基于贝叶斯机器学习的生态模型参数优化方法研究

参数优化方法是准确估计生态模型参数、降低其不确定性的有效手段。本文提出一种基于贝叶斯机器学习的No-U-Turn Sampler（NUTS）生态模型参数优化方法。NUTS是一种高效的参数优化方法,每次取样中利用递归算法生成候选参数集（二叉树）推断参数的后验信息,如果满足约束条件“非U型回转”,不断构建子树更新参数;否则,记录本次抽样的“最优”参数集,并开始下一次取样,直到获取足够样本。该算法在每次取样中充分优化参数,避免因随机游走行为产生冗余抽样,提高了参数优化效率。本文以千烟洲亚热带人工针叶林碳通量模拟为例,基于Pymc3框架利用NUTS参数优化方法实现了碳通量（Net Ecosystem Exchange,NEE）模型参数反演,并与Metropolis-Hastings（MH）方法进行对比。结果表明,本文算法的参数值达到稳定波动时的抽样次数减少了85%左右,参数优化效率提升3倍左右。参数优化后,2种NEE模型中7个参数不确定性降低10%~53%。此外,NEE模拟效果明显提升,模拟值与实测值的R²分别提高23%和17%,RMSE分别降低3%和4%。综上所述,本文提出的参数优化方法对生态领域的参数估计或数据同化工作具有一定的借鉴意义。     

湖南省慈利县农作物富硒程度调查评价与富硒农业发展建议

针对慈利县重点区域进行了农作物富硒程度专项调查,通过建立湖南省相关农作物富硒标准,发现采集的十种农作物中大部分达到富硒标准,其中油菜、水稻、玉米等硒平均含量最高,分别达到0.285mg/kg、0.136 mg/kg以及0.118 mg/kg。通过对比根系土硒含量,研究了十种作物硒富集规律,得出了各农作物硒富集系数,油菜硒富集能力最强,其次为大豆、大蒜。针对慈利县土壤、农作物含硒量特征提出了慈利县富硒农业发展的建议。     

Stability analysis of slope in strain-softening soils using local arc-length solution scheme

Soils with strain-softening behavior — manifesting as a reduction of strength with increasing plastic strain — are commonly found in the natural environment. For slopes in these soils,a progressive failure mechanism can occur due to a reduction of strength with increasing strain. Finite element method based numerical approaches have been widely performed for simulating such failure mechanism,owning to their ability for tracing the formation and development of the localized shear strain. However,the reliability of the currently used approaches are often affected by poor convergence or significant mesh-dependency,and their applicability is limited by the use of complicated soil models. This paper aims to overcome these limitations by developing a finite element approach using a local arc-length controlled iterative algorithm as the solution strategy. In the proposed finite element approach,the soils are simulated with an elastoplastic constitutive model in conjunction with the Mohr-Coulomb yield function. The strain-softening behavior is represented by a piece-wise linearrelationship between the Mohr-Coulomb strength parameters and the deviatoric plastic strain. To assess the reliability of the proposed finite element approach,comparisons of the numerical solutions obtained by different finite element methods and meshes with various qualities are presented. Moreover,a landslide triggered by excavation in a real expressway construction project is analyzed by the presented finite element approach to demonstrate its applicability for practical engineering problems.     

Landsat时序变化检测综述

时序变化检测已成为当前Landsat数据主流的变化检测方法。本文从检测算法对比、时序数据构建和精度评价等方面对Landsat时序变化检测进行回顾和评述,进而提出Landsat时序变化检测当前所存在的问题,及其所面临的挑战。Landsat时序变化检测算法可大致归纳为轨迹拟合法、光谱-时间轨迹法、基于模型的方法3大类,这些算法大多基于森林扰动提出;变化检测常用指标有波段型、植被指数型、线性变换型、组合型4大类,每类指标的优势不同,可综合多类指标以更全面地检测不同扰动类型。尽管Landsat时序变化检测已取得长足发展,但仍然面临诸多挑战,其中最大挑战是缺少一致性的参考数据集进行变化检测精度评价。     

Climate projections of spatial variations in coastal storm surges along the Gulf of Mexico and U.S. east coast

Using statistically downscaled atmospheric forcing, we performed a numerical investigation to evaluate future climate's impact on storm surges along the Gulf of Mexico and U.S. east coast. The focus is on the impact of climatic changes in wind pattern and surface pressure while neglecting sea level rise and other factors. We adapted the regional ocean model system(ROMS) to the study region with a mesh grid size of 7–10 km in horizontal and 18 vertical layers. The model was validated by a hindcast of the coastal sea levels in the winter of 2008. Model's robustness was confirmed by the good agreement between model-simulated and observed sea levels at 37 tidal gages. Two 10-year forecasts, one for the IPCC Pre-Industry(PI) and the other for the A1 FI scenario, were conducted. The differences in model-simulated surge heights under the two climate scenarios were analyzed. We identified three types of responses in extreme surge heights to future climate: a clear decrease in Middle Atlantic Bight, an increase in the western Gulf of Mexico, and non-significant response for the remaining area. Such spatial pattern is also consistent with previous projections of sea surface winds and ocean wave heights.     

青藏高原北部活动层土壤热力特性的研究

利用2003年10月～2004年9月期间高原北部可可西里(QT01)、北麓河(QT02)、开心岭(QT05)、通天河(QT06)等地活动层土壤温度梯度、土壤热通量及土壤水分的观测资料,计算了高原北部活动层土壤的导热率、土壤容积热容量、导温率等土壤热力参数.结果显示,QT02、QT05、QT06三站导热率、导温率夏秋季节较大而冬季较小,容积热容量则相反,表现为秋冬季节大而夏季较小;QT01站导热率表现为春季大,夏季较小;表层土壤粒度较小及较低的土壤湿度是冬季导热率较小的可能原因;冻土的热力特征参量可描述为相应深度的温度、体积含冰量及土壤盐度的函数,土壤含水量是融土热特征参数的主要影响因子;土壤水分含量小于某一临界值时,导温率随土壤水分含量的增大而增大,反之则减小.     

Micro‐geomorphology determines community structure of biological soil crusts at small scales

Biological soil crusts (BSCs) are ubiquitous communities of diminutive organisms such as cyanobacteria, green algae, lichens, mosses and others associated closely with particles of surface soil, forming a cohesive thin horizontal layer. The ecological roles of BSCs affecting soil nutrient cycling, stability and hydrological processes, influencing the germination and establishment of vascular plants, and serving as habitats for numerous arthropods and microorganism have been well documented. We tested the hypothesis that micro‐geomorphological features determine the spatial distribution of BSCs by reallocating related abiotic resources at small‐ and medium‐scales in the Tengger Desert. Our results showed that higher soil pH and higher total potassium content in topsoil positively correlated with the colonization of cyanobacteria and algae in the earliest successional stages of BSCs, while increasing dust deposition onto the topsoil enhanced the development of lichen and mosses in the later stages of BSCs. Increasing soil moisture raised the proportion of mosses and lichen in BSCs, this will possibly change the ecological functions of BSCs, such as nitrogen‐fixation by cyanobacteria, due to the conversion from a complex to relative simple type of BSC. Micro‐geomorphology has created various habitats at a small‐scale affecting colonization and development of cryptogams. This paper considers the contribution of micro‐geomorphology to biodiversity in the extreme arid desert systems. Copyright © 2010 John Wiley & Sons, Ltd.     

深埋电极的地电阻率观测研究

首先研究了四极观测系统装置系数与电极埋深的关系;然后给出了点电流源在3层地壳模型的地表和第二层时,电源所在层的电位的解析表达式;最后将天津宝坻地区的电性结构简化成一个3层模型,计算给出了当地表层和基岩中的电阻率出现变化时,在地表和基岩上层开展四极地电阻率观测结果与供电极距和深度的关系.     

龙门山断裂带南北活动差异及复杂构造建模——基于地表破裂带和中小震精定位

位于青藏高原东缘的龙门山断裂带一直以来是研究青藏高原隆升、四川盆地含油气构造演化、南北地震带地震危险性分析的重点地区,对新构造运动、油气勘探、地震研究均有重要意义。自从汶川地震在龙门山断裂带上发生以来,各研究领域的专家利用已有的资料,基于各自对龙门山断裂带构造的认识,对该地震的发震构造问题开展了广泛的讨论,提出了多种发震构造模式,但大部分仅提出了一个或是几个二维的发震构造剖面,部分三维发震构造模型也仅是不考虑断层相互切割关系的简单化的有限元模型,且部分模型仅仅是提出了起始破裂处的发震模型。鉴于汶川地震长达240km的地表破裂带,仅建立局部地区的构造模型是远远不够的,而汶川地震地表破裂带和中小震精定位结果、震源机制解均表明龙门山断裂带的构造存在较大的南北活动的差异。因此,研究地震发生和传播机制需要建立整个龙门山断裂带的复杂三维构造模型。     

农作物产量对夏季累积高温的气候响应——以南京市为例

利用南京站1949—2008年逐日气温、逐月降水距平百分率资料及南京市统计年鉴数据,分析农作物产量对夏季累积高温的气候响应特征。结果表明：1949年以来南京市夏季累积高温呈波动上升趋势;最近60年（1949—2008年）和最近30年（1979—2008年）夏季累积高温的线性增长率分别为0.35℃/a和2.88℃/a;夏季累积高温与夏、秋粮产量均存在显著的负相关,相关系数分别为-0.66和-0.62,油料作物对夏季累积高温的变化也有一定的响应,但不如前两者显著,棉花产量对夏季累积高温变化不敏感。