A Study of a Framework of Integration of Heterogeneous Spatio-Temporal Data
Yoshihide Sekimoto and Ryosuke Shbasaki
Center for Spatial Information Science, Institute of Industrial Science
University of Tokyo
7-22-1, Roppongi, Minato-ku, Tokyo 106-8558, Japan
Tel: (81) -3-3402-6231 ex: (2562) ,Fax : (81)-3-3408-8268
E-mail : sekimoto@skl.iis.u-tokyo.ac.jp
Abstract
Recently many kinds of spatio-temporal data are repeatedly acquired for almost the same objects because of the recent rapid progress of data acquisition technology . we can raise various maps and remote sensing data of the same area as examples . These are heterogeneous spatio-temporal data. Some specific data integration problem has several solutions such as map overlay and fusion of the remote sensing data through all of them are note very satisfactory. However , we need to develop a generalized method of data integration based on each solution. This is because to will be expected to deal with an enormous and heterogeneous data in order t meet much higher demands such as monitoring of the complicated dynamics of a city. This study proposes a theoretical framework to integrate data of the object measured by different conditions, together with information to be provided, such as metadata in applying integration method.
1. Introduction
Recently many kinds of spatio-temporal data are repeatedly acquired for almost the same objects because of the recent t rapid progress of data acquisition technology. We cab raise an example of NOAA AVHRR data an Landsat TM data as heterogeneous -----different resolution , band and observation time--- remote sensing data of the same area. Spatial aspects of the heterogeneity are some gaps in the map boundaries, differences in resolution and geocoding gaps in remote sensing data and so forth . temporal components examples are the data observation time and acquisition time and acquisition time. Users , surrounded by may kinds of spatio-temporal data , according to their own demands, cannot help selecting either of a may data or just overlaying many data such as "rubber sheet method ". In the former case using single data it may not satisfy the various demands of users, while in the later case simple overlay of heterogeneous data may create meaningless information. This leads t the problem that we can not make effective use of many kinds of data . in this situation NSDI (National Spatial Data Infrastructure ) plan, which aims at the common sharing of information, is now under ways a governmental policy and it will creator a demand to integrate a wider variety of heterogeneous spatio-temporal data. Although specific cases of data integration may have several solutions like map overlay we need to develop a more generalized method of data integration . this is because it will be expected to deal with an enormous and heterogeneous data in order to meet much higher demands such as monitoring of the complicated aspect of a dynamics city. The objective of this study is therefore to show a theoretical , methodological framework of integrating data of an object from multi sources to estimate the status of the object, together with essential information like metadata provided in applying this methodology. Further more we show that were place the existing ad hoc methods in the framework and a proposed generalized method can be applied to more complicated case.
2. An example of map overlay as a process of data integration
In this chapter we briefly describe flow of the data integration through an example of map overlay to identify importance components in the data integration process. We assume that a use needs to estimate more accurate Cho-Co-Moku (CCM: lowest level of administrative zones ) boundary in Sinjku Ward by overlaying two boundary maps, Spatial Data Infrastructure (Digital Map 2500) by GSI (Geographical Survey Institute )a n Census Mapping System (CMS) data by Statistics Bureau of Japan.
2.1 Retrieval ad acquisition of observational data suitable for demand of users
User's need is to know more accurate boundary of CCM so as to get some suitable data form database, and SDI and CMS are chosen as suitable "observational " data by referring each metadata. But some gaps may be seen between boundary line features in these maps. W have to determine which lines should be merged considering the why the errors and gaps occur. Figure 1 shows that the sources of the gaps are generally divided into three steps .
Fig.1 Sources of errors and gaps
