Analytics

Spatial Analyst in ArcGIS for Desktop 10.4

ArcGIS Desktop 10.4 is now available for download!  For users of the Spatial Analyst extension, we have added five new tools for this release, and added some enhancements to several other tools in capabilities and performance.  Please read on to find out more.

Distance toolset
With the new Cost Connectivity tool you can generate a network of optimum paths connecting multiple areas.  This has applications in many areas of planning and operations, such as wildlife management and resource harvesting.

The cost distance tools (Cost Allocation, Cost Back Link, Cost Distance) and path distance tools (Path Distance, Path Distance Allocation, Path Distance Back Link) have been enhanced with four new parameters to give you more control over the impact of distance and the nature of the travelers.  More specifically, with these enhancements you can:

Extraction toolset

The Extract Multi Values to Points, Extract Values to Points, and Sample tools have been updated to better handle conditions when the locations being sampled are NoData cells for certain output formats.  Because the shapefile and Info table formats do not have a concept of <null>, a value of 0 was returned in those cases.  Now  a value of -9999 will be returned.

For Extract Multi Values to Points and Sample, some performance optimizations have been made to allow for faster execution against larger point datasets and with more input rasters.

Segmentation and Classification toolset

Four new tools are available for this toolset in this release.

The Create Accuracy Assessment Points tool creates randomly sampled points to be used in post-classification accuracy assessment.  The Update Accuracy Assessment Points tool updates fields in the attribute table of a ground truth data to compare ground truth points to the classified image.  The Compute Confusion Matrix tool generates a kappa index, which is an overall assessment of the accuracy of the classification.

The Train Random Trees Classifier tool employs a supervised, machine-learning classifier to generate an Esri classifier definition that is more resistant to overfitting than other classification techniques.

Final word

Please give these new capabilities a try, and be sure to let us know how they work for you.  We are always interested in getting feedback on our functionality, any problems you might encounter, and any suggestions you have for future releases.

About the author

Juan is a Product Engineer and Documentation Lead on the Spatial Analyst team. He has been working with raster analysis using Esri software since the good old days of ARC/INFO Workstation on UNIX machines.

Next Article

A classical visualization of earthquakes using ArcGIS API for JavaScript

Read this article