70 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 6. Bandung-lndonesia 1999 
SPATIAL AND TEMPORAL DATA HANDLING FOR REMOTE SENSING DATA 
R. Venantius Hari Ginardi 
Presented on ISPRS Mid-term Technical Commision VI Symposium, Bandung, 15-17 April 1999 
Lecturer on ITS Surabaya (Sepuluh Nopember Institute of Technology - Surabaya), postgraduate student on Msc in Information 
Tecnology for Natural Resources Management, SEAMEO-BIOTROP, Bogor-Indonesia 
ABSTRACT 
Processing multi-satellite imagery with different times of acquisition in image processing software is based on manual image 
selection by the user. Its efficiency depends on how accurate the user notes the file content, image acquisition dates and file location. 
Selecting image files based on spatial and temporal criteria is a process that helps the user to retrieve image file(s) more effectively. 
This paper proposes a Temporal Data Manager Development for Remote Sensed data that enables image data query, based on spatial 
and temporal information. This model can support development in the real time dynamic modelling using Satellite Imagery. 
1. INTRODUCTION 
Early use of satellite imagery was mainly based on image 
interpretation. Image processing tools , electromagnetics theory 
and coordinate system were combined for this purpose. The 
output was object figure or analysis such as land cover 
classification, NDVI, clay ratio, shape detection, topographic 
feature, etc. 
other programming for the advanced user. In some remote 
sensing software, this feature is called batch script or user script. 
The scope of programming is dependent on supported language 
and predefined object class. Much more customization can be 
made if remote Sensing software supports C language or .lava 
programming. 
2. SPATIAL AND TEMPORAL DATA HANDLING 
With a wider choice of sensor type and satellite source, more 
variation of spatial and spectral satellite imagery becomes 
available. It is also feasible to collect multisensor data and 
multitemporal data for certain area. With multitemporal data for 
specific area, more analysis and modelling based on change 
detection can be performed, using the advantages of currently 
cheaper and more detaileds satellite imagery. 
Current image processing sofware use manual data retrieval. A 
user should use his/her interpretaion of metadata to select which 
image file to retrieve. The management of metadata is in the 
form of text files that are organised depending on user. This 
paper proposes to develop a Temporal data manager approach 
for Remote Sensed Data, that enables features for data query 
based on spatial and temporal information. 
Basic information stored in remotely sensed data is spectral and 
spatial information. Managing temporal information is an 
additional possibilitiy for current remote sensing applications. 
Although a time stamp for acquisition time information is 
currently available in Remote Sensed Data, this information is 
not a part of data processing. The image file format is highly 
dependent on remote sensing application software. By 
managing temporal information, some operation can be 
performed for sequence of satellite imagery like : clouds 
removal, mosaic, and other image correction using other image 
information. 
Spatial information of image consist of information about 'what' 
feature exist in image file and 'where' is object position relative 
to other object. Spatial information in vector representation is 
known as topology. Temporal dimension can be handled in a I 
time type information format. There are two common time type 1 
information format : event time and transaction time. Temporal 1 
information systems can be sub-divided into four categories i 
based on those two type (Worboys, 1994) I 
s 
No support for 
transaction time 
Support for 
transaction time 
No support for 
event time 
Static 
Rollback 
Support for event 
time 
Historic 
Bitemporal 
Table 1. Temporal Dimension supported by system types 
This paper focuses only on historic type of temporal dimension 
to be applied in Remote Sensed Data since image file from 
satellite imagery can be considered as an entity with a single \ 
time stamp. Coversely transaction time is usually used in 
motion multimedia video that has duration time component. 
Storing multi-temporal data for one spatial entity, for example 
NDVI data from NOAA satellite for West Java on 7 consecutive 
days, has 2 possibilities : 
The Development of Temporal Database in GIS (called TGIS) 
is more applicable since GIS and Database Systems are closely 
related. Adapting a Temporal Database to Remote Sensing 
application is very difficult because of its different file structure. 
Remote Sensed data is similar to multimedia data. Since there is 
existing research to model and retrieve Multimedia data using 
temporal information, this approach can apply a temporal data 
manager into current remote rensing application. 
Some remote sensing software that is widely used supports user 
defined application development that enables customization and 
a. Store all 7 layers of temporal information in one dataset. 
This approach is widely used because it is easy to perform 
arithmetic modelling, but the independence of each 
temporal information layer is lost because they are 
bounded together. 
b. Store each image layer data of temporal information 
independently, and manage metadata for each dataset. This 
approach has a benefit from its independence, but requires 
manual user retrieval and metadata interpretation. The 
temporal data handling used in this paper is based on this 
approach.