In this study, the value of proxy data was explored for calibrating a conceptual hydrologic model for small ungauged basins, i.e. ungauged in terms of runoff. The study site was a 66 ha Austrian experimental catchment dominated by agricultural land use, the Hydrological Open Air Laboratory (HOAL). The three modules of a conceptual, lumped hydrologic model (snow, soil moisture accounting and runoff generation) were calibrated step-by-step using only proxy data, and no runoff observations. Using this stepwise approach, the relative runoff volume errors in the calibration and first and second validation periods were –0.04, 0.19 and 0.17, and the monthly Pearson correlation coefficients were 0.88, 0.71 and 0.64, respectively. By using proxy data, the simulation of state variables improved compared to model calibration in one step using only runoff data. Using snow and soil moisture information for model calibration, the runoff model performance was comparable to the scenario when the model was calibrated using only runoff data. While the runoff simulation performance using only proxy data did not considerably improve compared to a scenario when the model was calibrated on runoff data, the more accurately simulated state variables imply that the process consistency improved.

中文翻译：

---

利用小型农业流域的代理数据逐步预测径流

在这项研究中，探索了代理数据的价值，用于校准小型非流域盆地的概念性水文模型，即在径流方面是非流域的。研究地点是一个占地66公顷的奥地利试验流域，主要由农业土地利用（水文露天实验室）组成。仅使用代理数据而不使用径流观测值逐步校准了概念性集总水文模型的三个模块（雪，土壤水分核算和径流生成）。使用这种逐步方法，在校准以及第一和第二验证期间的相对径流量误差分别为–0.04、0.19和0.17，每月皮尔逊相关系数分别为0.88、0.71和0.64。通过使用代理数据，仅使用径流数据，与模型校准相比，状态变量的仿真得到了改善。使用雪和土壤水分信息进行模型校准，径流模型的性能可与仅使用径流数据进行校准的情况相媲美。尽管与仅根据径流数据校准模型的方案相比，仅使用代理数据的径流模拟性能没有显着提高，但更准确地模拟的状态变量意味着过程一致性得到了改善。