What's New for Spatial Analyst in ArcGIS Pro 2.9

11-18-2021 03:51 PM
Esri Contributor
2 0 180

For ArcGIS Pro 2.9, we made a number of improvements to the Spatial Analyst extension. Key amongst them is an exciting new capability for the Suitability Modeler. We added new functionality for calculating statistics for a moving window across multidimensional raster data. Read on for more details on these and other new and improved capabilities available with this release.


Where do I get it?

ArcGIS Pro 2.9 was released on November  11, 2021.

For a complete summary of all the changes in this release, have a look at the What’s New in ArcGIS Pro 2.9 blog post and video.


What’s changed for Spatial Analyst?

Listed here are the main functional areas that saw improvements over the last release:

  1. Suitability Modeling
  2. Interpolation analysis
  3. Local raster analysis
  4. Multidimensional analysis
  5. Raster reclassification
  6. Segmentation and classification
  7. Surface  analysis


1. Suitability Modeler

The Suitability Modeler in ArcGIS Pro 2.9 sees a significant advance in technical capability for working with servers. We also made a number of general enhancements and improvements in quality.

Suitability modeler on servers

With this release, the Suitability Modeler can now be also run on servers using ArcGIS Pro as a client. The suitability and locate maps you create can be shared on the server. To get access to these capabilities, you need to be signed in to an ArcGIS Enterprise portal that has ArcGIS Image Server configured for raster analytics.

A new Output type parameter on the Suitability and Locate tabs allow you to select between running the suitability modeling workflow locally or on servers.

Enhancements and improvements

Use the new Sources tab to: see the sources for each criteria, identify which criteria in the model are transformed, change the source for a criterion, and remove a criterion from the model. When changing the source for a criterion, a series of warning and error messages ensure the new source and the transformation being applied are appropriate. You can use this tab to convert local criteria into web imagery layers, which are the required input type to run the analysis on servers.

For the Range of Classes tab in the Transformation pane, you can now enter suitability values directly into the transformation table. Use the Classify button to change the number of classes. You can select from seven different classification methods.

We refined the interaction for multiple map views. You can remove criteria from the model simply by clicking a remove button next to the criterion. All messages from the underlying geoprocessing tools are now displayed.


2. Interpolation analysis

If you have used the Natural Neighbor tool to interpolate raster surfaces, perhaps you ran into difficulties with increasing numbers of input points? You will be happy to know we enhanced the tool to support very large inputs, up to approximately two billion points.


3. Local analysis

A new Percentile statistic option is available for the Cell Statistics GP tool and  Cell Statistics raster function. When that option is selected, a new Percentile Value parameter is enabled with which to specify the particular percentile (between 0 and 100) to calculate.

When the statistics operation is Median or Percentile, a new Percentile Interpolation Type parameter is enabled. Use this parameter to select the method of interpolation when the specified percentile value lies between two input cell values.


4. Multidimension analysis

You can now calculates statistics over a moving window on multidimensional data along a specified dimension with the new Dimensional Moving Statistics tool and Dimensional Moving Statistics raster function. Moving statistics is also known as moving window statistics, rolling statistics, or running statistics. Essentially, a predefined window around each dimension value is used to calculate various statistics before moving to the next. This capability helps in workflows such as smoothing out noise or anomalies across dimensions.

An exciting new statistic type is available with this tool. The Circular Mean calculates the mean for angles or other cyclic quantities, such as compass direction in, or months in a year. This setting enables a Circular Wrap Value parameter for  designating a value to wrap around to calculate the circular mean. In angle calculations, for example, the parameter should be set to 360 (degrees). This means the value 360 will be wrapped to 0, the value 370 will be wrapped to 10, and so on. Another application is for time calculations based on months in a year, where the circular wrap value should be 12. In this case, an input value of 13 will be wrapped to 1.

A new method is available for how to handle NoData values in the statistics calculation. In addition to the typical Data and NoData choices, the Fill NoData option will replace NoData values in the input with the result of applying the selected statistic on the values within the defined window.

The Aggregate Multidimensional raster function has a new Percentile statistics setting. This option enables a new Percentile Value parameter. A new Percentile Interpolation Type parameter becomes available when the statistics operation is Median or Percentile.


5. Raster reclassification

The Slice tool has four new methods available for reclassifying rasters. For the new Defined interval method and the two new Standard deviation methods, the new Interval size parameter determines the number of zones in the output raster. For the new Geometric interval method, the setting of the Number of output zones parameter determines the number of zones.

The new Change NoData to value for output parameter makes replacing NoData cells in the input raster to a value of your choice easier.

One small detail to note is that the parameter labeled as Base zone for output in ArcGIS Pro 2.8 and earlier is now labeled Starting value for output. Since the name for this parameter in Python remains the same, your scripts will continue to run as-is.


6. Segmentation and classification

The new Train K-Nearest Neighbor Classifier tool generates an Esri classifier definition file (.ecd) using the K-Nearest Neighbor (KNN) classification method.

Support for multidimensional raster input was added to the Create Accuracy Assessment Points and Update Accuracy Assessment Points tools, as well as commensurate parameters to select the dimension field.

For the Export Training Data For Deep Learning tool, the Metadata Format parameter has a new Imagenet keyword option for object detection labeling and object tracking. The RCNN Masks keyword can now be used for object tracking, not just object detection. The Crop Mode parameter is now applicable when the ImageNet keyword is set.


7. Surface analysis

For the Surface Parameters tool, four new Parameter type choices are available. Those are: Plan (projected contour) curvature, Contour geodesic torsion, Gaussian curvature, and Casorati curvature. See the Usage notes and the parameter table to learn about these options, but be sure to read the How Surface Parameters works topic for additional information.



We hope that you enjoy the updates and improvements that are available for you in Spatial Analyst for ArcGIS Pro 2.9.

About the Author
Product Engineer on the Spatial Analyst team. Started at Esri when it was ESRI.