Going with the Flow
Written by Cheryl Gerber
GIF 2012 Volume: 10 Issue: 1 (February)
By stringing together tasks and guided, interactive processes, GIS workflow has become the de facto standard framework for defining the work and the flow—how the work is routed—from initial task to completed process and actionable product.
The streamlined simplicity of workflow has harnessed complex geospatial data and systems in recent years, rendering GIS production more effective. One of the biggest benefits is integration. Military and intelligence are often working in separate acquisition, delivery and contracting offices, but workflow can serve to unite them.
“Geospatial data production leverages workflow, as it lends itself well to repeatable workflow processes with production-line capacity for the most efficient execution,” said Greg Pleiss, Esri solutions manager, Professional Services division.
GIS workflow automates common geo-processing activities, ensures standardization and consistency across operations, and manages geographically dispersed workforces. The software is growing more capable and flexible, with user-configurable tools to accommodate everyone, from those who define the workflow to those who execute tasks in standard ways. Step-by-step workflows now walk users through previously complex image analysis tasks.
Esri’s ArcGIS Workflow Manager serves to structure the work and flow in standard, repeatable GIS tasks as business processes across an enterprise. As users execute each step in a workflow, the Workflow Manager launches disparate applications, configured as part of the workflow behind the single interface.
“Software adds execution behind the process using dialogue boxes, like: ‘Is this imagery ready to use?’ If the user’s response is ‘No,’ and additional processing needs to take place, then the system takes the user back to re-task image collection. If it’s ‘Yes,’ then feature extraction would be the next step in that workflow,” Pleiss said.
ArcGIS workflow users can configure a question step or a series of choices. Each selection leads the user down a different path toward software to launch for the next task.
Guided Software
Many geospatial imaging companies today offer guided software to increase GIS workflow productivity. In its ENVI image processing and analysis software, for example, Exelis Visual Information Solutions (formerly ITT VIS) provides discrete, taskoriented workflows that usher users through step-by-step processes to create specific end products.
“We create interactive software processes to increase productivity through increased integration and automation,” said Peter McIntosh, solutions engineer, Exelis VIS Geographic Intelligence Systems Group.
One example is the process of developing a map workflow to create a helicopter landing zone (HLZ) for troops in theater. The HLZ map is then populated or published in the GIS for actionable intelligence.
GIS workflow expanded into cloud computing in 2010, when Esri’s ArcGIS Server 10 became available on the Amazon Elastic Computer Cloud (EC2). In 2011, PCI Geomatics released its GeoImaging Accelerator (GXL) large volume image production system on Amazon’s EC2 to support high resolution content production for the ESRI ArcGIS online platform. GXL uses high performance computing to optimize the speed of distributed automated workflows for industrial-strength production.
As part of its image preprocessing technology stack, PCI Geomatics also offers GeoImaging tools for ArcGIS, which supports more than 20 different satellite sensors, and Geomatica, a stand-alone desktop image classification package with the ability to build in automated workflow.
When mobile enhancements were added to GIS workflow on the cloud in 2011, the ArcGIS Server ArcPad Extension began to offer the latest release of ArcPad (10.0.2) to abet the process of field mapping, data collection and updates from the field to the cloud.
Most GIS workflow takes place at the tool and task or process level rather than the systems level, although GIS and non-GIS workflows unite at the systems level for decision support. Software tools supporting tasks often start with capturing imagery in such processes as creating an HLZ.
“A surveillance analyst seeking a process to identify an HLZ to publish to a broader audience first captures the imagery from satellite or sensors, and then starts the orthorectification workflow for accurate geospatial parameters. Next, the feature extraction workflow defines and classifies helicopters into vector-based, feature- classed information,” said McIntosh.
“To determine the quantity and quality of movement, the analyst enters and runs the change detection workflow to see what features have changed and how they have changed,” he said. “At the end of each workflow we provide the option to publish the result of the work product into the GIS.”
Step-by-Step Process
For example, the ENVI orthorectification workflow, integrated with Esri’s ArcGIS, provides a step-by-step process to remove geometric distortions introduced during image capture. The workflow produces a map with planimetric geometry and orthorectified imagery that is registered to a ground coordinate system with consistent scale throughout the image. Planimetric or flat plane geometry approximates the round surface of the earth by projecting it onto a flat plane.
To increase the accuracy and simplicity of the orthorectification workflow, Exelis added a method called rational polynomial coefficients, which allows a wider variety of sensors to be processed.
The software is designed to be easy to use. Users select the input image they want to orthorectify from various types of commercial or multi-spectral sensors, Internet Explorer or ArcGIS, then drag and drop the selected images into ENVI. They can confine image processing to specific areas of interest to reduce processing time. Next, they select the output parameters such as pixel size or file name and path. A preview feature allows users to check orthorectification results without having to process the entire dataset.
ENVI uses both feature extraction and classification workflow. However, feature extraction is object-based image analysis rather than pixel-based classification workflow. “The object-based approach is supplanting the traditional pixel-based approach since it puts out GIS-ready feature classes as vectors, not rasters or pixels, with a rich set of attributes,” said McIntosh. “Vector has properties associated with it so you can colorize.”
While raster images are based on pixels and grids of pixels creating bitmaps, vector images use mathematical relationships between points and the paths connecting them to describe an image. Therefore, vector graphics are composed of paths. Bitmaps require higher resolution and a spatial anti-aliasing technique to create a smooth appearance while vector-based graphics appear smooth at any size or resolution since they are mathematically described. Spatial anti-aliasing minimizes distortion artifacts when representing a high-resolution image at a lower resolution.
Change detection in geospatial imagery is another example of a previously tedious, though vital, analysis task that is greatly simplified by workflow. First, GIS workflow users select the images they want to analyze, such as two images of the same scene taken from different satellite sensors. Next, they might select a before and after comparison of two images of the same location, based on whether the data is raw, already processed or classified. They can choose a thematic change method to detect changes in specific features over time, such as buildings, roads or natural land cover to highlight changes that have occurred in categories.
Change thresholding allows users to set parameters to identify the magnitude and type of changes that have occurred. Data cleanup refines results to make features appear more visually realistic. Finally, users can preview, export and choose how to use the results, whether in a PowerPoint presentation, a geodatabase or directly into an ArcGIS file to create a map.
Web Capability
GIS workflow is expanding continually beyond pixels and desktops into the realm of vectors and servers. One of the most notable improvements to Esri’s ArcGIS Workflow Manager is the development of web capability. “We’ve increased focus on the development of the server side, by building out web capability through the use of REST interfaces,” said Pleiss.
Representational state transfer (REST) is a style of software architecture for distributed hypermedia systems such as the web. It is considered an alternative to the Simple Object Access Protocol (SOAP) interface, an existing standard developed by Microsoft for exchanging XML-based messaging on the web.
SOAP was designed to integrate with other standards but is exclusively XML-based, whereas REST information exchanges can be in any of more than 350 Multipurpose Internet Mail Extension types. SOAP exchanges messages between SOAP nodes, whereas REST both captures information from resources and updates resources with information.
Esri has made use of the flexibility inherent in the REST architecture. “SOAP exchanges verbose XML responses back and forth across the web. It’s work intensive, whereas REST architectural style reduces every resource on the web to a URL. It’s a lighter weight, simpler way to do web services communications,” said Pleiss.
ArcGIS Workflow Manager uses Open Geospatial Consortium (OGC) standards for GIS data integration and the Business Process Execution Language (BPEL) for standards-based workflow. BPEL is a standard executable language set forth by the Organization for the Advancement of Structured Information Standards (OASIS).
BPEL is for specifying actions in business processes with web services. BPEL processes export and import information by using web service interfaces exclusively. OASIS is a global consortium that drives the development, convergence and adoption of e-business and web service standards.
Another ArcGIS Workflow Manager development is the addition of spatial notifications, most notably for change detection. Notifications are more organizationally significant than they might seem initially. They serve to share vital information in a timely fashion across departments and can streamline project management.
“We provide a plug-and-play framework that allows the configuration of notifications when certain changes happen to geospatial data,” Pleiss said. “You can plug in your chosen notification type, such as email.”
“A lighthouse is a nautical, aeronautical and topographic image, so all three would subscribe to notifications of any change in the lighthouse,” Pleiss said.
Spatial auto-notifications and alerts are rule based. If there were a change in the lighthouse, for example, then subscribers who had chosen to receive notifications that way would receive an email indicating the nature of the change.
The flexibility of the notification function reflects Esri’s continued build-out of service-oriented architecture capability in GIS workflow. “It allows the plug-in of your own notification engine, which can trigger a new business process in another part of an organization,” Pleiss continued. “This lays the groundwork for cross-business unit collaboration in a proactive and automated fashion. And it enables the seamless integration of GIS and non-GIS users and business processes.”
Esri has been building GIS workflow capability to integrate business processes and bring non-GIS users into the GIS environment. “In the last two years we have undertaken a program to implement workflows that brings in the contracting officers who are the liaison between the agency and the contracted company for a seamless transfer of geospatial data,” said Pleiss.
Esri is in the process of rolling out GIS workflow at the executive and management level for actionable intelligence in the decision-making process. “It provides a window of visibility for up-to-the-minute, live workflow status without having to be a heavy GIS user. It allows executives to prioritize resources, tasks and activities,” he said.
New Technologies
GIS workflow increasingly has incorporated new technologies into the process. For instance, PCI Geomatics recently completed beta development of its synthetic aperture radar (SAR) GeoImaging tools for analysis of SAR images in ArcGIS.
Users can create application-specific workflows since the SAR tools are integrated in ArcGIS. “The current most common application workflow for SAR imagery is maritime domain awareness, which detects ships on the ocean for the Navy and Coast Guard,” said Kevin Jones, director of marketing and product management, PCI Geomatics.
SAR is a form of radar that uses relative motion between an antenna and its target to provide distinctive, long-term coherent signal variations, which are used to obtain finer spatial resolution than is possible with conventional beam scanning. SAR is implemented by mounting a single beam-forming antenna on a moving platform such as an aircraft or a spacecraft. The target image is repeatedly illuminated with pulses of radio waves at different antenna positions. The echo waveforms are then rendered coherent, stored and processed for an image of the target region.
“SAR satellite sensors cut through all weather, day or night. You can collect images regardless of clouds or darkness based on back scatter or sound bouncing,” said Jones.
“To realize the potential of SAR imagery, we want to translate it from the technical realm to the application level so end-users can use this technology to solve business problems,” he said. “PCI has integrated 40 SAR-specific algorithms into its GeoImaging tools.”
The new SAR tools for ArcGIS include coherent change detection, classification of multi-polarized imagery and advanced utilities to filter and analyze imagery. The tools support commercially available data from numerous satellites.
Despite the simplicity of workflow, the huge volume of geospatial data behind it can still clog installed systems. “GIS workflow on the cloud has [input/output] issues at multiple levels. Loading these large volumes of geospatial data uses huge bandwidth on the cloud, so we had to optimize our code to reduce the number of read/ write operations to optimize I/O operations in cloud processing,” noted Jones.
“Sometimes the cheapest and fastest way to get large volumes of images loaded onto the cloud is still to FedEx it to the data center,” he added.
I/O issues are a key hurdle in cloudbased geospatial data processing, from the initial transfer of earth observation imagery data to the actual processing of content. “But at the same time, there is improved scalability on the cloud and it saves on hardware purchases and maintenance,” Jones said.
Workflows will continue to evolve to solve the problems users are facing. “We keep defining more workflows to address specific problems. Down the road, we’ll have workflows that say: ‘Go find IEDs in a specific location.’ We are moving toward making it one-click, easy flow analysis,” said McIntosh.
“The workflows will become increasingly simplified as the algorithms underneath them become increasingly sophisticated,” he said. ♦
