深度学习PaddlePaddle框架支持下的遥感智能视觉平台研究与实现

百度深度学习PaddlePaddle框架支持下的遥感智能视觉平台,能够运用深度学习技术实现遥感影像的智能建模、训练和解译。本文通过深入分析PaddlePaddle图像分割模型库PaddleSeg的图像处理深度学习算法模型DeepLabV3+、U²-Net及RetinaNet,开发设计了遥感智能视觉平台,实现了遥感影像的地块分割、变化检测和斜框检测等专业功能。研究表明：遥感智能视觉平台提取的图斑总面积是目视解译的80%、有效图斑比例为76%、错误图斑比例为18%,实现了快速有效的遥感图像智能处理。     

河流泥沙输移过程中矿物风化的碳汇效应初探——以长江干流为例

本文首先提出了河流泥沙输移过程中泥沙中的钙镁矿物溶蚀消耗水体中的CO2并具有碳汇功能的观点。基于前人长江干流从源头到入海口和支流2003~2007年期间4次河流悬移质泥沙的化学元素组成和矿物组成资料,分析悬移质中CaO、MgO含量和方解石、白云石含量变化特征,定量计算了这些取样点悬移质泥沙的CO2总碳汇能力和非永久性、永久性碳汇能力,分析了不同碳汇能力沿程变化规律及其原因。碳汇计算结果表明：寸滩—大通河段1956~2000年期间泥沙输移过程中钙镁矿物溶蚀产生的总碳汇量、非永久性和永久性碳汇量分别为2572万t/a、1700万t/a和872万t/a。由于输沙量减少,寸滩站—大通站河段的总碳汇量、非永久性和永久性碳汇量2006~2019年期间较1956~2000年期间相应分别减少了1852万t、1224万t和872万t。三峡水库年均淤积量1. 145亿t,损失总碳汇量675. 6万t,相当于三峡电站减排二氧化碳8580万t的7. 9%。全球河流入海年输沙量126. 1亿t,以寸滩- 吴淞口河段碳汇功能0. 060 t/t计,总碳汇量7. 57亿t相当于全球岩石风化碳汇总量10. 56亿t CO2的71. 6%。河流泥沙输移过程中钙镁矿物溶蚀的碳汇量具有重要的作用,其溶蚀速率大于原地风化。     

河北矾山超钾质岩浆氧逸度及含水性特征的磷灰石地球化学记录

产于华北克拉通北缘的晚三叠世矾山超钾质碱性岩杂岩体主要由单斜辉石岩和辉石正长岩等组成,以发育内生磷灰石-磁铁矿矿床而闻名。本文选取矾山杂岩体中的磷灰石为研究对象,通过测试磷灰石的主微量元素及原位Sr-Nd同位素组成以探讨母岩浆的氧逸度及含水性特征。研究表明,矾山磷灰石以轻稀土元素富集、重稀土元素亏损为特征,显示出缺乏显著Ce异常的“右倾”的稀土配分型式,且具有较低的Ga含量及较高的δEu值,暗示杂岩体具有较高的岩浆氧逸度,这也得到了磷灰石Mn氧逸度计的支持。所有磷灰石以高Sr、低Y为特征,具有较高的Sr/Y(>30)和La/Yb(>90)比值,且Sr/Y与La/Yb存在明显的正相关关系,表明矾山超钾质岩浆具有较高的含水量。矾山磷灰石的原位Sr-Nd同位素投点位于地幔序列,明显偏离地壳趋势。母岩浆的“氧化且富水”特征是矾山杂岩体得以发生磷灰石-磁铁矿矿化的重要因素,其主要与板片俯冲过程中蚀变洋壳对岩石圈地幔的交代作用有关。通过对比矾山杂岩体与全球范围内其他地质环境中磷灰石的成分数据,发现具有“高Sr、低F、少Cl”特征的磷灰石可作为识别超钾质碱性岩杂岩体(及其相关金属矿床)的...     

Critical conditions for natural gas charging and delineation of effective gas source rocks for tight sandstone reservoirs

Although it has been shown that the potential of tight‐sand gas resources is large, the research into the mechanisms of hydrocarbon charging of tight sandstone reservoirs has been relatively sparse. Researchers have found that there is a force balance during hydrocarbon charging, but discriminant models still have not been established. Based on the force balance conditions observed during gas migration from source rocks to tight sandstone reservoirs, a calculation formula was established. A formula for identifying effective source rocks was developed with the gas expulsion intensity as the discrimination parameter. The critical gas expulsion intensity under conditions of various burial depths, temperatures, and pressures can be obtained using the calculation formula. This method was applied in the Jurassic tight sandstone reservoirs of the eastern Kuqa Depression, Tarim Basin, and it was calculated that the critical expulsion intensity range from 6.05 × 10⁸ m³/km² to 10.07 × 10⁸ m³/km². The critical gas charging force first increases with depth and later decreases with greater depths. The distribution range of effective gas source rocks and total expelled gas volume can be determined based on this threshold. This method provides new insight into and method for predicting favourable tight‐sand gas‐bearing regions and estimating their resource potentials. Copyright © 2014 John Wiley & Sons, Ltd.     

Deep convective clouds over the northern Pacific and their relationship with oceanic cyclones

Based on combined Cloud Sat/CALIPSO detections, the seasonal occurrence of deep convective clouds(DCCs) over the midlatitude North Pacific(NP) and cyclonic activity in winter were compared. In winter, DCCs are more frequent over the central NP, from approximately 30°N to 45°N, than over other regions. The high frequencies are roughly equal to those occurring in this region in summer. Most of these DCCs have cloud tops above a 12 km altitude, and the highest top is approximately 15 km. These wintertime marine DCCs commonly occur during surface circulation conditions of low pressure, high temperature, strong meridional wind, and high relative humidity. Further, the maximum probability of DCCs,according to the high correlation coefficient, was found in the region 10°–20° east and 5°–10° south of the center of the cyclones. The potential relationship between DCCs and cyclones regarding their relative locations and circulation conditions was also identified by a case study. Deep clouds were generated in the warm conveyor belt by strong updrafts from baroclinic flows. The updrafts intensified when latent heat was released during the adjustment of the cyclone circulation current. This indicates that the dynamics of cyclones are the primary energy source for DCCs over the NP in winter.     

鄂尔多斯盆地深部富铀烃源岩提供铀源可能性的实验研究

鄂尔多斯盆地上三叠统延长组长7段石油优质烃源岩以富铀为特色,而盆地侏罗?白垩纪地层以赋存大量砂岩型铀矿而著称.为探索深部长7富铀烃源岩为浅部砂岩型铀矿形成提供铀源的可能性,选择盆地南部的富铀烃源岩、油样以及具有代表性的碳酸铀酰作为反应物,在中?高温高压还原条件下分别进行烃源岩生排烃?排铀模拟实验和油?铀关系模拟实验.结...     

大陆深部地壳物质成分识别方法综述

本文系统总结了大陆深部地壳物质成分识别研究方法和元素丰度合理性检验的方法,以期为大陆深部地壳物质成分探测提供技术方法体系。深部地壳物质成分识别的主要方法有:①因构造运动抬升出露到地表的中下地壳剖面;②地表出露的高级变质地体;③产于火山岩中的深部地壳捕掳体,如麻粒岩捕掳体;④地球物理测深资料与深部岩石物理性质的高温高压实验测定结果之间的拟合和⑤壳源岩浆岩源区地球化学示踪。元素丰度合理性检验的方法主要有地表热流和元素比值法。     

不同卸荷速率下岩石强度特性研究

卸荷条件下的岩石强度特性研究对于分析开挖作用下岩石工程的安全性具有极为重要的意义。卸荷试验中影响岩石强度的因素很多,包括岩性、卸荷点的位置、卸荷应力路径和卸荷速率等,以锦屏大理岩为对象,重点研究卸荷速率对强度的影响。提出用屈服接近速率来统一表征加荷和卸荷下的应力增加或减小的速率,通过比较加荷和卸荷条件下的屈服接近速率表明,当围压卸荷速率为常规试验轴向应力加荷速率的0.2～0.3倍时,两种应力路径下接近破坏时的屈服接近速率相当。通过2D弹塑性细胞自动机的数值试验和大理岩的室内卸围压试验来验证理论分析的结果,表明:围压卸荷速率越大,岩样的强度就越高;当卸荷速率为常规轴向加荷速率的0.2～0.3倍时,卸荷和加荷路径下的强度相当。从而得出结论:对于各种卸荷应力路径,如果卸荷点处于弹性范围内,当接近破坏时的屈服接近速率一定时,应力路径对强度的影响不明显。     

大断面小净距大帽山隧道现场监控量测及分析

结合大帽山隧道的工程实践,通过围岩内部位移、拱顶沉降、围岩压力和锚杆应力的现场监控量测工作,研究复杂地质条件下大断面小净距隧道双侧壁导坑法施工时围岩的稳定性。阐明分导洞开挖时围岩内部位移的变化趋势、特点及位移场,相邻导洞施工时的相互影响,围岩与支护结构间的相互调整变形机制,拱顶沉降捕捉的变形小于围岩实际变形的原因,支护结构的压力和锚杆应力状态及其与围岩位移的变化关系。监测结果表明,大断面小净距隧道Ⅴ级围岩段的破碎带采用现有的施工工艺和支护参数是可行的,围岩变形可控,支护结构的支护效果显著,围岩基本稳定。研究的方法、分析和结论可为类似条件下隧道工程的设计、施工、监测和进一步的理论研究提供参考和借鉴。     

基于PSO-PP的围岩稳定性评价

围岩的稳定性评价是一个复杂的不确定系统问题。结合投影寻踪方法、粒子群算法和逻辑斯谛曲线函数,建立了围岩稳定性评价的粒子群优化投影寻踪(projection pursuit based on particle swarm optimization,PSO-PP)模型。该模型一方面采用粒子群算法优化投影指标函数及逻辑斯谛曲线函数参数,确保了模型的准确性;另一方面利用逻辑斯谛曲线函数建立投影值与经验等级之间的非线性关系。模型的测试结果显示了良好的精度,实例分析结果与实际状态完全一致,表明该模型在围岩稳定性评价中的可行性和有效性。