The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
risk at a given location and a specific time are returned to the 
decision support subsystem. 
The last task performs some statistical analyses and evaluations, 
and generates some synthesized incident reports. On most 
occasions, many decisions are dependent on the statistical data 
that are drawn on the foundation of the effect areas and risk 
levels of an incident given by specific prediction and evaluation 
models. This is a case, for example, the number of the people in 
the high risk region in chemical contingencies, which is figured 
out based on demographic census data and risk levels, is critical 
for medical service agencies to arrange their rescue scale. 
Therefore, some common statistical computations are 
implemented in the decision support subsystem, and some 
ordinary integrated data involving effect region and population, 
emergency route system and evacuation scale are generated. 
The last results given by the decision support subsystem 
provide well guides for emergency managers to issue task 
commands reasonably. On the other hand, these data also are an 
important basis to perform further decision making analyses for 
different emergency response agencies. For example, medical 
service agencies can establish detailed rescue plans on the basis 
of rescue scale, incident location and emergency route system. 
3. DISCUSSION 
A great span of organizations, including governmental agencies, 
the private sector and nongovernmental organizations, are 
always involved in emergency response. A centrally organized 
command center and a spectrum of distributed operation 
agencies are widely considered as an applicable organization 
framework to effectively respond to different kinds of 
emergencies (Turoff et al., 2004; Murray, 2007), and have been 
explicitly prescribed in many emergency plans (e.g., the 
National Emergency Response Program for Public Incidents of 
China and the National Response Framework of the U.S.A.). 
The most significant function of the command center is issuing 
orders that activate or deactivate specific response tasks, like 
evacuation, and coordinating activities of numerous operation 
agencies to accomplish the tasks, which are critical to 
successful emergency response. To achieve the function, many 
requirements, involving real-time situation perception of 
incidents and integrated analyses basing on collected multi 
sources data, are usually indispensable. For satisfying these 
requirements, emergency DSS should not be constructed in a 
monolithic structure but an open architecture capable of 
integrating data from different organizations and adding needed 
resources as data and models easily. 
The DSS - Eplan described in this paper gives such a design 
with an open architecture. It is built on three systems of the 
CyberSIG platform, and under their supports data and models 
from different organizations are able to be added into the DSS 
easily and participate in decision making analyses. The 
openness of the DSS mainly covers the following three aspects. 
Open access. The client/server architecture and the loosely 
coupled structure make web-based access to the DSS available. 
The feature assures emergency managers and operators are able 
to employ the system expediently. When an incident occurs, 
first responders can input incident-related data collected on the 
spot through the web interface of the system. Similarly, 
decision makers are able to utilize the interface to perform 
model computations remotely. The results given by the system 
are organized in web pages and transferred back to decision 
makers. 
Open resources. Resources from different organizations, for 
example, data and models, can be easily enrolled into or 
removed from the DSS. The adopted web service standard that 
is designed by W3C to achieve interoperability over a network 
gives great facility for organizations to package their resources 
as services demanded by the system, and the popular SOA 
architecture implemented by the system makes the register and 
the withdraw of services convenient. By some basic 
management functions provided by the web interface, 
organizations are able to manage their resources remotely. 
Open models. The completely component-centric modeling 
approach brings powerful integrated modeling competency. 
Models developed by different organizations for specific 
emergencies can be integrated into the DSS as basic 
components. Emergency analyses and decision making models 
can be created in short time by coupling these basic or 
composed models residing in the system. When new models 
with higher precision are developed, the old ones can be 
replaced with little difficulty by uploading the new components 
to the server of Isim and making some modifications to the 
metadata of the old ones. 
The open characteristic makes the DSS suitable for response to 
different incidents. In fact, the system provides a framework for 
emergency DSS, which consists of a series of ordinary elements 
of emergency DSS, for instance, incident-centric dataset, 
prediction and analysis models, and defines an approach that 
can compose these elements into an operation process. When 
extending to a new kind of emergency, most commonly, three 
steps are needed to create an emergency decision making 
process. First, organizations register their data that are required 
by decision making into the DSS. Second, emergency managers 
and specialists provide incident prediction and analysis models. 
Lastly, emergency managers create an analysis process for this 
kind of contingencies. 
Emergency response usually involves a series of tasks that are 
selected to be executed due to the type and scale of an incident, 
for example, logistics, evacuation, and various rescues. Among 
these tasks, many decision makings are carried out, and 
accordingly some models that are able to assist these tasks are 
needed. Therefore, more models should be developed and 
integrated into the DSS besides the emergency prediction and 
analysis models. In general, at least three aspects of models are 
to be provided for constructing a perfect emergency DSS: 
incident prediction models, transportation optimization models 
for evacuation, logistics models. Incident prediction models, 
gas dispersion model for nuclear or chemical incidents (Chang 
et al., 1997; Alhajraf et al., 2005; Baklanov et al., 2006; 
Soensen et al., 2007) as an example, give the effect areas of an 
incident and evaluate its possible risk. Transportation 
optimization models yield an optimized arrangement of travel 
routes and destinations for evacuees under some specific 
objective functions and optimization formulas. Some typical 
optimization formulas include the shortest route(Campos et al., 
2000), the shortest evacuation time (Pursals and Garzón, In 
Press), the minimum cost (Yamada, 1996; Cova and Johnson, 
2003), and the maximum traffic flow(Dunn and Newton, 1992). 
Logistics models (Bakuli and Smith, 1996; Fiedrich et al., 2000; 
Yi and Ózdamar, 2007) produce an optimized resource 
allocation scheme aiming at maximizing the utilization of 
available resources. Except for the three categories, risk