FRAME REPRESENTATION OF
ECOLOGICAL MODELS IN FORESTRY
PLANNING
Tao Chen
Jian Kang Wu
Mikio Takagi
Takagi Lab, Dept. of Electronics Engineering
Institute of Industrial Science, University of Tokyo
7-22-1 Roppongi, Minato-ku, Tokyo 106, Japan
Abstract
How to build an image interpretation expert system and other relevant expert systems
within a knowledge-based pictorial information system is an ad hoc theme in geographical
information system(GIS) field. It plays an important role in building a synthetic information
system and in enhancing the GIS system performance. In this article, we present a forestry
regional resource planning sub-system, which uses frames to represent the ecological models
and other relevant expert knowledge. With the computer aided planning module, we can
give in real time the geographic data satisfactory interpretations, perform in parallel certain
kind of fuzzy reasoning to obtain the satisfied decision. The results can be used to update
the data stored in the information system.
KEY WORDS: GIS, Query Language, Knowledge Representation, Expert System
1 INTRODUCTION
An important part of geographical information system is
computer-aided regional resource and environment manage-
ment, such as city planning, agricultural planning, forestry
management, and etc.. These often require system perfor-
mance such as efficient storage, flexible manipulation, and
intellectual usage of large amount of and variety of spatial
data. However, most existing practical systems are designed
from view point of geo-science. These systems emphasize
cartographics rather than system flexibility and intelligence.
Therefore these systems often demand well-trained opera-
tors. The exploration of the system performance depends
heavily on the intelligence of the operators. Because of the
system architecture, these systems are not able to support
computer aided planning based om some application mod-
els. Researchers from the field of computer science have pro-
posed some very promising ideas and designs[1,2]. Here, we
present a computer aided forest planning subsystem which
employes frame representation of ecological knowledge and
other expert knowledges. Its reasoning process consists of a
forward reasoning of Bayes classifying of Landsat imagery,
à backward reasoning schema using frame representation of
knowledges and a reasoning module using spatial consistency
model. The system models forestry ecological information,
including effects of environment, human management activ-
ities (cutting and planting), plan of forest authorities. The
model is then represented as frames. The forestry inven-
tory is then carried out by reasoning using Landsat data,
geographical data from database and knowledge stored in
frames. In the following sections, we first give a brief intro-
duction to the background environment (knowledge based
geographical information system KGIS), and then describe
frame representation of ecological models in details.
== TUE
[ DBMS, incuding DDL,DM. H Self-Descriptive Files
|
kK
a
reason knowledge
control base
Image Spectrum Knowledge :
Processing an Leaning
Library Ecological Models
Figure 1: Block-diagram of KGIS
2 SYSTEM BACKGROUND
The system block diagram of KGIS is depicted in Fig.1. The
central part is a spatial database management system. It
manages relational base, frame base and physical database.
The database description is stored in database dictionary.
Procedures used to manipulate either logical data or physi-
cal data are stored in procedure library. It consists of rela-
tional algebra operations and image algebra operations (bi-
nary image logic operations, geometric operations, opera-
tions for spatial relationship verifications). The input mod-
816
bpd ONY pp SY € 5 oc
an. QUES Hd
I