针对大坝变形监测数据呈现无规律特性和传统分解方法的不足,本文基于混合模型的思想提出了基于LMD-LSTM的大坝变形混合预测模型.首先利用局域均值分解(LMD)方法将原始数据进行分解,提取出不同频率的分量,再用长短期记忆人工神经网络(LSTM)预测模型对各个分量分别进行建模,最终将各个分量的预测值叠加重构以获得大坝变形预测值.工程实例表明,LMD-LSTM模型预测结果与实际观测值拟合较好,其精度指标MAE、MAPE和RMSE分别是1.2 mm、4.12×10-5％和2 mm.相较于LSTM和EMD-LSTM模型的精度指标,该文提出的混合模型预测精度更高,为大坝变形预测提供了一种新方法.

滑坡灾害作为一种与人们生命紧密相连的常见自然灾害,对其进行危险性评价研究极为重要.以江西省九江市修水县作为研究区域,以修水县243个滑坡地质灾害点作为研究对象,根据对修水县滑坡灾害的发育特征和关联因素的分析,选取了九大评价因子,利用信息量模型和AHP对修水县进行滑坡地质灾害危险性分区.其按危险程度分为极低、低、中等、高和极高5个危险区,分别占总面积的4.25％、14.97％、32.14％、35.17％和13.58％.综合研究区滑坡概况对各危险区进行分析,为研究区的地质灾害预防提供建议.

Urban development has advanced the accumulation of steel resources from the natural environment to urban structures, thereby aggravating the deterioration of the natural environment and increasing concern for the circular economy. Accurately understanding the spatiotemporal evolution of urban steel stocks is necessary to better maintain sustainable socioeconomic development. This paper constructs the consistency and continuity of multisource nighttime light (NTL) data and the results of modelled in-use steel stock (IUSS) in the Belt and Road (B&R) areas at a 1 km resolution. The multiscale IUSS spatiotemporal evolution in the B&R areas was quantified using geographic information analysis methods. The results revealed that the evolution of IUSS in the southern and northern parts of the B&R has significant differences. From 1992 to 2004, the evolutionary state of the IUSS showed drastic changes, while an overall trend of balanced development was maintained from 2004 to 2019. High-intensity IUSS is mainly concentrated in China, Russia, and India, and wealth level is the primary factor driving the development of IUSS. Combining the results of the development state analysis and spatial correlation analysis shows that central India and central and western China are potential hotspot areas for the flow of steel resources in the future. The results of this study can help develop differentiated and refined management strategies for the flow of anthropogenic steel resources.

洪灾数据类型多样、结构复杂、数据量巨大,随着时间累积给传统平台带来巨大压力.针对这个问题,通过结合大数据处理技术,将洪灾数据存储到云平台中,提高洪灾数据查询效率.根据洪灾数据的特征,将类型和种类进行分类,并通过HBase与HDFS相结合对洪灾数据存储进行设计.最终实现了洪灾数据的分布式存储、查询与管理,通过设计索引,提高了洪灾数据的查询效率和速度.应用结果表明,随着数据增长,研究设计的存储方案查询时间逐渐接近并小于传统平台的查询时间,能够提高洪灾数据利用效率.