Rainfall-runoff models must be calibrated and validated before they can be used for urban stormwater management. Manual calibration is very difficult and time-consuming due to the large number of model parameters that must be estimated concurrently. Automatic calibration offers as a promising alternative, ideally supporting a user-independent and time-efficient approach to model parameters estimation. In this article, we test the use of a state-of-the-art standard package (PEST, Parameter ESTimation, http://www.pesthomepage.org/) for the automatic calibration of a rainfall-runoff EPA-SWMM (Storm Water Management Model) model developed for a small suburban catchment. Results reported in the paper demonstrate that the performance of automatically calibrated models still depends on a number of user-dependent choices (the level of catchment discretization, the selection of significant parameters, the optimization techniques adopted). Through a systematic analysis of the results, we try to identify the guidelines for the effective use of automatic calibration procedures based on modeling assumptions and target of the analysis.

中文翻译：

---

使用PEST对一个郊区小流域的EPA-SWMM模型进行自动校准：一个案例研究。

在将雨水径流模型用于城市雨水管理之前，必须对其进行校准和验证。由于必须同时估算大量模型参数，因此手动校准非常困难且耗时。自动校准可以作为有希望的替代方案，理想地支持用户独立且省时的模型参数估计方法。在本文中，我们测试了使用最先进的标准软件包（PEST，Parameter ESTimation，http：//www.pesthomepage.org/）对降雨径流EPA-SWMM（暴风雨）的自动校准为一个小型郊区集水区开发了“水管理模型”模型。本文报告的结果表明，自动校准模型的性能仍取决于许多用户相关的选择（集水区离散程度，重要参数的选择，采用的优化技术）。通过对结果的系统分析，我们尝试根据建模假设和分析目标来确定有效使用自动校准程序的准则。