 Hello, I'm Hans van der Kwast, lecturer at IHE Delft Institute for Water Education. In this lecture I'm going to explain about projections in GIS. After this lecture you'll be able to explain why we use projections, to describe the advantages and disadvantages of projections, to choose the right projection for your purpose, to use EPSG codes, and to explain the difference between on-the-fly projection and the projection of layers. Why do we need projections? Well, the problem is that the Earth is in 3D, it's a sphere, and we often want to present maps on a flat piece of paper or on a screen. Therefore we need a coordinate reference system, or spatial reference system, which defines a translation from 3D to 2D coordinates. So we want to go from the 3D coordinates of the globe to a 2D coordinate system on the screen or on paper. To represent coordinates on the globe we use latitude and longitude coordinates in degrees, where zero latitude is defined as the equator. When we move towards the north pole the values increase positively towards 90 degrees north on the north pole, and negatively towards the south with minus 90 degrees on the south pole. For the longitude, zero degrees is defined as the prime meridian over greenwich. The values increase positively towards the east with 180 degrees around the international dateline, and negatively towards the west. We call this a geographic coordinate reference system, and the datum that's often used in this system is the WGS84. You will learn more about datums later. If we want to present the location of IHG Delft Institute for Water Education in degrees, that will be 52 degrees north, 0 minutes and 32 seconds. And it will be 4 degrees, 20 minutes and 16 seconds east. The notation of degrees, minutes and seconds is often very difficult for computer software to read. Therefore we convert the degrees, minutes, seconds, DMS coordinates to decimal degrees, DD. If we take the latitude of the IHG Delft building, 52 degrees, 0 minutes and 32 seconds, we can convert it quite easily to decimal degrees. We take 52, and we add the amount of minutes, which is 0 over 60 minutes in a full degree, plus 32 seconds over 3600 seconds in a full degree, and this gives then 52.0089 degrees north. Can you try the same for the longitude? We take 4 degrees, we add 21 minutes over 60 minutes, and we add 16 seconds over 3600 seconds, and that makes 4.3544 degrees east. So you see it's very easy to convert between DMS and DD. So the reason of map projections is that we have to go from a 3D globe to a 2D map on paper or on a screen. Have you ever tried to peel an orange and to keep the skin without breaking? In fact that's the same problem if we go from a 3D world to a 2D map. We distort properties such as area, scale, shape and direction. It is not possible to preserve all these properties. To solve this problem we have basically three families of map projections. First there's the cylindrical projection that is designed to preserve distances or areas. If you imagine the globe and we fold a sheet of paper around it in the shape of a cylinder and we project everything that's on the globe on the cylinder that surrounds the globe and then we cut the cylinder open, then we get the cylindrical projection. The second type are the conical projections. In this case we fold a sheet of paper in the form of a cone or a hat around the globe and we project everything that's on the globe on the sheet of paper and cut it open. Then we have the conical projection which preserves the angles. The third type are the planar projections. They preserve distances. So when we have the globe we can put a sheet of paper on one of the sides and project everything from the globe to the sheet of paper. That's in fact a disk and that is the planar projection. All projections have advantages and disadvantages and there are distortions of angular conformity, distance and area. So we can minimize the problems of one distortion by choosing a certain projection. There are also projections that compromise all distortions and all means that there is inaccuracy but it minimizes all the distortions of area, angular conformity and distance. An example is this Robinson projection or the Winkle-Triple projection. If we want to preserve angles we use the conformal or orthomorphic projections. A famous one is the Mercator projection which we see here on the picture. It's very often used as school maps in the US and used for navigation and meteorology. The problem is that it results in a distortion of the proportion of the area and the distortion is much larger if we are further north or south from the equator. For example if we look at Greenland it looks as large as the whole continent of Africa. The true size, there is a nice website thetruesize.com where we can move a country over other countries to compare the true size. Like in this example we see how the United States of America is comparable to part of North Africa and the Sahel. A very frequently used projection in this group of projections is the Universal Transverse Mercator or UTM projection. It is a global projection so you can use it for any location on the globe which is divided in 60 equal zones each 6 degrees wide in longitude from east to west and it starts numbering with number one at the international date line and goes up to 60. When we use it we need to indicate if it's north or south of the equator. The origin is on the equator at a specific longitude that changes using a false north and false east. So north and south are used in the indication of the projection to distinguish between northern and southern hemisphere. For example if we look at Uganda it's in UTM zone 36 north and then we can express the location of Kampala as 452,611 meters east and 36,127 meters north. It's very easy to find a map on the internet with all the zones indicated. So we can see for the Netherlands that it is in UTM zone 31 north. Depending on your location you need to use a different datum. So in US you use North American datum. In Europe we use WGS84 and in Africa around equator in east Africa we use the ARC 1960 datum. Later in this lecture you will learn what a datum is. There are also projections with equal distance, the equidistant projection. They keep the scale constant, the distances are accurate and examples are the plate curry equidistant cylindrical projection and some others. They are used for radio and seismic mapping and navigation. An example of this is the United Nations logo which is the azimuthal equidistant projection. So the distances are preserved but of course the distortions are large from the center. There are also projections with equal areas where the areas are more proportional. So here we see the MOLVAD Equal Area Cylindrical Projection where Greenland is now more proportional to the continent of Africa. But it gives a lot of distortions in shapes in the angular conformity. These maps are often used in maps for education or as a general reference. Let's look at some terminology that I've already mentioned in the previous slides. There's datum, there's spheroid, there's geoid, there's false northing and false... If we use projections we need a reference level. And the problem is that the globe is not a perfect sphere. So we can use the spheroid but that will not be a good fit. We can also choose to use the ocean level but that is also not very constant and very difficult to use of course at land. Then there's a reference ellipsoid which is a fit. We can use a local plumb line, we can use the continent or we can use the geoid. The geoid is the Earth gravity field which is also not constant in space but is a very accurate reference if we want to measure a reference elevation at a very precise level. For most purposes we use the datum which is a localized approximation of the ellipsoid. So examples are WGS84 or ARK1960 or North American datum. I've already mentioned a few times the false northing and false easting. What this means is best illustrated using the Dutch projection which is called the Rijksriukstelsel or Amersfoort. Originally the origin of this projection was at the church tower in Amersfoort a city in the center of the Netherlands. However there was a problem for people using it offshore like the Navy because they had to deal with negative values. So instead of redefining completely the origin they moved this origin artificially by introducing a false northing and a false easting. Since 1970s the origin has been moved with a false northing of 155 kilometers and a false easting of 463 kilometers. Since then all the territories of the Netherlands are in positive values and this is referred to as the Rijksriukstelsel or RD new projection. Now with all these projections which one will you choose? That depends very much on different factors. Are you studying a specific country or is the region consisting of multiple countries? If it's a specific country and you do a project for a government you normally use the national projection. If it's more regional it is often more useful to use a global projection such as UTM. It also depends on the type of analysis. Different projections have advantages and disadvantages and definitely if you do DEM analysis, work with elevation models or do measurements it is not a good idea to use the latitude-longitude but to use a projection. It also depends on the availability of data. Is a lot of national data available then you also stick to the national projection. If you rely on global data a lot then it's maybe better to stay in a global projection. The bottom line is that you and your team have to decide on the projection before you start using GIS. You need a common reference system for each project for your organization and even in a country you need to agree on that. And you can always fall back on the global coordinate systems such as UTM or the geographic coordinate reference system latitude-longitude if you don't need to do calculations. So we've learned a lot about projections and coordinates but now how do we use this more practical in GIS? Fortunately some decades ago the European Petroleum Survey Group has standardized the projections in so-called EPSG codes and they are cataloged and most open-source GIS tools support this catalog. Tools such as QGIS and GDAL definitely use this. If we want to express the projections of the Netherlands Amersfoortrex, Riechstelsel, NU in one number in the EPSG code we use the code 28992 and if we share this with everybody in the Netherlands they will all use the right projection in their tools. The same we can do for the UTM zone of the Netherlands on the 31 north on the date on WGS84 and there it is 32631. If we would stay with latitude-longitude on WGS84 we use the EPSG code for 326. But how do you find these EPSG codes? Well, if you know the name of the projection you can use the keywords in spatialreference.org to look in the catalog for the right EPSG code that belongs to that projection. If you don't know which projections are used in your country well the first step is of course to contact your national mapping agency but if that is very difficult you can go to the website epsg.io and look for a country or the region. It uses the internet to see how often in a certain country a projection is used and then you can base your choice on that and find the right EPSG code. If you have a shapefile which is already projected then it often comes with the .prj file. If you upload that file to prj2epsg.org it will return the EPSG code that you can use. It's very important that all your layers in a project are in the same projection if you want to do calculations with it. That's a bit confusing because if you want to visualize layers with different projections your GIS software often will use the so-called underfly reprojection to quickly on the fly reproject them to the same projection. The first map you enter into the software determines often the projection of your project. Then the second map if it has a different projection it will not ask you but it will automatically on the fly reproject it to the projection of the first layer. But it will not change the file so you might think that both are in the same projection but in fact what it did is only visually reproject the second layer to the same projection as the first layer. So if you want to do calculations you need to really use a reprojection function in the GIS to reproject it. So you have to really be aware that underfly reprojection does not change the projection of your files. It only does it on the map canvas. And you can avoid a lot of mistakes by realizing that and doing a real reprojection. There are three cases that I want to finish with. When you get data there's a case where the projection is known and that the projection is also assigned to the file. That means you don't need to take action and your GIS will automatically recognize the projection that is used. Sometimes you get a file and the person told you the projection or you know the projection but it has not been assigned to the file. Therefore we can assign the projection to the layer in the layer properties. This should not be confused with reprojection. This is just tagging the projection name and assigning it to the layer. If the projection is unknown we need to geo-reference the layer registering it to a coordinate system and rectifying it. This is one of the exercises in my GIS course. There's also a video that explains this.