(1)Measurement data. The measurement data of 
environmental disaster is the raw information that is 
received from the disaster monitoring processes, and 
various variables related to the disaster. The data 
provides the primary input for disaster management. 
It represents the current status of the disaster. 
Measurement data can be divided into two groups 
according to the general characteristics of 
management policy of environmental disaster: 
persistent and perishable. The persistent data 
consists of measurement data, whose use is long- 
term, and therefore needs to be maintained 
permanently in the database. On the other hand, 
perishable measurement data is of limited time use, 
so that its current value is valid only until the 
disaster characteristic is being monitored. 
(2)Structural data. In contrast to measurement data, 
structural data is composed of static disaster 
information. Unlike measurement data, structural 
data is valid even when the disaster does not occurs. 
Most of structural data is stored at initiation time of 
disaster management system. 
(3)Control data. Control data captures the current 
selection of control decision for disaster. The process 
for changing an existing set of control decisions is 
usually completed by the disaster managers of the 
policy hierarchy. Alternatively, the changes may be 
automatically triggered as a function of the 
information in the measurement data. In addition to 
the current settings of control decisions, the control 
database also stores a library of predefined control 
decision settings that reflect the appropriate settings 
for a variety of common environmental disaster 
patterns and distribution. 
Thus, disaster management systems based on policy 
hierarchy are responsible to monitor, interpret, and 
control environmental disaster. 
4.3 Influence management 
The role of influence management of environmental 
disaster is to manipulate the adjustable control 
decisions in real time so that the disaster influence can 
be efficiently controlled in order to reduce the disaster 
loss. Influence management from analysis for policy 
hierarchy is divided into two task as the followings. 
(1)Influence evaluation that finds how changes in 
control decisions reduce the influence of 
environmental disaster; and 
(2)Decision making on how to adjust the control 
decisions. The first task is essentially equivalent to 
find a relationship between the disaster influence 
and the control decisions, and may be required to 
estimate the disaster influence. The second one is to 
decide what control decision is selected for 
controlling the disaster influence. 
4.4 Influence evaluation 
776 
The analytical techniques, such as probability theory, 
can be used for the influence evaluation of 
environmental disaster. However, they require 
unrealistic assumptions and tend to be mathematically 
untractable as the structure of the influence measure 
becomes complex. On the other hand, discrete-event 
simulation is a viable alternative to analytical 
techniques. Its major advantage is that it can be 
modeled with much less stringent assumptions, and 
more complex performance measures can be handled 
with relative ease. However, discrete-event simulation 
usually suffers from significant computational burden 
because a single simulation run represents only one 
realization of a stochastic process. In order to obtain an 
accurate influence estimate under a given 
environmental disaster, several independent runs are 
needed, and these runs should be repeated. 
4.5 Decision making 
In the policy hierarchy, this task requires control 
decision optimization, and can be accomplished by the 
above learning and inference methods. 
5. DISASTER CONTROL 
The fundamental goal of environmental disaster 
management is to be able to control the influence of the 
disaster. The disaster control mechanisms can be 
classified along two dimensions: local versus global 
and automatic versus manual as the followings. 
5.1 Local control 
Local control mechanisms rely on local data collection 
and local decision models related to management of 
environmental disaster. The local refers to specific 
components of the disaster as opposed to the disaster as 
a whole. The advantage of local controls is that they 
incur fewer decision overhead, since decisions are made 
locally with local data. Due to this locality of the 
operation, local control processes are unaffected by 
other local control decisions. 
5.2 Global control 
Global control processes rely on all disaster data and 
global decision models related to management of 
environmental disaster. Clearly, global control 
processes are capable of optimizing performance of 
total disaster control decision. However, they are more 
vulnerable to faults related to environmental disaster 
and have greater information overhead since decisions 
require all data of disaster. 
5.3 Automatic control 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
In 
Sy 
th 
sui 
Sy: 
tas 
Sy‘ 
sh 
eas 
ag 
An 
ag) 
agi 
prc 
Ele 
an) 
des 
pra 
ear 
rel: