Tag Archives: Image Server

Imagery Best Practices

Supporting Imagery and Lidar in the ArcGIS platform has been around for a long time. In the ArcGIS Server Space, Image Server became available at 9.3.1. Since then it has evolved to Mosaic Datasets, Image Services, Raster Functions and now raster analytics. Now imagery is really an integral part of the ArcGIS platform. However, it is only as performant when the imagery is managed and configured optimally.

I often get asked,

  • what format should I store my imagery in?,
  • How many images can be in a Mosaic Dataset?
  • How should I structure my imagery? and
  • what is the maximum number of images per folder?

When answering these questions I have drawn on past experience and advice from Esri. Now though Esri have compiled all this information into an Excellent Centralised resource Imagery Workflows – Best Practices https://doc.arcgis.com/en/imagery/workflows/best-practices/what-are-best-practices.htm

In the Imagery formats and Performance section it details topics such as:

  • File format suitability
  • Recommended imagery formats
  • Reformatting imagery
  • Pyramids
  • Statistics
  • Working with large mosaics
  • Storage system performance

If you’re going to be managing imagery and lidar I recommend you reads these documents. They are comprehensive and invaluable. I must admit I have been doing this for 12 years now and there is information on Lidar management that I did not know about.

This is just one component of the ArcGIS Imagery Workflows documentation Esri has just produced.


Viewing Imagery in a different light

Imagery is the most commonly requested basemap of all the Esri base maps available. It provides the spatial context for your authoritative data that vector base maps cannot. Many questions can be easily answered by visually inspecting the imagery, for example how far is that tree away from the power line? or is that a new swimming pool that does not have a building permit?

There is though more to imagery though than visually inspecting in the Red, Green and Blue visual spectrum. Many of today’s satellites capture information in the Yellow, Near Infrared and Far Infrared portions of the electromagnetic spectrum. In fact the latest World View satellite has 16 bands. So how can this extra analytical data be accessed in a web service?

Image services provide an underutilised feature called functions. These are algorithms that can be applied to an image service on-the-fly to reveal hidden information in the imagery. When an image service published, it is not just the visible band that can be published but all bands from a satellite image can be made available. In its simplest form a function can swap these bands being used to display. For example from Red, Green, Blue to Infrared, Red, Green. Another common analysis is Vegetation health through an NDVI analysis. The difference in chlorophyll reluctance and absorbance of plants in the Infrared and Red portion of the spectrum.  The best part about using functions is that they are only calculated on the portion of image shown on the screen at the resolution displayed and no new image is created, just an interpretation of the image, all performed server side on-the-fly and then sent to the client. There are some very good examples on the Amazon Landsat 8 demonstrator site found at http://www.esri.com/landing-pages/software/landsat/unlock-earths-secrets

Continue reading

Replacing Rasters in an Image Service

In some cases it might be necessary to replace a source image in an Image Server service (.ISDef) and then to rebuild the overviews for the affected area. This might be in situations where new imagery has become available or where there were problems with an original. This blog details the steps required for doing this (without having to rebuild the service from scratch):

– First make a back up of your image service (.ISDef).

– Using the ArcMap selection tools, select the source image to replace (this should select all associated overviews at the same time). You might need to zoom to the source image to achieve this.

– Remove the selected images from the service: Image Service > Advanced > Remove Raster Dataset. In the dialogue box that comes up, tick ‘Remove raster data set’, ‘Delete raster process definition files’ (as long as the selected files are not being used by another service) and ‘Delete derived images’.

– Once the images are removed, build the service: Image Service > Advanced > Build. In the dialogue box, tick on the ‘Compute pixel size’ and ‘Create service boundary tick’ options.

– On completion of the build, add in the replacement source image: Image Service > Advanced > Add Raster Data Set.

– Build the overviews once more: Image Service > Advanced > Build. Again tick on the ‘Compute pixel size’ and ‘Create service boundary tick’ options.

– Next define the overviews: Image Service > Advanced > Optimise > Service Overviews > Define (accept default options)

– Make sure the overviews look correct and then build overviews a final time: Image Service > Advanced > Build. This time in the dialogue box, tick on the ‘Generate derived image’ and ‘Compile service’ options.

– Your image service is now updated

John H