分布式水文模型软件系统研究综述

分布式水文模型软件系统作为分布式水文模型的技术外壳,是模型应用的重要技术保障。当前分布式水文模型应用呈现出多过程综合模拟、用户群范围广和计算量大的特点,对分布式水文模型软件系统的灵活性、易用性和高效性提出了更高的要求。本文首先分析了分布式水文模型应用的主要流程,之后从应用视角对现有分布式水文模型软件系统的特点进行了归纳,主要结论为:①软件系统按照模型结构灵活性的高低分为以下3种类型:不支持子过程选择和算法设置,不支持子过程选择、但支持算法设置,同时支持子过程选择和算法设置;②根据用户操作数据预处理软件方式的不同,参数提取方式分为菜单/命令行式和向导式;③按照模型的程序实现方法分为串行和并行方式,按照模型运行环境分为本地和网络模式。现有软件系统在灵活性、易用性和高效性方面存在如下问题:一是尚未解决模型结构灵活性和对用户知识依赖性之间的矛盾;二是现有菜单/命令行式和向导式的参数提取方式步骤繁琐,难以实现参数的自动提取;三是模型大多为串行方式和本地模式,容易遇到计算瓶颈问题。最后从模块化、智能化、网络化及移动化、并行化和虚拟仿真等方面探讨了分布式水文模型软件系统的发展趋势和研究方向。

Review on distributed hydrological modelling software systems

Distributed hydrological modelling software systems are crucial because they provide technical support to the application of distributed hydrological models. Currently, applications of distributed hydrological models have exhibited new characteristics including multi-process synthesis simulation, a wide range of users, and intensive computation. Because of these new characteristics, the existing software systems are facing great challenges with respect to flexibility, usability, and efficiency. This paper reviews existing software systems for distributed hydrological models. Firstly, we analyzed the distributed hydrological modelling applications workflow including model structure determination, parameter extraction, model running, and calibration. The characteristics of existing software systems are discussed: (1) model structure flexibility of the existing software systems is divided into three types: no support of process and algorithm selection, only support of algorithm selection, and support of both process and algorithm selection; (2) parameter extraction methods of the existing software systems are divided into menu/command line and wizard method; (3) computing forms of the existing software systems are divided into parallel computing and serial computing; (4) computing modes of the existing software systems are divided into stand-alone and network mode. Secondly, we summarized the limitations of existing software systems with respect to their flexibility, usability, and efficiency. The limitations include the following: (1) contradiction between model structure flexibility and user knowledge dependence-the more flexible the model structure is, the more knowledge users need to have; (2) the existing methods of parameter extraction are too fussy for non-expert users; (3) the serial and stand-alone softwares usually encounter computing bottleneck as the appliaction scenario is data and/or computing intensive. In the last part of this paper, the emerging trends of distributed hydrological modelling software systems are discussed. These include (1) Modular modelling. The modular development ensures software reuse, but it is not enough when scale or semantic is unmatched, so the ontology knowledge needs to be considered; (2) Intelligent modelling. Using expert knowledge to realize model structure determination and parameter extraction and combining expert knowledge and optimization algorithm to parameter calibration is needed in future work; (3) On-line modelling. The development of cloud computing and network techniques makes on-line modelling practical. In addition, mobile terminals with powerful computing and storage capacity could be potential application platforms. This means that special user interface and data format are needed; (4) Parallel computing. Taking full advantage of new parallel programming standards (CUDA, OpenCL) and exploring the finer granularity parallelizability is an emerging trend. In addition, virtual simulation is another important trend.