Environmental Engineering Reference
In-Depth Information
The Model Component Explorer
The Model Component Explorer (MCE) is a Windows application developed
by CRA for inspecting model components to reveal interfaces, domain classes
and VarInfo values for each variable, simple and composite strategies and their
parameters, inputs, outputs, and associated strategies. Given the component
architecture implemented, the tool can be used to show the input and output
variables within the domain classes of any model. The tool also allows XML files
to be saved with schemas for each domain class and strategy, allowing them to be
uploaded into a shared ontology (see paragraph Web resources).
APES Tools for Integration in Broader Modelling Systems
Given its component-based structure, APES can be run not only from a user inter-
face, but also using a command line procedure. This allows the model to also be
called from applications developed using languages which have no binary compat-
ibility with .NET (e.g. Java) provided these applications are on a machine running
Windows. APES is used in the software system SeamFrame (Wien et al., this volume)
as an external component. As SeamFrame is implemented in Java, it requires APES
to run as executable files. Support for integration can also be provided if the appli-
cation which acts like a client is a web application, provided that such an application
exposes web services and includes rich clients. In this case, some of the applications
implementing a user interface can also be used. APES includes tools for integration.
The Parameter Estimator
Parameter estimation is a major aspect of crop modelling. Together with the functional
forms of the equations, it is a major determinant of prediction quality. Parameter
estimation is a difficult and time-consuming exercise and requires expertise not
always available in a modelling project. The purpose of the Parameter Estimator
(PE), developed by PRI and INRA, is to provide software to automate model
parameter estimation. The Parameter Estimator consists of functions in the R
statistical computing the language (R Development Core Team
2007)
. Models such
as APES are coupled to the PE through specific functions. APES is written in the
C# computer language and is, for the purposes of the PE, exposed as a Microsoft
COM object. This makes it possible to use the R-to-COM bridge (Baier
2007)
to
execute an APES simulation run from within R statements. The R software requires
the following information: the observed data, the name of the model to be run, the
paths to the input files for the contexts of the data (for example climate, soil and
management files for each context), the list of model parameters and indicators as
to which are to be estimated and finally information related to the correspondence
between the data and the model output. The R routine sets the parameter values
