Keywords
high performance computing, distributed computing, hydrological modelling
Start Date
1-7-2006 12:00 AM
Abstract
Our paper describes experiential work in developing distributed versions of several hydrological and non-point source pollution models. Work in parallelization of the calibration of environmental models and in the population of belief network representations of these models has required us to developing a client-server framework. Newer models often have large computational needs in calibration. Similarly, replacing computational models with belief network representations suggests the utility of parallel computation for the Monte Carlo techniques to generate conditional distributions for responses of outputs to the model parameters. We are required to maintain the integrity of the original model and to decouple interfaces from their computational engines. This has resulted in the development of sophisticated “wrappers”, and middleware components for communications and task management.
Experiences in Sharing Environmental Models in Distributed Environments
Our paper describes experiential work in developing distributed versions of several hydrological and non-point source pollution models. Work in parallelization of the calibration of environmental models and in the population of belief network representations of these models has required us to developing a client-server framework. Newer models often have large computational needs in calibration. Similarly, replacing computational models with belief network representations suggests the utility of parallel computation for the Monte Carlo techniques to generate conditional distributions for responses of outputs to the model parameters. We are required to maintain the integrity of the original model and to decouple interfaces from their computational engines. This has resulted in the development of sophisticated “wrappers”, and middleware components for communications and task management.
