Technology
National Geo-spatial Data Infrastructure: Theories and Technologies
Data is acquired by different agencies at different times, in different formats as per their immediate applications requirements. All this data exists in unrelated archives. There is considerable duplication of data as inter agency co-ordination is a rare characteristic. There are no means of ascertaining data holdings. The data characteristics and formats are not known explicitly. Relating diverse data sets is difficult due to lack of standardisation of data formats.
Standards
The answer to the above problems lies in evolving a set of standards that need to be followed by all data generators. Three major groups of standards are required. They relate to formats, contents and exchange. The format is implicit in a hardcopy such as a map or a table but in digital form, which is required in a SDI, the format has to be explicitly spelt out in the form of headers and lookup tables. The contents have to be standardised as per the subject. For example, a Level 1 Land Cover map should mean the same thing in terms of content to all users of the map. If we need to move data from one system to another, or from one software to another, or merge two independent data sets, we must have data exchange formats which the source and recipient entities understand.
The Survey of India toposheets provides a very good national standard for topographic maps. However, thematic maps have standards defined by the source organisations. There may be cases where no standards exist. In general, geoid, projection, scale, and minimum mapping unit, mapping error and registration accuracy specify spatial data objects. Hierarchy of classes specifies themes. Standards are also laid down for the source data, update cycles and for verification and validation. Conversion of analogue data to digital also involves errors and standards are laid down to ensure that these are held within tolerable limits. This is of importance when legacy data has to be converted from paper to digital form.
The issues of standardisation can be very contentious because scientists tend to draw up their own standards for specific tasks. Bringing all these ideas to a commonly agreed standard requires a great deal of interaction and discussion to iron out the issues. Standardisation may require some amount of legacy data restructuring and this must be done carefully keeping changes to a minimum.
Over and above the data standards it is also necessary to standardise data base structures, field names, file naming, conventions, etc. Many of these items can become technology specific. It is necessary to avoid technology specific standards because it is expected that the technologies in use will be heterogeneous. Further, legacy systems have to be protected hence technology specific standards must be avoided. Standardisation of hardware and software must be avoided. Rather stress should be laid on interoperable systems that provide continuity of legacy systems.
There are several standardisation efforts under way. At the international level there is the effort being pursued by the International Standards Organisation through its ISO/TC211 initiative. Nationally there are several efforts. The US has its Spatial Data Transfer Standards, SDTS. The Canadians are examining the Spatial Archive and Interchange Format, SAIF. In India, we have two standards: the National Resources Information System, NRIS, Standards and the Survey of India Digital Vector Data Standard, DVD. The former is a content-based specification while the latter addresses the data exchange issues. An attempt has been made to merge the two to form a single content cum transfer specification. These are the first steps in the direction of evolving a national standard for India.
Standardisation can become an unmanageable behemoth. Consequently, most efforts remain at the research level. There is a need to temper the rigour of standardisation with the flexibility of implementation. An interesting approach in this regard is the initiative of the Open Geodata Inter-operability Specification, OGIS of the OGIS Consortium. This specification seeks to standardise processes rather than contents or formats. This allows systems, software and applications adhering to the OGIS to easily exchange data. In the final analysis, we will require a mix of standardised formats and content as well as processes to realise the SDI.
The Spatial Framework
The spatial framework is very crucial and it is important that this be decided upon in the beginning. In its simplest form, it is a frame of latitudes and longitudes with intermediate tic marks aimed at providing an invariant reference for all spatial data sets. However, most users need some basic references. Thus, it also can include ortho-rectified imagery, elevation, bathymetry, geodetic control, transportation, administrative boundaries, etc. Data is generated by several agencies but need to be used together. Hence, all data has to be registered to this framework so that they can be related to each other. The framework must meet the mapping accuracy desired by the applications. The choice of the geoid and the projection systems has a bearing on the accuracy. Further, the accuracy is also a function of the scale of mapping. In India, we have the advantage of having an excellent cartographic database in the Survey of India topographic sheets. These are based on the Everest spheroid and Polyconic and Lambert Conformal Conical projections. The framework of this system is ideally suited for providing the structure for a spatial database. However, as this framework is shared with the Armed Forces there are severe restrictions on the use of the framework. Consequently, the SDI as conceived above will become ineffective as a facility for the open community.
An alternate framework is under discussion and should be operationalised in a short time. This framework will serve all databases from 1:1million to 1:25000. This can comfortably cater to the needs of the administration for planning and monitoring purposes. However, this will not be sufficient for project implementation. This level requires databases at 1:10,000 or larger scales. These maps are usually cadastral maps using local projections. Interlinkage of these maps with the projected framework is an involved task and no standardised procedures exist. With the increased use or Remote Sensing for thematic mapping, such inter-linkages are essential and hence this is an area in need of urgent attention.