IMPROVEMENT OF QUALITY CONTROL SYSTEM IN CROATIA S. Lemajić *, N. Rožić, M. Rapaić Croatian Geodetic Institute, Savska 41/XVI, 10144 Zagreb p.p.19, Croatia - hgi@hgi.hinet.hr KEY WORDS: Geoinformation project, Data upgrade, Quality control, Quality Plan, Croatian Geodetic Institute, ISO ABSTRACT: Croatian Geodetic Institute (CGI) was founded in February 2001 and started with professional work at December of 2001. CGI is an independent institution but working closely with State Geodetic Administration (SGA). According to Law on State Survey and Real Estate Cadastre (Croatian Parliament, 1999) CGI is responsible for scientific geodetic and surveying works and for quality control of geodetic products produced by geodetic companies. At the beginning of professional work CGI started with performing of digital orthophoto control and analogue topographic maps control. At that time the CGI quality system was not in accordance with ISO standards. The Book of Ordinance (State Geodetic Administration, 2001) was the key document for ensuring quality control. Lack of software systems and hardware infrastructure, supporting data necessary for quality control, procedures and formal QC documentation were the main reason to start establishment and development of new CGI Quality System for quality control of geodetic products. Croatian-Norwegian Geoinformation Project (CRONO GIP) started in June 2002. Implementation of component c) of the Norwegian assisted Project in Croatia for enhancing the capabilities of the Croatian Government to build a National Spatial Data Infrastructure has according to Terms of reference (Statens Kartverk, 2001) three objectives: CGI shall be able to control the quality of topographic data produced by geodetic companies for SGA, CGI shall be able to perform satisfactory controlling of the quality of topographic maps and orthophotos produced by geodetic companies and CGI shall be able to prepare draft regulations and standards in line with international standards, to be applied by geodetic companies for assuring and documenting quality in the production of geographic data. Before Implementation of CRONO GIP results, CGI performed the Test Quality Control. The Dataset, which was the subject of Control, CGI have controlled according to Specification which are made for Data Upgrade. The Results with Reports of Test Control are delivered to Producers for corrections. After Corrections the Structure of Dataset was acceptable for further methods of control. - 1. INTRODUCTION At the end of 2001 CGI started with professional work in the field of control of orthophoto in the scale 1:5000 and control of topographic map in the scale 1:25000. The control of these products was not based on ISO rules. Lack of experience, equipment and specifications which were needed for control, demanded to make full control of these products. Also, before establishment of CGI there was no institutional system for making quality control but CGI started with improvement of the quality of products. CRONO GIP which started in June 2002 had a main goal to improve quality control system in CGI. Consultant Team made Inception study report (Program Management and Mapping PMM, 2002) for producing detailed project plan, for reviewing all activities defined in Terms of Reference, for verifying the project objectives and local awareness for the Project, setting guidelines and procedures for execution and building the team. The base activity of CRONO GIP which is connected with Croatian Geodetic Institute is to establish capacity in CGI to control the quality of geodetic data produced by geodetic companies for State Geodetic Administration. From the beginning, the planned activities if the Project are: Specification and verification of controlling procedures and methods for CGI according to ISO Verification of existing documentation with comparing to ISO standards Drafting new regulations and standards * Slavko Lemajić, dipl.ing.geod. slavko.lemajic@hgi.hinet.hr Croatian Geodetic Specification and procurement of equipment (hardware and software) Education in the use of new equipment (training) Practical execution of controls on products and making the reports. The realization of the Project is based on a set of documentations: Products and product specifications Quality plan Procedures and checklists Reporting. 1.1 Products Product is a set of work results, which can be used for several purposes. The Product is defined by Product Specifications. For each Product, requirements are set to the Process leading up to the final product. These processes are subject to Process Specifications. The process specification will mainly be required to process documentation. The Process Specifications are subject to more frequent changes due to development of technology and methodology, compared to the product specifications that are of a more permanent character. Product Specifications and Process Specifications are related to Quality Elements and Quality Sub elements ISO/DIS 19113 and ISO/DIN 19114 (International Organization for Standardization, 2001 and 2002). Each of these elements is described with: Definition of the quality element/sub element Institute, Savska 41/XVI, 10144 Zagreb p.p.19, Croatia – Specification Producers documentation requirements. The control of these elements is described with respect to: Sampling strategy Accept/reject rules Quality control method. Following products are defined: Aerial Photography and Ground Control Points Scanned Photo Aerial Triangulation Topographic Data Digital Terrain Model (DTM) Topographic map in scale 1:25000 (TK25) Digital orthophoto in scale 1:5000 (DOF). 1.2 Quality Plan Quality Plan defines the organization, policies and procedures applied by CGI when undertaking quality control of product deliverables. Developing and inspection plan and execution describes how individual inspections are planned and undertaken. Inspection of a product may consist of many individual inspections. Developing of inspection plan describes the steps to follow when developing an Inspection Plan. The result of this examination shall be written in the Inspection Plan for the Project. The same Inspection Plan template can be used over again for new projects of similar nature. Execution shows the steps to follow when undertaking the Inspection Work. The inspection itself must be new for each project however the procedure can be the same for similar projects. a) The development of a control plan follows the next six steps: Identification of dataset When identifying the dataset, the relevance of the data should be emphasized. Identification of data quality elements The control might involve one or more quality elements. The control plan will identify all data quality elements and data quality sub elements relevant for the product or service. Identification of data quality measure The data quality measure has to be defined. Identification of tolerances Requirements for geographic information specified as tolerance are standard parts of the specifications for the data set. Identification of data quality scope The geographic area for the control and groups of objects has to be identified. A choice has to be done between full inspection and sampling. If sampling is chosen, the size of the scope and geographic location has to be identified. Identification of control method The control method has to be defined. b) Execution of inspection contains the following: Undertake control measurements and calculate Values for data quality measures True values must be obtained, usually by control measurements. Corresponding values are selected from the dataset. The difference between the value of the dataset and the true value is the discrepancy. Based on all differences, the value for the current quality measure can be evaluated. Compare calculated values for quality measures with tolerances. Values from the control measurements shall be compared with the tolerance of the data quality measure. It has to be decided whether the quality corresponds with the product specifications or not. The result must be significantly worse than the tolerance before the dataset is rejected. Approval and handling of gross errors/discrepancies When the control is finished, it has to be decided whether the dataset shall be approved, rejected or if the control shall be expanded. Reporting Reporting can be done in two ways: Prepare a special inspection report Report as metadata. It is recommended to report by means of inspection reports. 1.3 Procedures and Checklists A Checklist is a Formal description of a task mainly built up of boxes to be filled in, ticked or signed. All Procedures and Checklists are controlled by a system of sponsor and custodian where both signatures are required before the documents are released for use. Custodians are responsible for the distribution and revision control of all documents assigned to them. Sponsors authorize the content of the document for use. As a general rule, sponsors are departmental managers. Data quality methods are divided into two main classes, direct and indirect. Direct methods determine data quality through the comparison of the data with internal and/or external reference information. Indirect methods infer or estimate data quality using information on the data such as lineage. The direct evaluation methods are further sub classified by the source of the information needed to perform the evaluation. a) Direct Evaluation Methods Types of Direct Evaluation Methods: internal and external. Means of Accomplishing Direct Evaluation: for both external and internal evaluation methods, there are two considerations, automated or non-automated and full inspection or sampling. Full inspection Sampling b) Indirect Evaluation Method The indirect evaluation method is a method of evaluating the quality of a dataset based on external knowledge. This external knowledge may include, but is not limited to, data quality overview elements and other quality reports on the dataset or data used to produce the dataset. 1.4. Reporting Reporting of Data Quality Evaluation Information is possible as: Reporting in Metadata Reporting in a Quality Evaluation Report Reporting Aggregated Data Quality Result. It is recommended to report by means of Inspection Reports. 2. TEST QUALITY CONTROL OF TOPOGRAPHIC DATA Before the end of Project CGI has preformed test quality control of topographic data. Preliminary tasks that are done can be described as: Verification of existing data according to CROTIS (State Geodetic Administration, 2002) Data upgrade specification (PMM, 2003) Quality control of upgraded data. 2.1 Verification of existing data 2.3 Data upgrade quality control For verification of existing data six Croatian private companies delivered a part of data from dataset that were the object for comparison with Croatian Topographic Information System (CROTIS) data model. Another objective was how to bring the existing data up to new Specification. The task was to determinate differences between datasets and existing specification. It resulted in the the report about the existing data based on examples of six companies. Findings were divided on: Global remarks to existing specification (or incomplete specification) Global remarks to all companies Specific remarks for a single company. 2.2 Data upgrade specification Data upgrade specification describes: Content of specification Terminology used in specification Deliverables Specification for geometry and topology Specification for topographic data DTM specification. According to this Specification companies have upgraded the Data. The main activity in pre qualification process was changing attributes and topological cleaning the data. Such prepared data will be ready for input into database. The integral parts of specifications are: Object selection criteria Mapping catalogue Library of symbols Quality elements. In the specification, there is also the terminology described that is partly based on Simple Feature Specification by Open GIS Consortium. According to specification, the contractors should deliver final production report and the dataset. Final production report contains Production metadata and other information, Descriptions of topology processing procedures and Descriptions of quality control procedures for important processes. Dataset to be delivered is Topographic vector data in 2D and Digital terrain model (DTM) vector data in 3D. All data should be delivered in „dgn” format (Microstation). Coordinate system of dataset, file names ad media are described too. Specification for geometry and topology describes that delivered data should contain only line string and point geometries. Data topology describes that polygons are not allowed but polygons should be automatically generated from the data. The same rule with requirements for topology is described. Specification of topographic data describes groups of feature classes and representation in „dgn” files. Representation of feature classes is described in „Mapping catalogue” and selection criteria in „Selection criteria” for data capture. Distribution of feature classes is divided by groups, levels and files. The specification also describes which feature classes and in what way they should be represented. There are following feature classes: Toponyms, Buildings, Utility, Transportation and elevated features, Vegetation, Hydrographic and digitized feature. DTM Data Specification contains Digital Terrain model and DTM data processing procedures. DTM features are spot heights, mass points, break lines, form lines and areas of unreliable DTM. In DTM data processing procedures are data preparation, data vectorization, data editing, data specification, quality control and delivery formats. State Geodetic Administration (SGA) contracted data upgrade for six map sheets of Topographic map in scale of 1:25000 (TK25) with six geodetic companies. They upgraded existing dataset according to Data Upgrade Specification and Mapping Catalogue. Before starting the job consultant team of CRONO GIP made training for geodetic companies. Companies delivered upgraded data with technical reports. CGI, in coordination with CRONO GIP Team, made a strategy for quality control. The Strategy contained: Quality Plan Quality methods with realization Results of control Reports. 2.3.1 Quality Plan Quality Plan defines requirements for quality control of upgraded data and Quality Management System. Quality system in this Quality Plan covered Product of Topographic Data. Quality Plan covered following steps: Control of deliverables Control according sources Control of structure of data Control of topology Control of DTM dataset. Geodetic companies delivered mainly the following files: - YYYY-MM-DD-<ContractorID>-<MapNumber>-BUILDINGS.dgn - YYYY-MM-DD-<ContractorID>-<MapNumber>-UTILITY.dgn - YYYY-MM-DD-<ContractorID>-<MapNumber>LANDCOVER.dgn - YYYY-MM-DD-<ContractorID>-<MapNumber>-DIGITISED.dgn - YYYY-MM-DD-<ContractorID>-<MapNumber>-TOPONYMS - YYYY-MM-DD-<ContractorID>-DTM.dgn - YYYY-MM-DD-<ContractorID>-DEM-GRID.dgn - YYYY-MM-DD-<ContractorID>-REPORT.pdf, where YYYY-MM-DD is date of deliverry, ContractorID represent name of company, MapNumber represent map number and at the end content of files with file format. Examples of topographic data and DTM data are shown on Figure 1 and Figure 2: Figure 1: Example of topographic data Figure 4: Unprecise creating of road centerlines Visual control of DTM was done with the help of derived shade (tiff format) and DEM (dgn format) as shown on Figure 5. Some questionable parts on raster file are checked. Figure 2: Example of DTM data Sources which are used for control were Data inventory report, Data upgrade specification and Mapping catalogue. For the control of toponyms there were aslo the old maps of TK25 used. Data structure control included was done acoording to mapping catalogue which describes level, color, line style and weight for linear objects, cell name for point elements and representation point (text) for areal objects. Control of topology included checking closing areas, duplicate lines and boundaries, open boundaries, number of representation points in areas, missing representation points in areas, detecting unused boundaries, etc. Control of DTM included also checking of structure elements in files and deriving DEM from delivered elements and detecting of possible errors. 2.3.2 Quality methods There are two methods for quality control: Visual control with Microstation and SCOP Automatic control of data structure and topology (using FME). Microstation was used for visual control of data. There was a system to control elements, which were wrongly represented before data upgrade process. Mostly, bridges were wrongly represented, roads were without center lines, waters were not represented under bridges (Figure 3) and culverts, creating centerlines were also not enough precise (Figure 4). Figure 5: Visual control of DTM Automated quality control was done with FME (Feature Manipulation Engine) software. Partly based on FME training and partly with help of CRONO GIP consultants in CGI the FME factory files were used for checking the structure of datasets and for checking topology. The inputs for control of dataset structure were all legal levels, colors, line styles and weights for linear elements, for point elements input there was a list of symbol (crotis.cel file) and for area elements a list of codes (representation points). FME detected all open boundaries, unused boundaries (which are not necessary for creating area topology), missing classifications points (text codes), duplicate classifications points, overlapping lines and boundaries, self intersected lines and boundaries. 2.3.3 Results of control Figure 3: Water line not represented under bridges After the control, the results were divided to: Comments connected to undefined specification Comments connected to misinterpretation of specification Comments to topological processing. In the process of quality control CGI detected some cases which are not enough described or are not described in Specification. Typical example of that case is joining of road center lines. There was no rule for joining narrow paths with road center lines which are very important for creating road network topology (Figure 6). 2.3.4 Reports For each company CGI wrote report, which includes: Introduction Authority for control List of Sources Content of control Findings General and single comments Comments connected to topologic processing Conclusion. 2. CONCLUSION Figure 6: Narrow path not connected to road centerline Some producers interpreted object in their own way. Typical example is representing of culverts (Figure 7) Croatian Geodetic Institute has done Data upgrade Quality Control of structure and topology processing of topographic data in short period of time. Control of accuracy and correctness of representation and classification have not been included in this process. Complete quality system should be implemented at the end of the project. Even though complete quality system was not implemented, CGI has done quality control in part and results of control were very useful. Before completing the specification, some elements will be changed and enhanced and the difference between producers reduced to accepted level. There is also a team in CGI for all tasks and challenges. This test quality control confirmed that contention. REFERENCES Croatian Parliament, 1999, Law of state survey and real estate cadastre, Narodne novine br. 128/99, Zagreb. Figure 7: Different ways of representing culverts Most comments connected to topology processing were in using wrong attributes for linear elements (wrong color or weight for lines and boundaries). Distribution for errors and companies is shown in Table 1. Error type Invalid element type Invalid line attribute Invalid boundary attribute Unused boundaries Missing classification point (textual code) Unknown classification point Excess classification points Self intersecting lines Self intersecting boundaries Invalid cells Company C1 C2 ● C3 ● ● C4 ● ● C5 ● ● C6 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Statens Kartverk, 2001. Terms of reference for the implementation of component b) and c) of the Norwegian assisted Project in Croatia for enhancing the capabilities of the Croatian Government to build a National Spatial Data Infrastructure. Appendix A to the Contract, Oslo, Norway. Program Management and Mapping, 2002. CRONO GIP Croatian Norwegian GeoInformation Project - Inception study report, Zagreb, Croatia. Program Management and Mapping, 2002. CRONO GIP Croatian Norwegian GeoInformation Project – Data Inventory report, Zagreb, Croatia. Program Management and Mapping (PMM), 2003. CRONO GIP Croatian Norwegian GeoInformation Project – Data upgrade specification, Zagreb, Croatia. International Organization for Standardization, 2001, ISO/DIS 19113 Geographic information – Quality principles. International Organization for Standardization, 2002, ISO/DIS 19114 Geographic information – Quality evaluation procedures. State Geodetic Administration, 2001, Book of Ordinances, Zagreb, Narodne novine br. 55/2001, Zagreb. Table 1: Distribution error types for each company State Geodetic Administration, 2002, Croatian Topographic Information System CROTIS, Zagreb.