Rain gauges continue to be sources of rainfall data despite progress made in precipitation measurements using radar and satellite technology. There has been some work done on assessing the optimum rain gauge network density required for hydrological modelling, but without consensus. This paper contributes to the identification of the optimum rain gauge network density, using scaling laws and bias-corrected 1 km × 1 km grid radar rainfall records, covering an area of 28,371 km2 that hosts 315 rain gauges in south-east Queensland, Australia. Varying numbers of radar pixels (rain gauges) were repeatedly sampled using a unique stratified sampling technique. For each set of rainfall sampled data, a two-dimensional correlogram was developed from the normal scores obtained through quantile-quantile transformation for ordinary kriging which is a stochastic interpolation. Leave-one-out cross validation was carried out, and the simulated quantiles were evaluated using the performance statistics of root-mean-square-error and mean-absolute-bias, as well as their rates of change. A break in the scaling of the plots of these performance statistics against the number of rain gauges was used to infer the optimum rain gauge network density. The optimum rain gauge network density varied from 14 km2/gauge to 38 km2/gauge, with an average of 25 km2/gauge.

中文翻译：

---

基于雷达降雨分析的水文建模最佳雨量网密度识别

尽管在使用雷达和卫星技术进行降水测量方面取得了进展，但雨量计仍然是降雨数据的来源。在评估水文建模所需的最佳雨量计网络密度方面已经做了一些工作，但没有达成共识。本文有助于确定最佳雨量计网络密度，使用缩放定律和偏差校正的 1 公里 × 1 公里网格雷达降雨记录，覆盖面积为 28,371 平方公里，在澳大利亚昆士兰东南部拥有 315 个雨量计。使用独特的分层采样技术对不同数量的雷达像素（雨量计）进行重复采样。对于每组降雨采样数据，二维相关图是根据普通克里金法的分位数-分位数变换获得的正态分数开发的，这是一种随机插值。进行了留一法交叉验证，并使用均方根误差和均值绝对偏差的性能统计以及它们的变化率来评估模拟分位数。使用这些性能统计数据与雨量计数量的比例的中断来推断最佳雨量计网络密度。最佳雨量计网络密度从 14 km2/gauge 到 38 km2/gauge 不等，平均为 25 km2/gauge。以及它们的变化率。使用这些性能统计数据与雨量计数量的比例的中断来推断最佳雨量计网络密度。最佳雨量计网络密度从 14 km2/gauge 到 38 km2/gauge 不等，平均为 25 km2/gauge。以及它们的变化率。使用这些性能统计数据与雨量计数量的比例的中断来推断最佳雨量计网络密度。最佳雨量计网络密度从 14 km2/gauge 到 38 km2/gauge 不等，平均为 25 km2/gauge。