**谷歌研究团队开发AI模型，可提前7天预测未监测河流洪水**
近日，国际知名科学期刊《自然》（Nature）发表了一篇来自Google Research洪水预测团队的最新论文，揭示了一项重大技术突破：人工智能（AI）模型能够在未安装水位测量仪的流域提前7天预测洪水。该模型由Grey Nearing及其团队开发，利用5680个现有的水位测量仪数据进行训练，以预测那些未被直接监测的河流在接下来7天内的日径流情况。
在评估过程中，谷歌的AI模型与目前全球领先的洪水预警系统——全球洪水预警系统（GloFAS）进行了对比。结果显示，该AI模型在同日预测准确率上与GloFAS相当，甚至在某些情况下表现更优。这一创新技术有望极大地提升全球洪水预警的及时性和准确性，特别是在那些缺乏监测设施的地区，将为防洪救灾工作提供宝贵的预警时间。
论文标题为“Global prediction of extreme floods in ungauged watersheds”（全球未监测流域的极端洪水预测），标志着人工智能在自然灾害预警领域取得了新的进展。这一成果的发布，不仅为科研界提供了新的研究方向，也为全球公共安全和环境管理提供了有力的科技支持。
英语如下：
**News Title:** “Google Research AI Model Breaks Record: Accurately Predicts Undetected River Floods Seven Days in Advance”
**Keywords:** Google AI, Flood Prediction, Nature Paper
**News Content:**
A Google Research AI model, developed by Grey Nearing and their team, has been revealed to predict floods in unmonitored rivers up to seven days ahead, according to a recent paper published in the renowned scientific journal *Nature*. The model, trained using data from 5,680 existing water level gauges, forecasts daily discharge in rivers that are not directly monitored.
During evaluations, Google’s AI model was compared with the world’s leading flood early warning system, the Global Flood Awareness System (GloFAS). The results showed that the AI model was comparable to GloFAS in same-day predictions and even outperformed it in certain instances. This innovative technology holds the potential to significantly improve the timeliness and accuracy of global flood warnings, particularly in regions with inadequate monitoring infrastructure, providing valuable lead time for flood mitigation efforts.
Titled “Global prediction of extreme floods in ungauged watersheds,” the paper signifies a new milestone in AI’s application in natural disaster early warning systems. The publication not only offers new research avenues for the scientific community but also furnishes robust technological support for global public safety and environmental management.
【来源】https://mp.weixin.qq.com/s/GoOPqLtdYvPv3_no7GJUJQ
Views: 1