GeoConnections - Canadian Geospatial Data Infrastructure Working Group Meeting Tools for Framework Data
May 31st, 1999
Timothy V. Evangelatos
|
|
||||||||||||||
Peter Paul opened the workshop by welcoming the more than 70 participants. He noted that the composition of the workshop was about 1/3 from industry and the remainder from federal and provincial government agencies. He felt that this was a good mix for the exchange of ideas. He then introduced Pierre Lafond who was the facilitator for the workshop.
Using a slide presentation (Attachment 3) Pierre Lafond provided a brief background and context for the workshop. Although a focus for GeoConnections is on making geospatial accessible through the Internet, the provision and use of such data will depend heavily upon the availability of appropriate tools and the community is looking to the commercial sector to provide them. The first workshop on data integration and tools (Reference 1) was hosted by the NRCan Center for Topographic Information in November 1998, and the conclusions reached are described in the next presentation. In February 99 and April 99 the GeoConnections Secretariat also organized two national meetings to deal with framework data and a number of related issues. Reports on these meetings are available (Attachment 2 and 3). In February 1999 the federal government announced the $60M GeoConnections program which includes a component for partnering with industry which will be the topic of a 3rd presentation. These presentations would be followed by reports on a number of related activities. Pierre introduced the workshop objectives which were:
1) Data Integration Tools Workshop: Sylvain Latour had organized and chaired the first workshop on tools. He saw that agencies such as the Inter-Agency Committee on Geomatics and Canadian Council On Geomatics were focusing heavily on data but not paying enough attention to the ensuring that appropriate tools would be available for creating, integrating, and using the framework data. In his experience, when users try to integrate geospatial data, they may spend over 80% of their time and resources in trying to integrate data from various sources. His vision is to reduce this effort by 1 or two orders of magnitude. At the last workshop, a consensus was achieved on four important aspects (Attachment 4) which are summarized below:
Henry Kucera felt that the time-line suggested by the first tools workshop was fairly conservative and he felt that it could be cut in half. Sylvan agreed that if we can work together effectively the full CGDI infrastructure could be realized much sooner.
Costas Armenakis inquired about the inclusion of a server for image based data in the CGDI vision. Sylvan replied that such data would be handled in the similar ways as other types of data and his discussion really applies to all forms of geospatial data.
2) Information Interoperability Institute (III): Bill Lalonde is the acting president of III, a not-for-profit organization, whose goal is to foster global information interoperability through such standards as OGDI (Open Geospatial Datastore Interface) and RDA (Remote Data Access), two related elements that must become part of the CGDI implementation strategy. OGDI is a Application Programmer’s Interface (API) that "opens" the door to geospatial data, whereas RDA is a similar specification for non-spatial data. The Secretary of State is providing $436K to establish III and an additional $400K is expected from DND and GTIS, to cover R & D and the integration of RDA and OGDI. The proposed III business plan predicts a $1.5M budget over five years which will be achieved through memberships and related R&D projects.
Last week III hosted a 2-day technical workshop. The first day dealt with presentations by software vendors who have implemented the ODGI API. They described, and in some cases also demonstrated their solutions. Nine vendors discussed their work and some of the problems they had in implementing the OGDI specification. Up-to-now OGDI has been supported by an ad hoc group and the lack of a totally fixed, robust, well defined specification has caused some difficulties as various implementations differ slightly. As Bill explained, the purpose of the III is to solve issues such as these. The emphasis on the second day of the workshop was on what III is all about. Many issues were discussed, for example the processes to be implemented to deal with extensions to the core specification, OGDI support and promotion, relationship of III to the international standards bodies, where III are actively participating and in some cases leading certain initiative (e.g. OGC Web Mapping Testbed). A report on the III Technical Workshop is being prepared and will be available shortly. A new temporary web site is being installed at http://132.156.30.81/iii and a permanent home will be available soon.
Rupert Brooks wondered about the relationship of OGDI with other international standards efforts and if III would use their model. Bill replied that a key activity of III is to contribute to ISO TC211, ISO SC34 (SQL/MM), and OGC, as well as CGDI. In response to Henry Kucera’s question about memberships Bill noted that a preliminary list is on the III web site.
3) Funding Opportunities for Framework Tools Development: Paula Rojas provided a summary of the GeoConnections seven program elements which are reproduced below (Attachment 6):
Henry Kucera asked about the skills network. He wondered if there were opportunities for skills transfer. Paula argued that although training is important, there are other programs that are focusing on that aspect.
As a public servant that may be involved in proposal evaluation as well as the program advisory committee, and at the same time a probable partner in some of those proposals, Rick Morrison was concerned about the possibility of a conflict of interest. Paula noted that evaluators would be excluded from rating any project in which they were partnering. Rick continued with a concern about what he called the practice of "geoconnectivism"- which is a challenge, and that of "vendorism" - which is easy to do. He asked the workshop: how do we avoid the latter and move the former ahead? Paula commented that the vision for CGDI is that of a shared leadership and we must work together to make CGDI successful.
Neil Anderson added that the Canadian Market is quite small and often governments do a disservice to small businesses. The government may help get a business or service created in Canada, but then the company finds that the local market is too small to support the business and it folds. This led to Neil’s question about the opportunity for leveraging potential projects with international partners. Paula replied that this is fine. The proposal must come from a Canadian company but there are no restrictions about who it partners with. (I thought that the IP had to remain in Canadian hands.)
4) GEOIDE R&D Investment Program: Daood Aidroos provided an overview of this $10M program which may provide opportunities for leveraging resources to build the CGDI. Over 30 SME’s, 25 universities, and 6 government departments, are collaborating in 20 projects (Attachment 7). Daood described the organizational and legal structure and gave examples on why GEOIDE has been formed. A problem for the country are the large number of geomatic data islands that need to become linked to enable decision makers to make informed decisions. He concluded with a description of a proposed evaluation framework for the projects and the program itself
5) The GEOIDE Network - Opportunities for strategic investments: Jeff Edwards provided a description of the investment opportunites and priorities for GEOIDE (Attachment 8), which are reproduced below :
As Pierre Lafond had mentioned in his introduction the three background papers would be followed by short presentations from several federal agencies who are involved with some aspect of framework data tools.
6) Marine Geomatics/DFO: Paul Bellemare gave an overview of DFO’s activities which cover a wide range of tasks involving geospatial data. Some of these are:
|
|
Paul listed the wide variety of geospatial data sets collected by DFO (Attachment 9) and then focused on the following three data sets that have been identified as being part of CGDI framework data:
Tim Evangelatos asked if the work done by the ACZISC on standards such as developing a single definition for "coastline" would be incorporated in the new effort for the MSDI. Paul responded affirmatively. He sees the ACZISC being an active participant in the work.
Henry Kucera noted that members of the Mercator Alliance have been looking at the coastal zone marine infrastructure requirements. Paul acknowledged that priorities for the coastal zone are recognized and things are moving. He added that ACZISC is pulling parts of this together.
Daood Aidroos asked if there were any "killer" applications identified in the coastal zone or marine areas. Paul felt that there were many opportunities to help make data available to so that managers could make better critical decisions. One example that he suggested was the management of the ecosystem.
7) DND Framework Activities: As Dave McKellar had been called away, Jennifer Hum-Millar was asked by DND to prepare and give the presentation. Currently DND’s framework activities cover the three areas of:
With regard to GEOIDE, they have two related projects. One is to improve decision making using WWW and the second is the development of automated techniques to extract, generalize and access geospatial information from hyperspatial remotely sensed data (Attachment 10).
Neil Anderson noted that issues of the timely updating of geospatial data has not been adequately addressed and should be a major consideration. Diane Richardson added that the updating process is being investigated in certain GEOIDE projects. Similar to DND’s work they are looking at extracting features from satellite imagery, but a lot of research is still required.
8) Statistics Canada and Elections Canada: These two agencies have been collaborating on creating a National Road Network and Sheila Rambeau and Pierre Lermusieaux chose to give a joint presentation. According to Sheila, Statistics Canada’s main application is the census and they require a national network upon which to conduct the enumeration. It is important for them to have data on roads and street addresses. Elections Canada require similar information as they must know where people live As the two agencies have common requirements, a partnership between the two agencies was created to maintain the road and address data base. Formats and software compatibility are issues and Pierre will expand on them later. Other features such as coastline are also important in order to provide context for the roads boundary data.
Sheila concluded the presentation by listing a wish list for the two agencies. It includes:
Antoine Terjanian was interested in the problem of rural addressing. He reported that in Quebec civic addresses exist for all citizens and wondered if Statistics Canada was taking advantage of that. Sheila responded that there is a great deal of variety in how rural addresses are handled across Canada. Provinces like NS and NB that have implemented 911 are in good shape. In general this has been more of a municipal and telephone company activity rather than a provincial responsibility. The prairies have little 911 coverage. She added that getting access to appropriate data, which is often owned by the telephone companies and can be a problem. Further the assignment of new addresses is not done consistently across the country.
9) Center for Topographic Information: Christine Parent and Pierre-Yves Boisvert also provided a joint presentation. Christine provided an overview of what has been happening at the Centre for Topographic Information (CTI) with respect to the CGDI Data Alignment Layer (CDAL), while Yves described plans for the NTDB.
Pierre-Yves described the four main areas of interest for CTI (Attachment 12). These include
Antoine Terjanian noted that there are other sources for updating the NTDB besides the Landsat 7 data. A major upgrade of the cadastre has just been completed in Quebec, and could be used. Sylvain agreed and CTI would prefer to work with the provinces.
10) Framework Tools Requirements of the Atlas of Canada: Rupert Brooks described the efforts of his division to bring together many different data sets into a solid framework to support a variety of mapping applications at scales of 1:1M and smaller. A base is being built from VMap Level 0 data and there are a number of requirements for new tools. The three main areas where tools are needed are generalisation, conflation and toponymy/text placement (Attachment 13). From the 1:1M base, generalisation to various scales down to 1/30M are required. In addition, generalisation operators for aggregation, smoothing/ simplification, selection, symbolisation and exaggeration are required. Internally to realize these goals two projects are underway. The first one in cooperation with Dr. Diane Richardson (CCRS) involves the generalisation of hydrology networks, and the second one on developing polygon aggregation routines is being done in collaboration with Dr. Lesley Chorlton (GSC). Rupert concluded by expressing interest in working with other partners to improve and develop the tools needed to carry out their work. Rick Morrison asked if users would be allowed to extract data from the National Atlas and use it in other applications. Rupert replied that through GeoGratis users are able to obtain some National Atlas Data. More will be available soon once they have completed their upgrades to make the data more analytically sound. Hydrology is the first priority with roads coming next. Rick further noted that GeoGratis provides a file-based download and he wondered about plans for providing an API for direct access to the National Atlas. Rupert noted that there will be an on-line map-making program with an available API that allows users to create maps on-line and then view them on the screen as a GIF file. Peter Paul added that images are provided because they do not have the right to provide all the vector data free of charge. As for providing vector data, chunks of the VMap Level 0 data are available at no charge but there are no plans to provide a custom extracting mechanism. Henry Kucera indicated that the data warehouse implemented at NRCan has the capability to both extract the data and to provide GIF images.
11) Agriculture Canada: Peter Schut gave a short overview of the situation in the area of national soil data. Canadians have being collecting soil data since the early part of this century. Many surveys have been conducted, and in the 1970’s much of this data was digitized and used in GIS applications. Until 1990 things were going well and 75% of Canada had been mapped through provincial/federal cooperation. Now funding to maintain and support this data has been drooped at both federal and provincial levels. At present Ontario has three (different) copies of the database. Manitoba, Nova Scotia, and the federal government each have their own copy. Each data set has been worked on independently and so they all differ. The challenge for Peter is how to bring all these datasets together. He is proposing that under the CGDI banner, we try to develop the tools to not only bring such national data sets together but also to ensure that it can be effectively maintained. Daood Aidroos noted that insights like that from the past could help us avoid similar mistakes in the future. He asked Peter: based upon what he was hearing at this workshop was he more optimistic that we will be able to create initiatives that will be sustainable, and further, is the private sector the entity that will make it work? Peter argued that the private sector must be able to make a profit and in areas where this may not be feasible the government has a responsibility to act. In some cases the community will have to pressure the government. Neil Anderson added that in the development of infrastructure, as exemplified in the development of utilities (e.g. electricity, telephone, etc.), whether they are private or public, the community as a whole is now looking at out-sourcing to the private sector. Organizational structure must be examined since there may be better structures for the management of geospatial data and he suggested it would be of value for government agencies to look at out-sourcing the geospatial "stuff."
After lunch Pierre Lafond and the organizing committee suggested following five topics for discussion in breakout groups:
Following a short discussion
it was decided to combine topics 1) and 5) and create three breakout
groups to deal with different aspects of generalization. The context
for the discussions were the tool requirements and the groups were given
approximately one hour to work on their topic and then provide a summary
of their conclusions to the rest of the group.
1) Cartographic Generalization: Peter Paul reported for this groups which included several government agencies and one industry representative. Discussions centered upon weeding out detail for map presentations and the group considered the problem from four aspects:
2) Database Generalization: Rupert Brooks reported for the group that discussed this aspect of generalization. He noted that there were a large variety of views and many different opinions in the group. One of the first conclusions was that there are some categories of features that can be broken into super and sub-types hierarchies, and tools to handle these categories are somewhat in reach. However, there are other categories of features, that do not fit this model and will be much more difficult to handle. Other conclusions were:
Henry Kucera urged the consideration
of a common model. He felt that we must have a canonical model with
common data types and repository tools to help specify domain models.
This could lead to the realization of developing the rule-based systems
for improved generalization as suggested by the group.
3) Visualization Generalization: Pierre Lafond reported on the discussions of this group. A lot of time was spent on discussing the problems associated with "names." Doug O’Brien had described the progress of ISO TC211 with the handling of names. Some of the problems of visualization of the names of objects that were identified are:
The group felt that the handling
of text for paper maps was well understood but the visualization of
text for the "digital world" is quite different. We require rules to
be added to tools for parsing of the text. These should follow the work
being done by TC211 and become a part of CGDI. Some of the rules suggested
for visualization are:
Rick Morrison felt that text
placement was not as critical as other aspects of tools to support the
CGDI. He suggested that the systems just provide a list of the names
and then let the users massage them. Henry Kucera added that a feature-based
model developed in BC several years ago would help resolve many of the
text visualization problems that have been raised.
Peter Paul noted that from the National Atlas point of view, users are really interested in a pretty cartographic product, and similarly for the large body of non-specialist users there is still a need for good quality cartographic product.
In response to a question about how much do we know about the needs of CGDI users, Paula Rojas responded that, at present, the GeoConnections Secretariat has assumed that the stakeholders know their clientele and/or are looking at their needs. Paula also acknowledged that there is a large group of potential non-specialist users who need to be communicated with and suggested that efforts with sustainable communities may help with that task. Phyllis Charlesworth added that, except for our traditional users, we really don’t know much about future users of the CGDI, and we must avoid assuming that we do. Henry Kucera agreed, and felt that we will see a major growth in on-line use and we must put the appropriate tools in place to log information about the users and provide feedback. Doug O’Brien felt that you will never know who all your users are and using text placement as an example, where we have identified two distinct kinds of users, he argued that it was important to ensure that enough information is carried with the data so that it will satisfy all users.
4) Alignment: Phyllis Charlesworth reported that her group discussed data integration and conflation in the context of the Data Alignment Layer (CDAL). Their initial observation was that CDAL does not meet all of the perceived requirements. Tools are needed to update it. Further we must manage expectations since the group felt that if CDAL did not meet the users’ need they would not come back again even when the full CDAL DBMS is implemented. In addition the group discussed the following topics:
Peter Schut suggested that one
of the most frequent users of CDAL is for integrating satellite imagery
and he wondered if CTI had any case studies that they could make available
on the web. Christine Parent responded affirmatively. The DAL web site
is being enhanced and CTI would be willing to add such case studies
to it.
5) Vector/Imagery Extraction: Jeff Edwards used a slide presentation to report on this group (Attachment 14). Three applications were discussed and the technology issues associated with them identified. The applications were:
6) Updating of Framework Core Data: Paula Rojas reported that her group was composed of 15 people split evenly between data supplier agencies and vendors. The group began with a roundtable on the philosophy of updating national framework data. National coverage will be provided through an evolving distributed environment determined by our stakeholders and some form of distributed updating capabilities. Following a discussion of a number of issues it was found that the tools could be grouped into three categories such as:
A) Tools to compare differences in data and find changes
Paula concluded by re-iterating
the underlaying importance of an overall model for CGDI to which everything
would be mapped.
Costas Armenakis inquired if the group had considered data set versions to keep track of changes to the features, or would the features be updated every time there is be a change. Paula felt that both were feasible, and Tim Evangelatos noted the use of both schemes in the CHS where new electronic charts are issued periodically with actual feature updates being carried out between the issuance of the new versions of the charts. All of this is done using an international standard, namely S57.
Ron Graves is recommending that his agency, Statistics Canada, adopt a business model that would support "time travel", i.e., the ability to look at data over time. He acknowledged that this would be a more complex model than is in place now. Paula noted that the group discussed the importance of the data model and it will be up to the framework node of the CGDI Program Advisory Committee to lead the development/implemetation of it.
The idea of a dynamic database bothered Phyllis Charlesworth. Keeping track of people’s actions in a distributed environment, with multiple sources of data, all integrated and presented on a single window, did not make sense to her. This led to a discussion on how to track changes which, unfortunately, had to be curtailed before reaching firm conclusions due to lack of time. Phyllis ended the discussion by commenting that she would not invest resources in such complex tools.
Pierre Lafond began a summary of the workshop by asking participants where do we go now? He felt that the presentations had been appropriate and highlighted the requirements for data integration tools. He acknowledged that although a set of tools in relation to framework data integration had been achieved we did not have enough time to prioritize them and Pierre asked what should be done with the list.
Henry Kucera suggested that GeoConnections should be charged with determining the tool priorities and feed recommendations back to the community for future work. Paula added that those considering the submission of proposals to GeoInnovations 99 can use what they learned from the workshop. This knowledge will also so be used when GeoInnovations 2000 is developed. A report on the workshop will be prepared and made available on the GeoConnections web site.
Henry Kucera argued that it was important to maintain the momentum that has been generated at the workshop and suggested that there is a need for a formal way to track and measure progress. Sylvain Latour felt, however, that the actual tool development should not be done at these open workshops but should be done elsewhere.
Due to the pressure of time Pierre Lafond has to stop the discussion. He closed the workshop by thanking Sheila Rambeau and Statistics Canada for organizing and making the Simon Goldberg Conference Room available. He also thanked the Secretariat for organizing the workshop. Paula Rojas thanked Peter Paul and Rupert Brooks for proposing the workshop and developing the program.
Note: In the invitation to this workshop, participants were invited to submit a short description of their ongoing framework work and a wish-list of projects they would like to be involved in. Attachments 11, 15, 16, 17, 18, and 19 are of this type.
6- References:
TOP