With Geographic Information System (GIS) tools such as those found in Esri's ArcInfo and ERDAS' Imagine becoming more widely available to general users, the ability to analyze geographic data practically becomes commonplace. The capabilities of these tools allow for the use of a multitude of methodologies for solving a single problem, with each methodology having the potential of resulting in a unique solution. In an environment where multiple users may be performing similar analyzes in support of decision makers, it is important to employ standard analysis techniques and present the results in a consistent manner.
Applications such as the Combat Terrain Information Systems (CTIS) Tactical Decision Aids (TDA) developed by Lockheed Martin (LMC) for the U.S. Army Topographic Engineering Center (TEC), provide a general platform for the distribution of standard analysis models and product generators. Using a structured workflow architecture, a robust data storage standard, and a programming language independent Application Programmers Interface (API), an application can facilitate the distribution of standard models as they emerge and, therefore, remove a users need to implement non-standard models of their own accord. This paper will address the CTIS TDA process and design as a methodology for rapidly providing standard GIS analysis models across a distributed group of users.
The CTIS TDA application provides U.S. Army Terrain Analysts with the capability to perform many types of terrain analysis and assists in the preparation of topographic products. CTIS TDA is a seamless integration of the Commercial-Off-The-Shelf (COTS) GIS products Esri ArcInfo and ERDAS Imagine, along with custom-developed C, Ada, and Unix shell scripts.
Analysis products were initially identified for integration into the CTIS TDA workflow process approximately 1-2 years prior to their release. Typical products ranged from intervisibility models (e.g. Masked Area Plots and Flight Line Masking) to standard feature/attribute queries (e.g. Helicopter Landing Zone and Bivouac Site prediction). Specific algorithms were selected according to availability. Many were ported directly from the CTIS legacy system, whereas others were extracted from existing software applications, as provided to LMC by TEC. Examples of the latter are the TIREM Path Loss Models for electronic frequency loss and the NRMMII Mobility models for on- and off-road mobility predictions.
A goal in the development of the CTIS TDA application was to provide a flexible, user-friendly tool for automated terrain data analysis and product creation that could be rapidly upgraded to accommodate new data formats and analysis techniques. It was felt that successful development of an application with these qualities would facilitate the integration of new models and their dissemination to a distributed user base. The key elements of the CTIS TDA design that would provide these desired qualities would be a structured but flexible workflow architecture, a robust data structure for the storage of source and output data, and the development of a non-embedded Application Programmers Interface (API).
A test of the CTIS TDA design was the integration of the Waterways Experiment Station (WES) Tactical Dam Breach (TACDAM) models into the existing CTIS TDA workflow process. The TACDAM models were introduced as a potential system capability late in the development effort of a scheduled CTIS TDA release. Successful integration of TACDAM would determine if the CTIS TDA design would support the rapid dissemination of an analysis model as it became available.
Due to the broad skill base of the typical end-user, a key design aspect of CTIS TDA was the simplification of analysis and production tasks without the removal of advanced COTS capabilities. This was implemented through the development of a flexible workflow process that accommodates the unique aspects of different terrain analysis models and automates their common aspects.
The CTIS TDA workflow provides a finite serial step process for data selection, data analysis, and product creation. The workflow is broken into four sequential steps: Data Setup, Parameter Specification, Footprint, and Finalize. The Data Setup and Parameter Specification steps are unique to each type of model, whereas the Footprint and Finalize steps employ product creation interfaces common to every model. The four step process was justified by the successful integration of all analysis products under consideration for CTIS TDA into the workflow process. By identifying the sequential steps associated with topographic analysis and map production, a generic graphical interface was developed that could ideally support the integration of any topographic analysis model. The following provides brief summaries of the four workflow process steps.
The Data Setup step allows for the identification of input data for a particular model. An example is a dialog for selecting a vector coverage.
The Parameter Specification step provides user access to all parameters that can be manipulated for a particular model. An example is a dialog that allows the user to specify the location and altitude of an enemy observation point for consideration in a masked area analysis.
The Footprint step provides the user with a template showing a default product layout. Existing layouts are based upon an assortment of National Imagery and Mapping Agency (NIMA) map product standards. By using default layouts/marginalia for all products, a consistent baseline for product content can be established across a distributed group of users. During this step, the user may alter the standard layout as necessary.
The Finalize step automates the actual creation of the final map product (digital form). This step automatically performs many of the tedious aspects associated with map creation. What may take over an hour manually, now takes seconds. In addition, the user is provided with access to the final map product through the ERDAS Imagine Map Composer application. Here the user can use existing COTS functionality to make changes to the final product and proceed to creating hardcopy and digital copies of the final product.
Since most of the CTIS TDA analyzes are source data dependent, it was critical that a storage technique be flexible enough to accommodate many complex raster and vector data standards. In addition, the selected storage technique had to also be compatible with the ERDAS and ArcInfo COTS products. For these reasons, source and output data for CTIS TDA is stored within an open hierarchical file structure with Imagine image and ArcInfo coverage formats serving as the respective storage objects for raster and vector based data sets. These file formats were selected because each is robust, open, and supported by the respective COTS products. The following sections detail key aspects of these formats, how they affected the CTIS TDA data storage design, and how this design supports the integration of new data formats into an existing application for geographic analysis.
The complexity of multi-layer vector data sets such as the National Imagery and Mapping Agency (NIMA) Standard Linear Format (SLF) and Vector Product Format (VPF) Interim Terrain Data (ITD) required that a rigid definition for data storage be created. Key design aspects for the CTIS TDA vector data structure were the methodologies used for the definition of projections, the storage of source specific metadata, and the implementation of data dictionaries.
ArcInfo offers many alternatives for defining the projection of source data, especially the Universal Transverse Mercator (UTM) projection. When considering zone, hemisphere, and standard ArcInfo projection definitions, four alternatives for the definition of the projection were identified. However, during the design process, it was identified that ERDAS Imagine only supported one of these options when the geographic area of the data fell in the southern hemisphere.
The generic ArcInfo and ERDAS vector importers (e.g. VPF and SLF) disregard format dependent metadata unless it is earmarked by the user for storage. Because many of the CTIS TDA analyzes are source dependent, it was important to collect the metadata so that it may be available for access when necessary (e.g. identification of relevant data standard). Acknowledging the open definitions for metadata being employed by the data formats under consideration, and that additional formats would be identified in the future, a generic INFO Table format was employed for all metadata records and their descriptions.
Data dictionaries provide a table for the conversion of feature and attribute codes to textual descriptions. Realizing that data standards are often dynamic, and again that additional formats would be identified in the future, a generic INFO Table format was employed for the definition of all feature and attribute codes for each individual data set.
Within the ERDAS Imagine image format, specific nodes are identified for the storage of metadata associated with the image formats supported by ERDAS Imagine importers. Acknowledging the existence of this methodology, a "standard" was easily identified for the storage of metadata for supported CTIS TDA raster data.
A drawback of using this methodology is that most, if not all, of the standard ERDAS Imagine image analysis functionality (e.g. image subset, rectification, etc.) disregard this information. Any historical information concerning the lineage of data is essentially erased. The following illustrates this drawback:
Possible solutions to prevent the loss of data lineage were the employment of rigid naming conventions, the modification of all ERDAS Imagine image access points, and the removal of access to the COTS functionality. All of these proved difficult to enforce and/or cost prohibitive. Even though data management would be less user friendly due to the lack of descriptive metadata, the existing ERDAS Imagine methodology was selected for use in the CTIS TDA data storage technique.
Through the use of open hierarchical file storage and the design of rigid definitions for the storage of data content, data integration is relegated to a single effort: the development of a data importer. Any raster/vector data format can be made available to CTIS TDA and the respective COTS products by simply developing an importer that adheres to the respective standards for data storage.
The final design approach that was implemented to ensure system flexibility and facilitate the integration of new analysis models was the development of an Application Programmers Interface (API) for the workflow process. This API currently allows for the easy integration of any X-Windows based application, including ERDAS Macro Language (EML) and ARC Macro Language (AML), into the workflow process. The goal of the API design was to provide an interface to the CTIS TDA application that would allow for the integration of any new analysis model with minor modification to compiled software. The API defines three insertion points for any new model into the TDA workflow process: Data Setup, Parameter Specification, and Footprint. Key elements of the CTIS TDA API design were the development of application communication software in C, AML, EML, and Unix executables and the definition of intermediate files for specific steps of the workflow process.
The workflow process operates as a parent executable that monitors the progress of its child applications, here being the executables associated with all four steps of the workflow process. Integration of a new analysis model only requires the development of the applications associated with Data Setup and Parameter Specification steps of the workflow process. The input parameters and exit routines for each of these executables are rigidly defined, with functions and/or executables being provided to fulfill the requirements of the exit routines. The following illustrates the process associated with the use of the CTIS TDA application communication software for the Data Setup and Parameter Specification steps:
The workflow process application handles all of the CTIS TDA database access that accounts for the maintenance of all products (e.g. current step the product resides, is it currently being accessed, etc.).
Intermediate files are primarily used to define the contents/layout of a map product within the Footprint step of the workflow process application. The following four files are required to create a map product for manipulation in Footprint and Finalize:
The TACDAM analysis models provide the capability to predict the effects of a dam breach. These models were originally developed in ArcInfo, independent of the CTIS TDA design. The models used SLF ITD (ArcInfo coverage) as source data; however, the format differed from that employed by CTIS TDA. In addition, the TACDAM process was integrated as a single application that did not initially fit into the four step workflow design.
Since the TACDAM SLF ITD data storage format used the ArcInfo coverage format, it was rather easy to modify the existing models to operate on the CTIS TDA SLF ITD storage format. Minor changes were required to accommodate projection definitions and nomenclature. The use of a common module for data access by all TACDAM analysis models facilitated this migration.
Modification of the original TACDAM application was slightly more complicated. The original application allowed for the simultaneous generation of multiple products, a functionality that was not currently implemented in the CTIS TDA workflow. In addition, the existing software did not use any of the application communication software that would allow an application to communicate with its parent workflow application.
To integrate the TACDAM models into the workflow process, all models were segmented so that each product could be generated independent of another product. This resulted in the development of a unique "Parameter Specification" application for each TACDAM model. Since the models all required the same source data, a single "Data Setup" application was developed for all of the TACDAM models. Throughout the development of the CTIS TDA applications, the appropriate application communication functions were implemented.
The development efforts associated with the TACDAM integration were carried out by WES with technical guidance from LMC. The data storage and workflow designs of CTIS TDA, along with its well-defined API allowed for this to be accomplished in a relatively short period of time without requiring significant development efforts from LMC.
The initial step towards facilitating the integration of analysis models into an existing software application is the conceptual definition of a user interface that is flexible enough to accommodate multiple types of analyzes. The design implemented with the CTIS TDA application employs such a concept through its workflow process user interface. This interface allows for the integration of any geographic analysis that can be broken down into a four step serial process:
The development of a data standard that is flexible enough to accommodate a myriad of data formats is also an essential element of a flexible system. By using ERDAS Imagine image and Esri ArcInfo coverage formats as storage objects for geographic data of the CTIS TDA database, new data formats can easily be integrated through the development of data importers.
A final aspect of the CTIS TDA design that facilitates the integration of new analysis models is the definition of an API. Through the use of the CTIS TDA API, models can be readily integrated into the CTIS TDA workflow with minimal modifications to the software architecture. In addition, this API has relieved restrictive dependencies on a specific GUI and analysis programming languages by providing support for C, EML, AML, and Unix Shell script functionality.
The TACDAM integration was a demonstration of the flexibility and robustness of the CTIS TDA design. By modifying the TACDAM application to make use of the CTIS TDA API and data storage formats, new analysis models were successfully integrated into an existing software application. This was done quickly, and with minimal effort, in response to the stated needs of the CTIS TDA end-user.
In today's environment of desktop GIS analysis, providing a medium for the timely distribution of analysis models as they emerge is the only way to insure standardized analysis across a large distributed group of users. The CTIS TDA application, the implementation of a flexible and robust software design concept, serves as a model for the design of GIS applications that facilitate the distribution of standard models for GIS analysis.
William R. Zanine
Senior System Engineer
Lockheed Martin Corporation
1301 Virginia Drive, Suite 305
Fort Washington, PA 19034-3217
Telephone: (215) 283-6966
FAX: (215) 283-6970
zanine@fwo.mds.lmco.com
--------------5012B8D6805--