The extreme sport of Parkour, or “Free Running,” is a discipline where athletes try to find the most efficient way from Point A to Point B. If that entails scaling a wall or jumping an alleyway, then so be it. It’s about seeing the obstacles not as something to reroute your entire path, but to be overcome through agility and outside-of-the-box problem solving.
Similarly, the ERDAS IMAGINE® Spatial Modeler is about solving geoprocessing challenges in the most efficient way—by removing the traditional dialog-driven workflow and instead streamlining a customized, repeatable means of getting the answer. By creating a model, you let the domain expert do the heavy scientific reasoning and then fully leverage that knowledge by enabling non-domain experts to run the models by only specifying inputs and a few user-friendly variables.
In this first blog we’ll talk about on-the-fly raster and vector analytics and LAStools for processing point cloud data. In a second blog we will discuss some of the exciting ways to extend the power of the Spatial Modeler beyond what comes with ERDAS IMAGINE.
The Beginnings of Geoprocessing: ERDAS IMAGINE Model Maker
ERDAS IMAGINE introduced the first graphic spatial modeler way back in the early 90’s. In fact, many of the dialogs for ERDAS IMAGINE actually used spatial models behind the scenes to run their analyses, such as this one for topographic normalization. But there were still many obstacles in the way, and running a model could be a time-consuming and not very user-friendly experience. In addition, the original version was raster-centric; all vector analysis had to be done on rasterized versions of the data.
Keep Moving: On-the-Fly Analytics
In 2010 we sat down and tried to figure out the most efficient way to allow you to construct models in an intuitive, friendly, powerful environment. We still wanted to leverage the huge variety of raster data access that is native in ERDAS IMAGINE, but in a more real-time environment. Inspired by the incredibly versatile ER Mapper technology we acquired in 2007, we wanted to integrate a way to preview results in real-time, making it easier to examine what-if scenarios so you will architect the most accurate, flexible and comprehensive model. The “pull architecture” enables this—the preview window “pulls” data through the model in real-time (in many cases).
This next-generation spatial modeler features an all-new, modern interface and provides real-time preview of results. Offering greater flexibility and efficiency, this new approach significantly enhanced core spatial modeling capabilities. We continue to enhance the Spatial Modeler by adding advanced operators like classification and de-hazing. Data analysts can now reach a new level of automation with robust operators for image segmentation, unsupervised classification, spatially-adaptive local brightness and contrast correction.
Moving forward, we are working to include the power and flexibility of ER Mapper in IMAGINE by including all of the ER Mapper workflows and wizards into ERDAS IMAGINE primarily via Spatial Modeler.
Jump the Gap: Vector Analytics
For years, one of the main gaps in the functionality of the Spatial Modeler was the inability to process vector data as vectors. Any vector inputs were immediately rasterized for processing. When ERDAS and Intergraph merged, we were able to pull from GeoMedia®’s rich set of real vector analytics. One of the strengths of GeoMedia is its real-time vector analysis. If you build a complex query and then change a parameter, the entire query system updates and you see the new output on-the-fly. This makes GeoMedia a perfect platform for Spatial Modeling. In fact, we are working on extracting the Spatial Model Editor from ERDAS IMAGINE and making it available in GeoMedia with significantly more vector analytics to follow.
Adapt Quickly: LAStools and Point Clouds
Raster and vector are the old guard in the Geospatial industry; LiDAR and photogrammetrically derived point clouds are the powerful upstarts, growing in popularity and usefulness. One of the most valuable aspects of our new modeler is that it lets you create models that bring disparate data sources together to help you get the full measure of information out of your data investment. However, there is such an explosion of technical advancements happening with point cloud analytics that we can’t possibly keep pace on our own. To win that race, we are working with the experts from rapidlasso to implement LAStools in our Spatial Modeler. This will significantly extend what our users can do with point clouds, particularly when the analysis extends beyond point clouds and into the raster and vector domain.
In the second post, we will examine how the Spatial Modeler is designed to help you set your ingenuity loose and tackle the really difficult challenges.