基于深度学习的全天空相机成像日间云量计算研究

云量是天文领域中地基光电望远镜站址选择的重要评价参数之一。针对全天空相机成像的日间云量计算存在的问题,提出一种基于深度学习的全天空相机成像日间云量计算模型。云检测层,模型通过构建通道加权-特征融合(channel weighting-feature fusion,CWFF)结构,从而加强对云层记忆能力和深层特征的提取能力以完成云检测任务;云量计算层,模型提出一种基于云检测模型的云量计算方法,有效提高云量计算的误差率。实验表明,该方法在云检测任务中的综合准确率超过95%,在云量计算任务中的平均绝对误差不超过5%。

基于深度极限学习机的卫星云图云量计算

卫星云图云量计算是卫星气象应用的基础,现阶段对其的研究未能充分利用卫星云图的特征,导致云检测及云量计算的效果不好。针对该问题,利用多层神经网络进行卫星云图的特征提取,并通过大量实验寻找到最优的深度学习的网络结构。基于度极限学习机对卫星云图的云进行检测和分类,再利用"空间相关法"计算云图中的总云量。实验结果表明,基于传统极限学习机的深度极限学习机能够充分提取云图的特征,在进行云分类时能够较清晰地区分厚云和薄云间的界限。相比于传统阈值法、极限学习机模型以及卷积神经网络,深度极限学习机的云识别率以及云量计算准确率更高,且所提方法比卷积神经网络的效率更高。

基于卷积神经网络的卫星云图云量计算

卫星云图云量计算是卫星气象应用的基础,目前的方法对卫星光学参数以及对卫星云图的特征利用率不高,导致了云检测及云量计算不准确。针对这个问题,利用卷积神经网络进行卫星云图云的检测,基于优化的卷积神经网络将云图分为厚云、薄云及晴空。在云检测的基础上利用"空间相关法"计算总云量。针对中国HJ-1A/B卫星图片的实验结果表明,通过对卷积网络结构及参数的优化卷积神经网络可以很好的提取云图的特征,云分类时厚云和薄云之间的过渡区域清晰,云的识别率以及云量计算的准确率都比传统阈值法、动态阈值法以及极限学习机模型的结果要好。

GRAPES新云量计算方案的引进和影响试验

在GRAPES模式中分别引进了EC云量计算方案和WRF云量计算方案,并与GRAPES现用的云量计算方案和简单云方案进行了对比数值试验,结果表明:(1)这4种云量计算方案都能较准确地模拟云分布,相对而言,EC云方案和简单云方案模拟地更为准确。(2)简单云方案模拟的总云量较多,GRAPES云方案和WRF云方案模拟的总云量偏少,EC方案模拟的总云量较为接近观测值。(3)在中国东部地区模拟的地面温度与观测基本接近,但在中国西部地区则误差较大;采用EC云方案时模式模拟的地面温度更接近观测值。(4)综合分析模拟的效果,认为EC云量计算方案模拟的效果最佳,可以作为GRAPES新云量计算方案引进的参考。

一次长江流域梅雨降水中三种云量计算方案的对比研究

采用NCEP分析场,选取2010年梅雨期长江流域的一次降水过程,分别基于Slingo方案、NCAR方案和钱氏方案,利用相对湿度计算云量,并以LAPS(Local Analysis and Prediction System)系统输出的云量分析场作为观测值,分别在高层(400 h Pa)与低层(850 h Pa),从宏观比较与统计分析的角度,与计算结果进行云量大小与区域分布的对比分析。结果表明,三个云量计算方案对云量中心位置的把握均较为准确,但对云量值的计算存在大小不等的误差。NCAR方案计算结果和LAPS输出场最为吻合,能够体现出云量大值区,但区域一般偏大;Slingo方案相较NCAR方案来说略差,但也能较好地描述云带分布;此外,钱氏方案计算出的云量值始终偏小,但其能够较好地描述云带轮廓与云量的分布特征。综合对比结果,NCAR云量计算方案比其余两者更优,且在低层(850 h Pa)表现尤为明显。

GRAPES_Meso区域模式积云计算方案引进及预报效果检验

云量在模式中的准确模拟既是一个重点问题,也是一个难点问题。经评估发现,在GRAPES_Meso(Global/Regional Assimilation and Pr Ediction System)模式中存在模拟云量偏少的问题,而ECMWF模式中的积云云量计算方案对云量模拟具有一定优势。鉴于此,本文借鉴ECMWF积云云量计算方案对GRAPES中尺度模式中的云量计算方案进行了适当改进,并通过影响试验模拟三维云量和地面温度来验证其改进效果。结果表明:较改进前,改进云量方案模拟的云量增多,单层云量更加准确,垂直方向云层结构改进效果明显;地面温度与观测值对比误差减小。综合分析模拟结果,认为改进云量计算方案在GRAPES模式中对云量和地面温度的模拟有较好的改进。

Digital Analysis of the Water Layout Ecological Wisdom in Traditional Chinese Rural Settlements:A Case Study of Liukeng Village in Jiangxi Province

During the long-term construction and development process of eliminating water disasters and promoting water conservancy in traditional settlements,a set of mature strategies that have simple ecological wisdom in water layout have been formed by adapting to the natural water environment and utilizing the regional water system.This study conducted a qualitative and quantitative analysis of the water layout strategies and their effects on Liukeng Village from three aspects:deciphering the water systematic pattern,calculating the spatial characteristics and quantifying the water environment,to explore the technical assistance and potential of water layout research in traditional rural settlements.The results indicated that Liukeng has an unambiguously systematic water layout pattern of source diversion,middle drainage and end purification of the water.Through 3D point cloud computing,it was shown that the site selection made accurate use of micro-topography and adopted the strategy of a multi-source water management.It formed an organic water system pattern,which provided sufficient water sources for all kinds of needs of Liukeng Village.The Dragon Lake in Liukeng held 83.0%of the precipitation,and the vegetation area accounted for 34.7%of the total area of Liukeng,which had high surface permeability and good middle drainage effects.Water detection showed that the Dragon Lake provided good water quality and purification.The purposes of this study are to fill the gap in previous non-quantitative research on water layout in traditional rural settlements,excavate the hidden information and value of settlements,and deepen our understanding of the ecological wisdom of the overall planning,layout and construction of water conservancy in traditional rural settlements.This knowledge can assist the win-win situation of water conservancy cultural heritage protection and modern utilization.It also provides useful inspiration and reference for properly dealing with the problems of rain and flooding,realizing the sustainability of water resources,and protecting the ecological environment in the process of the development and construction of village settlements in China.

Experimental cryptographic verification for near-term quantum cloud computing

An important task for quantum cloud computing is to make sure that there is a real quantum computer running,instead of classical simulation.Here we explore the applicability of a cryptographic verification scheme for verifying quantum cloud computing.We provided a theoretical extension and implemented the scheme on a 5-qubit NMR quantum processor in the laboratory and a 5-qubit and 16-qubit processors of the IBM quantum cloud.We found that the experimental results of the NMR processor can be verified by the scheme with about 1.4%error,after noise compensation by standard techniques.However,the fidelity of the IBM quantum cloud is currently too low to pass the test(about 42%error).This verification scheme shall become practical when servers claim to offer quantum-computing resources that can achieve quantum supremacy.