The global terrestrial water storage (TWS), the most accessible component in the hydrological cycle, is a general indicator of freshwater availability on Earth. The global TWS trend caused by climate change is harder to detect than global mean temperature due to the highly uneven hydrological responses across the globe, the brevity of global freshwater observations, and large noises of internal climate variability. To overcome the climate noise and small sample size of observations, we leverage the vast amount of observed and simulated meteorological fields at daily scales to project global TWS through its fingerprints in weather patterns. The novel method identifies the relationship between annual global mean TWS and daily surface air temperature and humidity fields using multi-model hydrological simulations. We found that globally, approximately 50% of days for most years since 2016 have climate change signals emerged above the noise of internal variability. Climate change signals in global mean TWS have been consistently increasing over the last few decades, and in the future, are expected to emerge from the natural climate variability. Our research indicates the urgency to limit carbon emission to not only avoid risks associated with warming but also sustain water security in the future.

全球陆地水储量（TWS）是水文循环中最容易获得的组成部分，是地球上淡水可用性的一般指标。由于全球水文响应高度不均匀、全球淡水观测的短暂性以及内部气候变异的噪音较大，气候变化引起的全球TWS趋势比全球平均气温更难检测。为了克服气候噪声和观测样本量较小的问题，我们利用每日尺度的大量观测和模拟气象场，通过天气模式中的指纹来预测全球 TWS。该新方法利用多模型水文模拟来确定全球年平均 TWS 与每日地表气温和湿度场之间的关系。我们发现，自 2016 年以来，在全球范围内，大多数年份中，大约 50% 的日子里出现的气候变化信号高于内部变异的噪音。全球平均 TWS 中的气候变化信号在过去几十年中一直在持续增加，并且在未来预计将从自然气候变率中显现出来。我们的研究表明，限制碳排放的紧迫性不仅是为了避免与变暖相关的风险，也是为了维持未来的水安全。