Keywords
hydrological modelling, model structure, intelligent modelling, modelling knowledge
Start Date
15-9-2020 8:20 AM
End Date
15-9-2020 8:40 AM
Abstract
A key step of hydrological modelling is to determine the hydrological model structure adapting to the application context of modelling, such as the modelling purpose, the geographical characteristics of the study area, and the data availability. Such a step is non-trivial for those modellers with limited hydrological modelling knowledge, who are often performers of hydrological modelling in real applications. Although integrated hydrological modelling frameworks have been proposed as important tools of aiding modellers from the burden on hydrological modelling, the determination of hydrological model structure still depends on the modellers. This means that the modellers have to manually select proper subprocesses as well as algorithms based on their modelling knowledge, then couple the corresponding modules to be a proper hydrological model structure. Improper hydrological model structures normally means bad model performance. To lower such burden on modellers, we propose an automated method of determining hydrological model structure. In the proposed method, three types of necessary knowledge on hydrological modelling (i.e., the knowledge of selecting subprocesses, the knowledge of selecting algorithms, and the knowledge of coupling modules) are formalized in RuleML (for the first two types of knowledge), and RDF (for the third type of knowledge), respectively. Then for an application requiring hydrological modelling, the proposed method uses the formalized knowledge to automatically suggest modellers with a proper hydrological model structure. Several hypothetical experiments and a real-life application were conducted to verify the applicability of the proposed method when combining with SEIMS, a recently-proposed integrated hydrological modelling framework. The proposed method combining with integrated hydrological modelling framework can provide modellers an intelligent modelling way to relieving their modelling burden.
An Automated Method of Determining Hydrological Model Structure
A key step of hydrological modelling is to determine the hydrological model structure adapting to the application context of modelling, such as the modelling purpose, the geographical characteristics of the study area, and the data availability. Such a step is non-trivial for those modellers with limited hydrological modelling knowledge, who are often performers of hydrological modelling in real applications. Although integrated hydrological modelling frameworks have been proposed as important tools of aiding modellers from the burden on hydrological modelling, the determination of hydrological model structure still depends on the modellers. This means that the modellers have to manually select proper subprocesses as well as algorithms based on their modelling knowledge, then couple the corresponding modules to be a proper hydrological model structure. Improper hydrological model structures normally means bad model performance. To lower such burden on modellers, we propose an automated method of determining hydrological model structure. In the proposed method, three types of necessary knowledge on hydrological modelling (i.e., the knowledge of selecting subprocesses, the knowledge of selecting algorithms, and the knowledge of coupling modules) are formalized in RuleML (for the first two types of knowledge), and RDF (for the third type of knowledge), respectively. Then for an application requiring hydrological modelling, the proposed method uses the formalized knowledge to automatically suggest modellers with a proper hydrological model structure. Several hypothetical experiments and a real-life application were conducted to verify the applicability of the proposed method when combining with SEIMS, a recently-proposed integrated hydrological modelling framework. The proposed method combining with integrated hydrological modelling framework can provide modellers an intelligent modelling way to relieving their modelling burden.
Stream and Session
false