But about what? And I thought, oh, here comes the questions. How can I explain my daughter what I'm going to do? And then I pick up the playmobiles, and then I explain what was the work. The work is just simple as this.

If you have a statistical area, an enumeration area, or a block area, you can count the people that are in that statistical area. You know that there are 10, 11, 12 people in that statistical area. But normally, you don't know inside that area

how that people is distributed. And then I explain to my daughter, the work that we've done is simply that we have the playmobiles, and we redistribute the playmobiles where they live.

And to make that, we use a simple map where we can see where is the forest, where is the water, because some people don't live in water, in lakes, in rivers. And that was my simple explanation to my daughter about the plugin.

And then she told me, oh, nice, the playmobiles are nice. About my presentation, I will show you how we did it. It's a solution that we use to have a raster map about population distribution,

because we need to have that map to make a plan. We need to have a raster map, because these statistical data, the population data, are stored in vector maps, in statistical enumeration areas. And we want it to be in a raster format.

And that's a problem that we, it's not new, but it's an old problem that it solved in other softwares, but it was not solved in KGIS. And that was the starting point for this work. OK, let's start with the objective that is,

as I told you, to make this raster representation. And another point is that you can check, you can download this plugin to use with your own data. OK, now I'm going to show you our solution.

It's not the only solution. There are many solutions. But we pick the solution presented by Holloway. That is just simple as this. We have to calculate the relative density of the mapping unit population for each land cover

type, forest, water, agriculture, urban, and so on. And then we have to pick to select the resolution that we want to have our final map.

And then we have to convert the population from the vector origin to raster. And then we have to calculate the parameter that is the expected population of the enumeration unit calculated

using relative densities. And that is an important step in this work because we used new data that was only available last year is the data about where the people live.

And in this slide, you can see how these equations, these formula was implemented. And this diagram is a simplification of the plugin.

For showing you that we started here with the statistical enumeration areas is the first point. We reused many functions from GRASS GIS.

The first was to convert it to raster. We need to develop there where the tabulation, we don't have any two for tabulate, cross tabulate, two layers.

And then we need to develop that tool. It was the hardest part to implement in this algorithm to tabulate with the land use. We need to tabulate the statistical areas with the land use in order to obtain the weight. And this was obtained by tabulating the two layers.

And then what we are doing with this diagram is obtaining all these parts of the formula. It's just simple as this. In order to, in the final step,

to calculate the decimatic map. The decimatic map is a representation of the population by the land use, by the different land uses, simple as this. Another problem that we need to solve is the densities or the wastes by land use classes.

And this was solved using the addresses open data layer. It's a point layer that the Statistical National Institute

is delivering to all the people in open data. And we use it to calculate these densities. In other works, this calculation of the densities is based on the image maps and satellite images.

Instead, we choose this layer because it's a new layer with good data for doing this calculation. This is the look of our plug-in.

I will just show you in the software. It's easier to explain you. And I will pass this slide. And in order to test if the plug-in was working, we issues GIMRANCHE municipality

because we are already doing work here, an emergency plan. And we choose GIMRANCHE for the tests. About the data to perform the tests. First one, we use as land use, the Korean land cover

that is available for all Europe. But you can use any other land use layer. The only point that if you intend to use the plug-in is you need to make a new recode file because we

are recoding the Korean land cover codes to new classes because they have more than 30 classes. And I'm going back to explain that we are recoding lots of classes to just these ones.

And if you want to use other land use layers, you need to make new recode files that are available in the plug-in folder. OK. Let's move forward.

The other layer is the typical statistical layer that we have in Portugal for all that represents the outing census geographic reference layer. And the other layer that we use is the addresses layer that was used to calculate

the relative waste of land use. If you want to use it, you can obtain it in Inspire atom download service provided by National Institute of Statistics. I don't know if Francisco Caldera is here, but thank you for making this layer available.

I think he's in the other auditorium. OK, and this is the result. And it was what we are expecting. Because if you can see, these gray areas are where the people don't live, rivers, some forested areas.

And it was really what we are expecting from our results. And if we have some results, we need to have some validation. And the first validation is the people that we have initially. It must be the people that we have in the final roster.

And that validation was clear for us. And then the other validation was how these restorations affect the population in each cell. And we found this histogram where we can see our differences between the results

of each cell and the population that is in the enumeration area. And it has a normal distribution. And that was what we are expecting also this type of distribution of error.

OK, concluding. For us, the tool establishes a new approach to obtain the relative weights for each length used. Because I didn't found any reference

that used these addresses open data for obtaining these weights. And this is something that was good for us using and see that works nice with our plug-in.

And then another issue is that this tool can be used in any place in Europe because Korean land cover, the land use, is available for all Europe. And you can check it in your own country.

OK, some issues that I would like to stress out is that we had some problems with the process because it's somehow time consuming. And I think that's a problem that we are trying to solve because it's using only some parts

of the memory of the computer. It's a thread problem that we are trying to solve. But the point is that it automates our process. Instead of making step by step all these steps that I show you in the diagram, it automates.

And it's simply for the user, the final user. You don't have to have specific knowledge about the symmetric mapping. You can pick, you can use our tool and obtain the final raster for the population.

If I have more, two minutes, I would like to show you how you can use it. Because I pass a slide very fast because I would like to show you it's lost, voila.

And there, you can see. Here, I'm going to close this. You just need to have these layers. This is the land use layer that I talked to you about.

It's the Korean land cover. And then you need to have these statistical areas. And what we did was this plug-in.

And this plug-in just picks, as input, two layers. It is the statistical layer. You need to have an ID field in order to tabulate, as I told you. You need to have a field that represents,

in our case, the population. And the second input layer is the land use, in our case, the Korean land cover, and the code that represents each land use class. And then the point is here, the classes and the ways

that you give to each class that were obtained from the open addresses data, OK? It's something that I didn't see any reference. And then you select the resolution. You can obtain a raster for 10, 25, 100,

whatever you want for your work. And then you save here your final raster that now I'm saving as a TIFF, but it's possible, sorry, for obtaining any other format, OK?

I finished my presentation. Thank you for your attention. For now, because we are testing,

I finished it last month. It's just in the GitHub. You can download it. But obviously, it can be uploaded in the repository of QGIS.

I think I understood that you used some GRASS processing, GRASS, and then some other implemented new algorithms. These new algorithms have been implemented within GRASS

so that the whole procedure is available also within GRASS or just in QGIS on the EuroScript. Yeah. The point is we try to reuse all the available algorithms. And I know that GRASS has a cross-tabulation tool, but it doesn't work for our proposed.

And then we need to develop it in Python. It was developed in Python. It's in the script. It's inside the script. And that's another, I think, that's another improvement that we must do, because it's the time consuming.

I think I didn't test it yet. It's that part. Thank you for the nice presentation. And also, thank you, everybody, for being here for the first part of this academic track session.

Second part that we start at 10 past 11. And so we have a break now. Thank you, everybody.