Abstract

The development of hydrological models that produce practically useful and physically defensible results is an ongoing challenge in hydrology. This challenge is further compounded in large, spatially variable basins with sparse data, where a detailed understanding of a basin’s hydrological response may be limited. This study presents an iterative and stepwise calibration strategy for model structure and parameters for a hydrological model of the 275,000 km² Liard River basin in northern Canada. The calibration procedure was optimized to exploit and represent available data at 29 stream gauges and included the use of multiple data sources to constrain model calibration and improve model function. A flexible modelling framework was used to allow the explicit inclusion of locally varied model structure within the calibration procedure. The final model exhibits strong performance in both calibration and validation, and represents significantly different hydrological responses in different portions of the basin well. The calibration procedure helped to identify differences in hydrological processes within the basin which have not been considered by other models of the Liard. The ability to modify model structure in order to account for different hydrological regimes in different parts of the basin is demonstrated to improve model performance locally and globally.

摘要

水文模型的发展产生了实际有用的和物理上可辩护的结果，这是水文学不断面临的挑战。在具有稀疏数据的大型，空间可变的盆地中，这一挑战更加复杂，那里对盆地水文响应的详细了解可能有限。本研究为275,000 km ²的水文模型提出了一种模型结构和参数的迭代逐步校准策略。利亚德河流域在加拿大北部。优化了校准程序，以开发和表示29个流量表上的可用数据，包括使用多个数据源来约束模型校准和改善模型功能。灵活的建模框架用于允许在校准过程中明确包含局部变化的模型结构。最终模型在校准和验证方面均表现出出色的性能，并且在盆地井的不同部分表现出明显不同的水文响应。校准程序有助于确定流域内水文过程的差异，而Liard的其他模型并未考虑这些差异。