The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
1125 
of the latest data and hence more dependable for decision 
makers. 2.5 Decision support subsystem 
2.4 Integrated Simulation Modeling System 
Isim is an integrated modeling system developed for satisfying 
a spectrum of applications based on digital city. It adopts a 
component-based modeling approach, and at the same time 
designs an extensible plug-in framework for giving support to 
different modeling paradigms (see Figure 4). In Isim, models 
encapsulated as components are taken as “building blocks”. By 
coupling these blocks, coarse-grain models are created and 
corresponding components for them are generated 
automatically. Three aspects of compatibility are checked when 
a coupling process is required, which are temporal 
compatibility, spatial compatibility and semantic compatibility, 
for the purpose of assuring the technical correctness and the 
actual significance of created models. In the system, models are 
managed on the basis of their metadata, and are able to be 
reused in different grains. Each model accepts two kinds of 
inputs: parameter and input variable. The first refers to the 
parameters of a model that need be specified only once during 
model execution. While the latter need to be set in each 
simulation step, and therefore are used to exchange data with 
other models. 
The system is constructed in client/server architecture and tries 
to provide a maximum of simplicity for users. The client is 
composed of a complete graphical user interface that enables 
modelers to create a model by some operations of dragging and 
dropping components represented as icons and to simulate a 
model. The server consists of a model base, a metadata base, a 
plug-in library, and a series of function modules for sustaining 
user access, model management and simulation control. Using 
the client, modelers can drive the server remotely to generate or 
execute a generic model, and explore data transferred from the 
server in multiple manners supported by the client, such as 
image, table, statistical figure and report. 
The decision support subsystem acts as the control center of the 
DSS. It drives the other three systems to produce necessary data, 
and integrates them to form a general report for emergency 
decision makers, according to prepared emergency processes. 
An emergency process prescribes some basic elements and 
regulates an operation process for a kind of incidents. Generally, 
it answers a series of questions as follows: 
What data are needed to support decision makings of these 
incidents? 
What models are required to support decision makings of 
these incidents? 
What data are needed by each model? 
How are these data obtained? 
How is this process executed? 
Where are the last results sent? 
When an incident happens, emergency managers can activate 
the decision support system by inputting incident-related data 
and call the decision support subsystem to execute a specific 
emergency process by the web interface. After that, three tasks 
will be executed sequentially to respond the emergency, which 
are data collection, model invocation and results integration 
(see Figure 5). 
£ 
Emergency 
Manager 
\ > < Colled Data > CZ 
Invoke Model > CIZ 
Integrate 
: Results 
Spatial Database 
Management 
System 
CyberSIG Service 
Bus 
Integrated 
Simulation 
Modeling System 
Figure 5. Decision support process overview 
Figure 4. Isim overview 
In the DSS, all emergency models are created and managed in 
Isim. At present, some models on prediction and evaluation of 
chemical incidents have been provided, and more emergency 
models are planned to develop. 
On most occasions, CSB is called by the DSS to obtain 
necessary data as the input of specific models for an incident, 
which assures more accurate prediction of the evolution and 
evaluation of the effects of an incident. The associations 
between incidents and emergency models, as well as emergency 
models and data, are predefined and recorded in the decision 
■ support subsystem. 
The first task collects necessary data for decision making 
analyses and incident prediction. Generally, in terms of source 
and content, these data can be categorized into three groups: 
incident-related, surroundings-relevant, and model-suited. The 
incident-related data are collected on the incident spot and 
inputted by emergency responders, which commonly describe 
the incident location, type, scale, risk level, primary effect 
region, and some other incident specific characters. The 
surroundings-relevant data are acquired from SDMS, and have 
same content as that of the fundamental urban spatial dataset 
described in section 2.1, but limited boundary. The data 
boundary is usually ascertained by the inputted incident type 
and effect region. The data that is necessary to model 
computations, except for the above two groups, constitute the 
model-suited data, for example, meteorology data, demographic 
census data. These data are obtained from other organizations 
under the support of CSB. The three groups of data are gathered 
and shape an incident-centric dataset. 
The second task invokes Isim to run specific models that are 
defined in the emergency process to predict the evolution of an 
incident and evaluate its effects. The values of the parameters 
and the input variables of the models are obtained from the 
dataset generated by the first task. The results calculated that 
generally include the effect areas of an incident and potential