 This video will show you how to take LIDAR point clouds and turn them into raster surface models using ArcGIS Pro. ArcGIS Pro does allow you to work directly with point clouds in their native LAS format. Simply add them to the map or scene from the catalog, and you can begin to view them adjusting attributes. Note that when you're zoomed out the point cloud won't display, it'll simply display the bounding box. I can choose to view only a subset of points by their classification or return attributes. Here you can see me going into the layer properties, going into the LAS filter dialog, and there I'm turning off all the classifications except those points assigned to class 2, the ground class. The points are still being displayed by elevation, but only those points with the ground class are displayed. Now I'll go back and activate all points regardless of their classification. You can do the same things with return values or classification flags. By going into the symbology menu, we can symbolize the points based on elevation, class, return, or intensity. Here I'm switching from elevation, which is the default symbology, to displaying it by the class, which is the point classification. Now switching it over again to the return information so you can see the return number for each point, and then finally back again to elevation. Although you can work with point clouds directly, if your workflow involves generating surface models, it's likely you'll want to put your point clouds within an LAS dataset. An LAS dataset is simply a container for point clouds, it exists outside of a geodatabase, and you can create one within any folder. Point clouds stored within LAS datasets have additional functionality that point clouds stored natively don't have. You can populate your LAS dataset with one or more point clouds using the add files to LAS dataset to your processing tool. Simply select the input LAS dataset that you created, and then choose the LAS files, or folder containing LAS files, and then click run to add those files to your LAS dataset. With our point cloud populated within our LAS dataset, we'll now take advantage of some of the functionality within the LAS dataset layer tools dialog. Instead of using the symbology dialog, I'm going to go over under the appearance tab and use the LAS points dialog to control the points that I'm viewing. Note that the LAS points dialog is not a replacement for the symbology dialog, but it's complementary and very useful, for example, quickly showing the ground only points or the non-ground points. Computing statistics on your LAS dataset can give you greater insight into your point clouds and help you determine crucial factors when you convert from point cloud to raster formats. You can compute statistics using the LAS dataset statistics tool. By default, statistics are stored within the LAS dataset, but you also have the option to export them to an external text file. The external file is simply tabular data, and the benefit of having it in this format is that you can open it up in a spreadsheet, database, or data visualization software package for further analysis. As you can see, the LAS dataset statistics contain information on both the returns and the classifications, including both total point counts, percents, z-values, and other useful information. Typically, when generating a surface model, you only want it to contain a subset of the points. This could be by return information or by classification code. The first step in this is to use the Make LAS dataset layer tool. In this particular case, I'm selecting my input LAS dataset. I'm creating a new layer. This is not an output file, but a virtual layer. Now I'm going to only include ground points, those that have assigned the LAS classification of 2 in this case, into this LAS layer. Clicking Run will produce a new virtual layer similar to a standard layer that you would create in ArcGIS Pro that contains a subset of this LAS dataset, in this case, all returns, but only those points that are classified as ground. I'm now creating a second LAS dataset layer, this time for those points of their first returns, either first of many or standalone first returns, but they can have any class code for their classification value. When I generate my raster datasets from these point clouds, I'll use the ground points layer to generate a DEM or digital elevation model representing the bare earth topographic surface. And then I'll use the first return layer to generate a DSM or digital surface model, a raster representation of the true 3D surface of all features, including buildings and trees, which wouldn't exist in the ground point classification and hence wouldn't be present in the DEM. Prior to interpolating your point clouds into a raster dataset, it's crucial that you understand the approximate resolution of your LIDAR point cloud. To do this, I'm going to access the LAS dataset statistics that we computed in an earlier step by right clicking on my LAS dataset, going into the properties and selecting statistics. A point spacing of 0.559 in this case means that every half meter, because our coordinate system is in meters, we have a LIDAR point in this particular dataset. When I computed the LAS dataset statistics earlier, we examined the tabular output. We can view those same statistics located here within the LAS dataset properties. Information on the classification, attributes, return, and classification flags is all available. To convert my first return and ground point layers to respective raster surface models, I'm going to go up to the data menu, click on the export button, and choose raster. This will launch the geoprocessing tool to export the LAS datasets to raster format. Starting off, I'm going to use the LAS dataset to raster geoprocessing tool to take my ground points layer to produce a DEM or digital elevation model. Once again, this represents the bare earth topographic surface. I'm storing my output raster in a geodatabase, but it'd be very appropriate to store it in a folder just by adding the .tif or .img extension. You'll notice that I have options to select the interpolation type. You'll probably want to play around with these, but I've had good success using triangulation and natural neighbor with no thinning. When I get down to the sampling type, I'm using the cell size, and this is where I want to apply my knowledge of the LAS dataset statistics, specifically the point spacing. My point spacing was 0.559 for this point cloud, but I'm going to drop the cell resolution down to 0.5 because my point cloud included some water areas where there are no LiDAR returns. It's the absence of those returns that allows me to make an educated guess as to what the actual sampling value or cell size should be. I'm now going to replicate that entire workflow with the only change being that my input LAS dataset is now going to be my LiDAR first returns layer. When the signal emanating from the LiDAR sensor comes into contact with a hard service, like a building or pavement, it's going to generate a single return. But when it comes into contact with something a little bit more porous, like tree canopy, there can be multiple returns from a single pulse. As a result, by using the first returns, we'll have a highest hit or digital surface model, DSM, representing the highest point of all the features surfaces across the landscape. You may find it useful to generate other surface models, such as a surface model from the lost returns or from even the intensity values. But for this particular example, we'll stop which is the DEM digital elevation model and the DSM, the digital surface model. Once those two processes are complete, I have my DEM, which represents the bare earth, and I have my DSM, which represents the top of all features. I'm now going to use some raster functions and subtract the DEM from the DSM to create an NDSM or normalize digital surface model. By subtracting the DEM from the DSM, I'll produce a layer that represents the height above ground, and that's what the normalize digital surface model is. As raster functions create virtual layers, this new output layer is not a raster file, but rather a virtual data set that exists within my ArcGIS Pro project. Now that I have my output NDSM layer, I can go into the table of contents and give it a more meaningful name. To improve the cartographic appearance, I'm going to go back into my raster functions, generate a hillshade, put that hillshade underneath my NDSM layer, apply some transparency, and then symbology to my NDSM layer. The result is that my data are now displayed in a way that makes it much more easy for me to identify features on the landscape. Jumping ahead, you'll see that I've used the same approach from my DEM and my DSM. I've generated a hillshade using a raster function and applied symbology from the symbology menu. Turning on and off these layers allows us to explore the differences between the DEM, the DSM, and the NDSM. The DEM is the bare topographic surface, and the pixel values represent the absolute heights relative to the vertical datum. The DSM is the true 3D surface, including features such as trees and buildings and utility lines, and the pixel values are also the absolute values relative to the vertical datum. Because the NDSM is the subtraction between the DSM and the DEM, the pixel values represent the height of features above ground. So clicking on the roof of an individual building, the pixel value represents that height in map units, which in this case are meters. To summarize the steps, we first populated our point cloud into an LAS dataset. We then generated LAS dataset layers for the appropriate points that we wanted to use in the raster surface models, and then we converted those layers to raster surface models before symbolizing them.