A machine learning approach for predicting computational intensity and domain decomposition in parallel geoprocessing

ABSTRACT

High performance computing is required for fast geoprocessing of geospatial big data. Using spatial domains to represent computational intensity (CIT) and domain decomposition for parallelism are prominent strategies when designing parallel geoprocessing applications. Traditional domain decomposition is limited in evaluating the computational intensity, which often results in load imbalance and poor parallel performance. From the data science perspective, machine learning from Artificial Intelligence (AI) shows promise for better CIT evaluation. This paper proposes a machine learning approach for predicting computational intensity, followed by an optimized domain decomposition, which divides the spatial domain into balanced subdivisions based on the predicted CIT to achieve better parallel performance. The approach provides a reference framework on how various machine learning methods including feature selection and model training can be used in predicting computational intensity and optimizing parallel geoprocessing against different cases. Some comparative experiments between the approach and traditional methods were performed using the two cases, DEM generation from point clouds and spatial intersection on vector data. The results not only demonstrate the advantage of the approach, but also provide hints on how traditional GIS computation can be improved by the AI machine learning.     

洪山泉流量动态系统分析

为查明洪山泉域的开采程度,在前人研究基础上,建立了现阶段泉流量-降雨量的线性模拟模型。通过对模型的系统辨识,对系统的"慢时变"现象作出了合理解释,并发现系统在1994年以后发生了"快时变",说明开采量从1995年开始迅猛增加。同时,由多年来泉流量序列的自相关分析,发现其宏观周期到2000年基本保持不变,说明虽然第二输入(即人为开采量)对该系统的作用已不可忽略,但尚未达到足以改变其宏观频率的程度。     

三峡库区四方碑滑坡稳定性与变形趋势预测

三峡水库建成后,库水位周期性涨落和暴雨产生的渗流作用导致大量古滑坡的复活或新滑坡的发生。以库区近水平层状结构的四方碑滑坡为例,依据库水位实际调动,将水位从175 m至145 m不同降速与50年一遇暴雨进行工况组合,计算4种工况下滑坡的稳定性及破坏概率。然后采用Geo-studio软件的Sigma模块对滑坡进行变形模拟,运用R/S分析方法判断滑坡的变形持续性,并结合野外调查情况,综合评价分析四方碑滑坡的稳定性。结果表明:滑坡在各工况下整体均处于基本稳定状态,具有低危险性;变形模拟结果显示滑坡前缘位移最大,与野外调查情况一致;各监测点Hurst指数均介于0.5~1,表明时间序列具有正持续性,在研究的时间限度内滑坡的局部破坏增强,应在汛期加强对滑坡前缘的巡查和预警。     

Triangulated spherical splines for geopotential reconstruction

We present an alternate mathematical technique than contemporary spherical harmonics to approximate the geopotential based on triangulated spherical spline functions, which are smooth piecewise spherical harmonic polynomials over spherical triangulations. The new method is capable of multi-spatial resolution modeling and could thus enhance spatial resolutions for regional gravity field inversion using data from space gravimetry missions such as CHAMP, GRACE or GOCE. First, we propose to use the minimal energy spherical spline interpolation to find a good approximation of the geopotential at the orbital altitude of the satellite. Then we explain how to solve Laplace’s equation on the Earth’s exterior to compute a spherical spline to approximate the geopotential at the Earth’s surface. We propose a domain decomposition technique, which can compute an approximation of the minimal energy spherical spline interpolation on the orbital altitude and a multiple star technique to compute the spherical spline approximation by the collocation method. We prove that the spherical spline constructed by means of the domain decomposition technique converges to the minimal energy spline interpolation. We also prove that the modeled spline geopotential is continuous from the satellite altitude down to the Earth’s surface. We have implemented the two computational algorithms and applied them in a numerical experiment using simulated CHAMP geopotential observations computed at satellite altitude (450 km) assuming EGM96 (n _max = 90) is the truth model. We then validate our approach by comparing the computed geopotential values using the resulting spherical spline model down to the Earth’s surface, with the truth EGM96 values over several study regions. Our numerical evidence demonstrates that the algorithms produce a viable alternative of regional gravity field solution potentially exploiting the full accuracy of data from space gravimetry missions. The major advantage of our method is that it allows us to compute the geopotential over the regions of interest as well as enhancing the spatial resolution commensurable with the characteristics of satellite coverage, which could not be done using a global spherical harmonic representation. The results in this paper are based on the research supported by the National Science Foundation under the grant no. 0327577.     

一个有限差分大气环流模式的并行效率

适合于大规模并行处理机（ＭＰＰ）的中国科学院大气物理研究所新一代九层大气环流模式（ＩＡＰ ９Ｌ ＡＧＣＭ）被发展成功。本文描述了并行模式的设计原则，给出了该模式的并行版本在我国几种主要的并行机上的并行执行效率数据。采用消息传递方式（ＭＰＩ）的区域分解方法用于实现程序的并行，结果表明，对于该模式，只有一维区域分解方法才能取得较好的并行可扩展性，即随着处理器的增加而能得到相应较好的并行加速比和并行效率。     

精密时钟单钟幂律谱系数的取值分析

精密时钟噪声可用五个幂律谱系数来表征其强弱。由于不存在实际可用的理想时钟作为参考,用以估计幂律谱系数的单钟噪声无法直接测出,而只能利用钟组内各钟之间的钟差。然而这是一个不适定问题．本文探讨了单钟幂律谱系数的取值范围与优化估计方法。     

Clay mineral,geochemical and Sr–Nd isotopic fingerprinting of sediments in the Murray–Darling fluvial system,southeast Australia

The Willyama Supergroup of the Broken Hill region in southern Australia consists of supracrustal sedimentary and magmatic rocks, formed between 1810 and 1600 Ma. A statistical analysis of nearly 2000 SHRIMP U–Pb zircon spot ages, compiled from published and unpublished sources, provides evidence for three distinct tectonostratigraphic successions and four magmatic events during this interval. Succession 1 includes Redan Geophysical Zone gneisses and the lower part of the Thackaringa Group (Cues Formation). These rocks were deposited after 1810 Ma and host granite sills of the first magmatic event (1710–1700 Ma). Succession 2 includes the upper Thackaringa Group (Himalaya Formation), the Broken Hill Group and the Sundown Group and was deposited between 1710 and 1660 Ma. These rocks all contain detrital zircons from the first magmatic event (1710–1700 Ma) and in some cases from the second magmatic event (1690–1680 Ma). The second magmatic event (1690–1680 Ma) was bimodal, resulted from crustal extension, and was coeval with deposition of the Broken Hill Group and deepening of the basin. With this event a mafic sill swarm focused in the Broken Hill Domain. Mafic sills lack any trace of inheritance, unlike the granitoids that commonly contain inherited zircons typical of the supracrustal sediments. Succession 3, the Paragon Group and equivalents were deposited after 1660 Ma, but before a regional metamorphic event at 1600 Ma. Metamorphism was closely followed by inversion of the succession into a fold‐and‐thrust belt, accompanied by a fourth late to post‐orogenic magmatic event (ca 1580 Ma) characterised by granite intrusion and regional acid volcanism (the local equivalents of the Gawler Range Volcanics in South Australia).     

基于FDR原理的冻土测量方法

在我国地面气象观测中,冻土的自动观测一直未能实现,为了解决这一问题,本文基于频域反射(Frequency Domain Reflectometry,FDR)测量原理,通过测量土壤介电常数变化实现冻土测量的方法,设计了一种基于平面电容传感器分层检测冻土的传感器,土壤冻结时,其内部水分会相变为冰,水的介电常数远大于冰,利用水冻结相变后引起介电常数急剧变化的特性,建立了基于土壤介电常数、地温反演冻土的数学模型,并进行了典型土壤实验室冻结试验及外场对比观测试验,结果表明:冻土传感器能正确分辨土壤冻结状态,测量数据与人工观测趋势一致,相关系数可达0.99以上,平均测量误差小于3cm,基于介电特性的冻土传感器可以准确连续测量土壤的冻结深度及其生消变化。     

衰减夹层GPR模拟的时频域全局反射误差

全局反射误差分析是深入研究探地雷达(GPR)吸收边界条件吸收效率的有力工具.基于常规完全匹配层(PML)的标准交错网格有限差分算法必须满足严格的CFL条件限制,即在单位时间步长内,不容许电磁波传播的距离超过单元网格尺寸.为了提高主区域所有网格节点的计算效率,并有效地吸收传播后期出现的低频隐失波,提出基于非分裂递归卷积完全匹配层(UCPML)的旋转交错网格(RSG)GPR正演算法.该算法采用不同的网格交错策略,并在边界条件中引入了吸收低频隐失波的自由可变因子,使得该算法允许选取较大的时间步长,提高了计算效率,并且实现了对低频隐失波的高效吸收.本文首先给出了RSG差分格式,推导了基于UCPML的RSG差分更新方程,探讨了数值色散的稳定性条件,然后以绕射现象严重的衰减夹层数值模拟为例,分别从波场快照、单道波记录、时间域反射误差(TDRE)、频率域反射误差(FDRE)四个方面分析了UCPML与常规PML的全局反射误差,说明了基于UCPML和RSG的GPR正演算法能更有效地吸收低频隐失波.     

基于全波形采样的激电多信息提取方法研究与应用

激电法是地质勘查中的一种重要方法,时、频域测量各具优势,但传统的时、频域测量实现方法相对独立.根据频域和时域可相互转换的理论,本文提出一种基于全波形采样的时频激电多参数提取的地球物理数据处理方法.采取时间域激电法的观测模式,接收机高精度同步记录整周期电压-电流全部采样点的波形数据,通过该文提出的处理方法,仅一次供电和测量即可提取多种时、频域激电参数:时域激电的视电阻率、视极化率和频域激电的多频视电阻率、视相位、视频散率、去耦后的视相位等参数.在云南保山某典型矿区开展方法的试验应用,提取了地质体大量的相关激电信息,总结了目标体的激电参数响应规律.应用结果表明,该方法便捷高效,抗干扰能力强,相较于传统的时域激电测量,多种参数组合可提升对目标地质体的认识,增强激电法的应用效果.