Flow responses to rainfall are investigated for three small chalk springs located within about 30 km of each other. A high degree of synchronicity is shown for the spring hydrographs, which exhibit a lag of about 50 days relative to a much larger local reference catchment. Rainfall–streamflow models with six or fewer parameters, calibrated using free-to-download software, account for about 75% of the variance in daily streamflow for the reference catchment, and between about 76 and 85% for the chalk springs. Several modelling issues are discussed related to computation of the daily effective rainfall that forms the input to a Unit Hydrograph part of the model. Descriptions are given of how and why, when the recharge area used for a spring is far too small, the modelling software generates physically unrealistic effective rainfall depths much greater than the rainfall, without affecting model-fit to streamflow or the calibrated values of model parameters (except one). Reasons are suggested why it can be pragmatically acceptable for computed effective rainfall to occasionally exceed the corresponding recorded rainfall by small amounts. Wider implications of the modelling results are outlined and some suggestions for further work are made.

研究了三个彼此相距约30 km的小型白垩弹簧对降雨的流量响应。春季水文图显示出高度的同步性，相对于更大的当地参考流域而言，它们表现出约50天的滞后。使用免费下载的软件校准的具有六个或更少参数的降雨-径流模型，约占参考集水区日径流量变化的75％，而对于白垩弹簧，约为76％至85％。讨论了一些与每日有效降雨的计算有关的建模问题，这些活动构成了模型的单位水位图部分的输入。描述了如何以及为什么，当弹簧的补给面积太小时，该建模软件生成的物理上不实际的有效降雨深度要比降雨大得多，而不会影响模型对流量的拟合或模型参数的校准值（一个除外）。提出了一些理由，为什么计算出的有效降雨量偶尔超过相应的记录降雨量小量在实际上是可以接受的。概述了建模结果的广泛含义，并提出了进一步工作的一些建议。