The motivation for improving gridded precipitation data lies in weather now-casting and flood forecasting. Therefore, over the past decade, Commercial Microwave Link (CML) attenuation data have been used to determine rain rates between microwave antennas, and to produce more accurate countrywide precipitation grids. CML networks offer a unique advantage for precipitation measurements due to their high density. However, these data experience uncertainty from several sources as reported in earlier research. This current work determines the reliability of rainfall measurements for each link by comparing CML-derived rain rates to adjusted weather radar rainfall at the link location, over three months. Dynamic Time Warping (DTW) is applied to the pair of CML/radar time-series data in two study areas, Israel and Netherlands. Based on the DTW amplitude and temporal distance, unreliable links are identified and flagged, and interpolated gridded precipitation data are derived in each country after filtering out those unreliable links. Correlations between CML-derived grids and rain observations from an independent set of gauges, tested over several rain events in both study areas, are higher for the reliable subset of CML than the full set. For certain storm events, the Kendall rank correlation for the set of reliable CML is almost double that of the complete set, demonstrating that improved gridded precipitation data can be obtained by removing unreliable links.

中文翻译：

---

从微波链路衰减导出的改进网格降水数据

改进网格降水数据的动机在于天气预报和洪水预报。因此，在过去十年中，商业微波链路 (CML) 衰减数据已被用于确定微波天线之间的降雨率，并生成更准确的全国降水网格。CML 网络由于​​其高密度为降水测量提供了独特的优势。然而，正如早期研究报告的那样，这些数据存在来自多个来源的不确定性。目前的工作通过将 CML 得出的降雨率与链接位置调整后的气象雷达降雨量进行比较来确定每个链接的降雨量测量的可靠性，超过三个月。动态时间扭曲 (DTW) 应用于以色列和荷兰两个研究区域的 CML/雷达时间序列数据对。根据 DTW 幅度和时间距离，识别和标记不可靠的链接，并在过滤掉这些不可靠的链接后得到每个国家的插值网格降水数据。CML 衍生的网格和来自一组独立测量仪的降雨观测之间的相关性，在两个研究区域的几次降雨事件中进行了测试，对于 CML 的可靠子集来说，比完整集更高。对于某些风暴事件，可靠 CML 集的 Kendall 等级相关性几乎是完整集的两倍，这表明通过去除不可靠的链接可以获得改进的网格降水数据。CML 衍生的网格和来自一组独立测量仪的降雨观测之间的相关性，在两个研究区域的几次降雨事件中进行了测试，对于 CML 的可靠子集来说，比完整集更高。对于某些风暴事件，可靠 CML 集的 Kendall 等级相关性几乎是完整集的两倍，这表明通过去除不可靠的链接可以获得改进的网格降水数据。CML 衍生的网格与来自一组独立测量仪的降雨观测值之间的相关性，在两个研究区域的几次降雨事件中进行了测试，对于 CML 的可靠子集来说，比完整集更高。对于某些风暴事件，可靠 CML 集的 Kendall 等级相关性几乎是完整集的两倍，这表明通过去除不可靠的链接可以获得改进的网格降水数据。