Water balance modelling in lateritic terrain

An analytical approximate model for unsaturated flow in a spatially variable field, coupled with infiltration and evapotranspiration at the upper boundary and a fluctuating water-table at the lower boundary, has been developed. The unsaturated flow equations depend on parameterizations of θ(Φ) and K(Φ). They are based on the notion of a moving, discontinuous front. The field heterogeneity refers to saturated hydraulic conductivity only. Horizontal variability is considered, and the flow medium is approximated as a set of uncorrelated, vertically homogeneous columns. Expectations and variances obtained with this approach have been compared with observations of the field hydrological processes. Three important aspects of the hydrology in this lateritic terrain are rapid water-table response, Hortonian surface runoff generation and soil suction variability. The stochastic conceptualization used explains to a high degree these characteristics, although some limitations are demonstrated.     

Derivation of planetary topography using multi-image shape-from-shading

In many cases, the derivation of high-resolution digital terrain models (DTMs) from planetary surfaces using conventional digital image matching is a problem. The matching methods need at least one stereo pair of images with sufficient texture. However, many space missions provide only a few stereo images and planetary surfaces often possess insufficient texture.This paper describes a method for the generation of high-resolution DTMs from planetary surfaces, which has the potential to overcome the described problem. The suggested method, developed by our group, is based on shape-from-shading using an arbitrary number of digital optical images, and is termed “multi-image shape-from-shading” (MI-SFS). The paper contains an explanation of the theory of MI-SFS, followed by a presentation of current results, which were obtained using images from NASA's lunar mission Clementine, and constitute the first practical application with our method using extraterrestrial imagery. The lunar surface is reconstructed under the assumption of different kinds of reflectance models (e.g. Lommel-Seeliger and Lambert). The represented results show that the derivation of a high-resolution DTM of real digital planetary images by means of MI-SFS is feasible.     

Simulation of groundwater dynamics in the North China Plain by coupled hydrology and agricultural models

We simulated the effects of irrigation on groundwater flow dynamics in the North China Plain by coupling the NIES Integrated Catchment‐based Ecohydrology (NICE) model with DSSAT‐wheat and DSSAT‐maize, two agricultural models. This combined model (NICE‐AGR) was applied to the Hai River catchment and the lower reach of the Yellow River (530 km wide by 840 km long) at a resolution of 5 km. It reproduced excellently the soil moisture, evapotranspiration and crop production of summer maize and winter wheat, correctly estimating crop water use. So, the spatial distribution of crop water use was reasonably estimated at daily steps in the simulation area. In particular, NICE‐AGR reproduced groundwater levels better than the use of statistical water use data. This indicates that NICE‐AGR does not need detailed statistical data on water use, making it very powerful for evaluating and estimating the water dynamics of catchments with little statistical data on seasonal water use. Furthermore, the simulation reproduced the spatial distribution of groundwater level in 1987 and 1988 in the Hebei Plain, showing a major reduction of groundwater level due mainly to overpumping for irrigation. Copyright © 2006 John Wiley & Sons, Ltd.     

SAS/ETS在变形监测数据处理中的应用

在综合考虑数学模型的预测精度与编程实现的可操作性基础上,提出了利用模块对变形监测数据进行时间序列分析,并概括介绍了该模块的3个主要命令使用方法,解决了时间序列分析在监测数据处理中实现起来非常繁琐的问题。最后通过与环境下动态灰色模型综合比较,证明了时间序列分析在变形监测数据处理中操作简单,模拟精度和预测精度都较高。     

利用人工鱼群算法优化高斯过程模型及应用分析

基于高斯过程利用传统共轭梯度法搜索超参数,存在对初始值依赖性强、获得局部最优解的不足。本文采用人工鱼群算法对超参数进行智能寻优,建立了基于人工鱼群算法的高斯过程模型对变形体形变进行预测分析。通过隧道和基坑两个工程实例计算比对分析,NN、SE和RQ 3种核函数中NN核函数的预测效果最好,平均相对误差分别为0.69%和1.06%。结果表明超参数优化模型的预测精度得到了较大的提高,改善了高斯过程算法本身存在的超参数求解方面的不足,效果明显。     

“一带一路”区域可持续发展生态环境遥感监测

2013年9月和10月,习近平主席在出访中亚和东南亚国家期间,先后提出了共建"丝绸之路经济带"和"21世纪海上丝绸之路"(简称"一带一路")的重大倡议。要全面保护"一带一路"区域生态环境,实现2030年可持续发展目标,是一个具有挑战性的问题。遥感技术对生态环境监测与评价发挥着十分重要的作用。本研究利用多尺度、多源遥感数据,对2015年"一带一路"区域的生态环境状况进行监测和分析,旨在提供可持续发展目标生态环境遥感监测的本底。本文选取了几个重要的生态环境方面开展监测与分析,主要包括宏观生态系统结构和植被状况、太阳能资源分布、水资源平衡、主要生态环境限制因素对经济走廊建设的影响、主要城市生态环境质量等。监测区域覆盖亚洲、非洲、欧洲和大洋洲的陆上区域。研究结果为生态环境评价与保护提供了有效的决策依据,有助于"一带一路"建设积极推进。     

基于高密度电阻率法的土体干缩裂隙动态发育过程精细监测研究

土体在气候作用下发育干缩裂隙是一种常见的自然现象,裂隙的存在会极大弱化土体的工程性质,诱发许多岩土与地质工程问题。为了实时掌握黏性土中干缩裂隙网络的发展状态,提出了一套基于高密度电阻率层析成像技术（ERT）的土体干缩裂隙动态发展过程精细监测方法。分别开展模型试验及原位试验,利用自行研制的测定系统持续采集电流-电位差数据,随后利用自行开发的有限元法电阻率层析成像（FemERT）系统进行数据处理,获取了裂隙网络在不同发育阶段的空间分布特征。结果表明：（1）ERT可以实现土体裂隙发育过程的精细监测,具备监测三维裂隙网络几何形态的能力,裂隙宽度的识别精度达到毫米级,裂隙深度的识别精度达到厘米级;（2）ERT的感度分布特征解释了裂隙发育对于土体电阻率的影响规律,测定电阻值时程曲线因裂隙产生位置的不同而呈现不同的变化规律;（3）反演电阻率及其相对变化率（Rev）可以直观表征裂隙网络在不同阶段的空间几何形态,凸显裂隙动态发育过程对于土体导电性的影响。     

Key Problems and Countermeasures in CO2 Flooding and Storage

Based on literature research in combination with the practice of CO₂ flooding and storage in Jilin Oilfield, this study assesses the key problems in CO₂ flooding and storage, proposing the corresponding countermeasures from five aspects of CO₂ gas source condition, namely geological condition evaluation, scheme design incoordination with other production methods, economic and effectiveness evaluation, together with dynamic monitoring and safety evaluation. The re...     

区域地质调查文本中文命名实体识别

作为我国地质调查领域最重要的数据源之一,地质调查报告中蕴含着丰富的地学知识及地质体描述等关键信息,准确高质量地抽取地质命名实体为地学知识图谱构建、知识推理及知识演化提供基础。笔者等在阐述地质命名实体识别任务基础上,分析地质实体不仅包含大量专业术语,还存在实体嵌套、大量长实体等领域特性,进一步增加了地质命名实体识别难度。笔者等提出一种基于轻量级预训练模型(ALBERT)—双向长短时记忆网络(BiLSTM)—条件随机场(CRF)模型的地质命名实体识别方法。首先利用ALBERT对输入字符上下文特征进行建模,并采用BiLSTM对其进行进一步上下文特征表征,最后采用CRF实现标注序列预测。实验结果表明,在构建的地质命名实体识别数据集上,相比于主流的命名实体识别模型算法,本文所提出的方法具有更好的抽取性能,提出的命名实体识别模型能为领域实体识别提供借鉴,同时为地学领域实体关系抽取和地学知识图谱构建提供有力方法支撑。     

IMERG卫星降水数据在嘉陵江流域的干旱监测效用评估

针对嘉陵江流域存在雨热同期,水旱灾害频发的现象,为快速且准确地把握流域内降水与干旱情况,利用覆盖范围广且分辨率高的网格化IMERG卫星降水数据对嘉陵江流域进行多时空尺度反演,并基于卫星降水数据采用标准化降水指数(SPI)对流域实行干旱监测。结果表明：1)根据分类指标与统计指标的计算结果,三种卫星降水数据中的IMERG-F能更准确地反映流域内的日降水量,与地面降水数据CC达0.737,整体高估地面降水2.6%,具有在干旱监测方面的应用潜力。2)三种卫星降水数据驱动的SPI指数在干旱监测方面存在一定的差异,IMERG-F驱动的SPI指数与地面降水数据驱动的SPI指数保持较高的一致性(CC>0.9),较近实时产品IMERG-F更能准确地呈现出流域的干湿特征。3)卫星识别降水与干旱监测的能力受地形地貌的影响,IMERG卫星降水数据在平原丘陵地带具有较好的适用性。