MOSS – A Historical Perspective
Back in 2004, I browsed several books on GIS, including one on the history of GIS. Out of personal interest, I checked out what the authors had to say about MOSS, the Map Overlay and Statistical System. While MOSS was mentioned several times, there was a lack of both historical accuracy and content. Since I “lived” the MOSS story, I felt that it would be interesting to go back, brush off a number of documents from the 1970’s and write a short but definitive history of the early days of MOSS. One item I quickly discovered in my research is that MOSS as a software product is still alive and well and available as OpenSource! Not bad for a GIS software product that was first developed in 1978.
The development and use of MOSS was a very important milestone in the evolution of GIS. MOSS was the first broadly deployed, vector based, interactive GIS. Second, it was the first GIS to be deployed for production use on mini-computers. Third, it was the first GIS to provide integrated vector and raster processing. Finally, and perhaps most importantly, dozens of States Federal agency’s staff were able to cost effectively use and learn about GIS at a time when there had been very little exposure as to the power and usefulness of GIS. In a sense, MOSS provided the educational springboard that allowed many of these agencies to use the lessons learned in implementing and using MOSS to grow and expand their GIS “reach”, purchasing and using more powerful, commercially supported systems.
In the middle 1970’s, coal-mining activities in the Rocky Mountain States began to accelerate. In response, the U.S. Fish and Wildlife Service was tasked with evaluating the impacts of strip mine development on wildlife and wildlife habitat. They were further tasked with evaluating and making recommendations regarding habitat mitigation. Professionals within the USFWS felt that the (at that time) nascent potential of GIS as an analysis and modeling tool was exactly what was required to aid the wildlife biologists map habitat and develop habitat mitigation scenarios.
With funding from the EPA Coal Program, the USFWS issued a Request For Proposals (RFP) in early 1976. As documented in the RFP, the scope of the project included doing a User Needs assessment, developing a GIS functional scope, evaluating existing GIS technologies, and making recommendations to the USFWS as to the appropriate course of action for the development and deployment of GIS technology. In late 1976, the contract was awarded to the Federation of Rocky Mountain States, a not for profit organization that eventually evolved into the Western Governors’ Policy Office. The USFWS group given responsibility to oversee the contract and participate in the User Needs assessment and demonstrations was the Western Energy and Land Use Team (WELUT). The project leader for the FRMS was Larry Salmon and the COTR for the Government was Ken Burnham. The complete FRMS GIS team came together in January of 1977 and consisted of myself, George Nez, Jim Gropper, and John Hammill. I was hired as the GIS architect and programmer.
For the first six months of 1977, we worked on two tasks: A User Needs Assessment and an Inventory of Existing GIS technology. The needs assessment involved interviewing wildlife biologists, natural resources planners, and other professionals that would be involved in wildlife habitat definition and habitat mitigation. The objective of the interview was twofold. First, to elicit the types of GIS functions and GIS environment that they would need in order to fulfill the mandated mission of the mitigation studies. The other was to begin educating them on the potentials uses and benefits of GIS. FRMS staff and WELUT staff traveled to numerous USFWS field offices and interviewed dozens of individuals. The results of the assessment were published in the summer of 1977.
Concurrently, I was doing an inventory of existing public domain and commercial GIS technology. Using previous work done by the IGU (International Geographical Union), personal contacts, and research, approximately 70 different mapping and GIS software packages were identified. Of these, 54 had enough documentation and basic required functionality to warrant further analysis. A basic set of information was collected on each system (hardware, operating system, basic GIS functions supported, etc.). In a sense, the survey and inventory was similar to those now published in various GIS magazines and industry analysts’ reports. The main difference is that we collected the information as opposed to sending out a survey and asking companies and individuals to complete the survey. The USFWS published the results of the inventory the summer of 1977. This document provided information on product name, programming language, required operating system, GIS capabilities, data model(s) used (vector vs. raster), level of documentation, and whether support were available – or not. This document is a valuable historical document as it has information and details of systems long extinct and forgotten.
The results of the User Needs Assessment coupled with new industry and technology knowledge allowed us develop a GIS Functional Needs document that specifically focused on the GIS requirements for the USFWS and the habitat mitigation process. We then were able to compare the functional specification against the inventory of existing systems. It became quickly apparent that none of the existing systems could meet even half of the requirements of the USFWS. Further, all the inventoried systems operated in batch mode. There was a general feeling among the USFWS/WELUT staff that an interactive GIS would be much more useful as part of a decision making process. Since I had used both batch and interactive GIS applications in graduate school, I strongly supported the interactive approach. So we had to make a “build or buy” decision.
A very critical and interesting discussion took place at this point. Some FRMS staff and some at WELUT were sold on the idea of using a batch mapping system called CMS (Computer Mapping System). Functionally, it was similar to Symap. Its primary output mode was the line printer. As far as I know, it is the only “GIS” ever written in COBOL! After weeks of argument, the decision was made not to use CMS. Also, since none of the commercially available systems at the time had the required functionality or they were way too expensive, the decision was but design and implement a new GIS based on the requirements uncovered by the needs assessment and documented in the functional specification. The stipulation by the decision committee was to use as much of the public domain code documented in the inventory as was feasible in the new system.
With this background of well-documented requirements and systems information, the design of MOSS began during the summer of 1977. Once the group agreed on the design, programming started. Eventually, a set of shared libraries of common functions, such as a graphics library, a math library and a text processing library were developed. These shared libraries with documented interfaces allowed for the very rapid development of the complete software system. Many of these base level subroutines were extensions of work done by myself and other geography graduate students in the GIS laboratory at SUNY Buffalo. The development environment was a CDC mainframe running the KRONOS operating system. Fortran IV was the development language. Graphics presentation and code development was done on a Tektronix 4010.
Since the decision was to implement an interactive system, we needed a user interaction language. Using the results of the user needs assessment, we were then able to define an English like language for user interaction for the new GIS. We decided to use simple action verbs (PLOT to plot a map) with simple subjects and modifiers (PLOT ROADS) to plot a roads map. We also raised the issue of how once a user has selected a map set, they could continue processing on the result of the select without having to constantly re-select the map or set of maps. We came up with the concept of “Active Ids”. These ID’s are equivalent to views or query sets. For example, if the user selected interstates from a road map, the result of this select was “remembered” by the system and given a unique identifier. The use could then reference this selected map set by its ID number. So, for example, the user could then enter PLOT 3 to plot map set ID 3 or they could enter OVERLAY (2 4) to perform polygon overlay on selected map sets 2 and 4. This approach made it much easier for users to develop and use sequences of commands to perform a given work flow or modeling operation. We further extended this concept so that users could save this save and reuse this information from working session to working session.
Using the defined user language, the functional specification, and the systems design for guidance, in late 1977, I began implementing high-level user action functions in the software. These higher-level functions accessed the shared libraries already previously built and tested. Extensive use of existing public domain code helped accelerate the development process. For example, code for point in polygon was extracted from software done my David Douglas, a polygon cross hatching and fill capability was “borrowed” from code by Waldo Tobler, and the vector to raster conversion software was a recoding into Fortran of the COBOL code in the CMS system. As a point of interest, the commitment to using existing public domain code continued through the full development and deployment lifecycle for MOSS.
This approach allowed for the rapid development and use of MOSS in a Pilot Project in 1978. The Pilot Project was done for several reasons:
Provide a validation of the capabilities and implementation approach for the MOSS GIS;
Help educate FWS staff on the user and power of a GIS;
Provide iterative feedback to be used to evolve and enhance MOSS.
Prove to the funding organizations that they were getting their money’s worth.
By 1979, version 10 of MOSS had been implemented on the CDC mainframe. This was the final version developed as part of the FRMS contract. At this time, I became an employee of the USFWS WELUT organization. Denny Parker, the founder of what is now GeoWorld, was my boss. My first task as a Federal employee was to port the CDC version of MOSS to a Data General C330 minicomputer running the AOS operating system. Another part of this project was to integrate MOSS with another software package that had been developed under contract by Autometric for the National Wetlands Inventory. The Wetlands Analytical Mapping System (WAMS) was an advanced map digitizing and edit package for topologically structured vector data. All work on these projects was completed in early 1980. As far as I know, WAMS was the first interactive digitizing system for capturing and structuring map data as topology in real time. But WAMS is another story worth of its own chapter in the history of GIS.
By the middle of 1980, the WAMS/MOSS software suite was ready for production use not only within the Fish and Wildlife Service but also in other Federal agencies. The Oregon state office of the BLM was the first non-FWS group to install and use MOSS and WAMS. Within a couple of years, MOSS was being used in the Bureau of Indian Affairs, multiple BLM State Offices, the Bureau of Reclamation, National Park Service, US Army Engineering Topographic Labs, Fish and Wildlife Service, and numerous State, Local and University organizations.
For the next several years, MOSS was extensively and successfully used by these agencies to complete numerous projects. In the process, hundreds of individuals gained hands-on experience using GIS technology in the US Federal Government. The USFWS staff and contractors developed complete systems documentation, detailed User’s Manuals and Guides, Reference Manuals, and tutorials. In the early 1980’s, a series of annual MOSS users’ conferences started. Perhaps the widespread use of MOSS helped break down the barriers to the use of GIS technology, helping to pave the way for the adoption and use of geospatial technology today.
But the MOSS legacy does not stop here. In the early 1980’s, Autometric under contract to the BLM and USFWS, provided on-going support and maintenance of AMS and MOSS. They also provided technology transfer services to install the software into both Federal and Local Government sites, provide training, and ongoing support. Autometric staff also ported the software to other operating system environments, such as VMS and UNIX. This work allowed Autometric to build an extremely experienced staff of GIS programmers and support professionals.
In 1983, the USGS EROS Data Center contracted with Autometric to do a study and design for enhancing MOSS to be based on a topological data structure as opposed to the full polygon approach. This activity was completed in 1984. While never implemented, the knowledge gained in the study was instrumental in the design of DeltaMap.
In 1984, a small group of the engineers decided the time was right to design and implement an entirely new GIS software package. A six month intensive design period ensued. The design took into consideration all the strengths – and weaknesses – of both AMS and MOSS. The team considered lessons learned in porting AMS and MOSS to UNIX. They also took advantage of the results of several projects that focused on the data structures and algorithms needed for real time collection and maintenance of topologically structured spatial data. By April 1985, version 0 of DeltaMap was publicly announced. Since design and implementation of DeltaMap reflected the strengths of AMS and MOSS, such features as the English oriented command language, session state information, and spatial views were maintained. However, contrary to some industry beliefs, DeltaMap represented almost 100% new code. Because DeltaMap used a topologically structured data model as opposed the full polygon data model of MOSS, very little MOSS code could be used to develop DeltaMap. DeltaMap – later GenaMap – was first installed at a customer site in 1986. GenaMap lives on today and is now an integral component of a broad set of applications for the telecommunications industry being deployed in Europe and South America.
So, from its somewhat humble beginnings in the US Fish and Wildlife Service, MOSS flourished for many years and enabled dozens of agencies and Federal offices to compile a comprehensive digital map database and complete numerous projects. Much of these data have since been converted into other GIS formats. All of the agencies that began their GIS experience with MOSS have moved to using a variety of commercial products. But the MOSS legacy lives on in having paved the way for these organizations being able to make better and more effective use of GIS technology to meet today’s policy and mandate challenges.
Some good references that detail the functionality that was available in MOSS are:
Reed, C.R. 1982, Map Overlay and Statistical System User's Manual, Western Energy and Land Use Team, U.S. Fish and Wildlife Services, Fort Collins, Colorado.
Reed, C., 1984, Map Overlay and Statistical System (MOSS), System Reference Manual, prepared by Autometric, Inc., Falls Church, VA
These references may be difficult to find. There is an on-line manual from the PC Version of MOSS as developed by Saul Katz:
http://www.ibiblio.org/pub/academic/geology/COGS/MOSS/mossdoc.txt