Automated terrain feature identification from remote sensing imagery: a deep learning approach

ABSTRACT

Terrain feature detection is a fundamental task in terrain analysis and landscape scene interpretation. Discovering where a specific feature (i.e. sand dune, crater, etc.) is located and how it evolves over time is essential for understanding landform processes and their impacts on the environment, ecosystem, and human population. Traditional induction-based approaches are challenged by their inefficiency for generalizing diverse and complex terrain features as well as their performance for scalable processing of the massive geospatial data available. This paper presents a new deep learning (DL) approach to support automatic detection of terrain features from remotely sensed images. The novelty of this work lies in: (1) a terrain feature database containing 12,000 remotely sensed images (1,000 original images and 11,000 derived images from data augmentation) that supports data-driven model training and new discovery; (2) a DL-based object detection network empowered by ensemble learning and deep and deeper convolutional neural networks to achieve high-accuracy object detection; and (3) fine-tuning the model’s characteristics and behaviors to identify the best combination of hyperparameters and other network factors. The introduction of DL into geospatial applications is expected to contribute significantly to intelligent terrain analysis, landscape scene interpretation, and the maturation of spatial data science.     

新源县复杂地形下的风电场风场结构特征分析

利用新源县风电场周边气象站及测风塔观测资料、ERA5-Land再分析资料和数值模拟结果,分析新源县风电场的风场结构特征。结果表明：（1）风电场周边气象站的逐小时10m风速均呈现出早、晚偏低,中午偏高的变化规律。喀拉布拉镇站和公安农场站的风向主要以东风和南风为主,肖尔布拉克沟站的风向则以南风为主。测风塔4302的风速随着高度变化不明显。测风塔4301-4306均存在南风和北风,但各风向占比有一定差异。随高度升高,测风塔南风风向呈现出东南转西南的趋势。（2）ERA5-Land资料不能很好地再现研究区域风场变化。（3）数值模拟的风场变化具有一定山谷风特征。夜间的风向以东南风为主,白天则低海拔河谷地带风吹向山顶,北部谷风逐渐主导东北风向。20时,风电场区域主要以西北风为主。     

Electromagnetic characterization of polar ice-wedge polygons: Implications for periglacial studies on Mars and Earth

Polygonal terrain is found in a variety of polar environments on Earth and Mars. As a result, many areas of northern Canada may represent ideal terrestrial analogues for specific regions of Mars - in particular the northern plains. In the Canadian Arctic, polygon troughs are commonly underlain by wedges of massive ice, with rare examples of other wedge types. If the same is true for Mars, this raises interesting implications for the processes that concentrate H₂O at the Martian poles. This study uses an electromagnetic induction sensor to investigate the electromagnetic characteristics of terrestrial polar ice-wedge polygons. Surveys were conducted in two regions of the Canadian Arctic using a DUALEM-1S dual-geometry electromagnetic induction sensor, which measures electrical conductivity in the first 1.5-2 m of the subsurface. At locations where strong geomorphological evidence of ice was found, polygon troughs corresponded to local conductive anomalies. Trenching confirmed the presence of ice wedges at one site and allowed ground-truthing and calibration of the geophysical data. Previously unknown bodies of massive ice were also identified through the use of this geophysical technique. This study shows that an electromagnetic induction sounder is a useful instrument for detecting and mapping out the presence of subsurface ice in the Canadian Arctic. Taking together with its small size, portability and ruggedness, we suggest that this would also be a useful instrument for any future missions to Mars’ polar regions.     

Comparison of ground- and aerial-based approaches for quantifying polygonal terrain network geometry on Earth and Mars via spatial point pattern analysis

Recently, a particular statistical method - spatial point pattern analysis (SPPA) - has been introduced as an effective means by which qualitative, observable variations in polygonal terrain network arrangements on Earth and Mars can be quantified. A number of ground- and aerial-based techniques are available from which to derive the required input data: the spatial (x-y) coordinates of all polygon trough intersections within the site. However, each of the data collection methods may contain some level of error. Thus, the overarching question addressed by this research is: “how are the results of SPPA affected by the method by which the input data were generated?” At two polygonal terrain sites in the Canadian High Arctic, we performed ground-based surveys using differential and non-differential Global Positioning Systems (GPS) as well as photogrammetric analysis of aerial and satellite images of varying resolution to determine the trough intersection coordinates. It was found that the most robust statistical results were produced when using data from a combination of differential GPS surveys and high-resolution (∼0.25 m/pixel) aerial images. Images of pixel size ≥1 m were found to be unsuitable for this type of analysis. With respect to the investigation of similar Martian landforms, HiRISE and MOC images of polygonal terrain sites in southwestern Utopia Planitia were analyzed. Our results show that it is strongly preferable to perform SPPA using HiRISE images, though an empirical model is outlined that could be used to correct for errors arising from the reduced resolution inherent to MOC images.     

分块地形坐标大气环流模式框架的计算稳定性及数值试验

用数值试验方法讨论了分块地形η坐标大气环流模式动力框架的计算稳定性; 采用实际初始场, 该框架能够长期稳定地积分.在积分前100d内总有效能量变化很小, 总质量守恒、各物理量场的数值和波形都得到很好保持, 这表明该框架稳定性好.在分析了该框架的积分结果后可知其合乎物理事实, 与IAP 2.0的动力框架相比, 能更好地反映出大地形的环流形势, 这说明有大地形时采用分块地形η坐标要较σ坐标好.另外, 还对青藏高原等大地形的纯机械强迫作了数值试验, 得到了地形强迫平均槽脊的分布特征, 给出了地形纯机械强迫造成的准定常行星波的图像. 这些工作表明采用分块地形η坐标处理大地形是合理可行的.     

DEM精度评估方法的研究与实践

研究地形模型精度的关键就是要找到影响地形模型建模质量的相关因素及其他们之间的误差关系,以检测现有的DEM模型的精度,同时规范DEM模型建立的精度标准,使DEM模型能够更加准确、逼真地反映实际地面特征是DEM精度评估和质量控制的难点问题。本文从几个方面对DEM的精度评估方法进行了研究与实践,为今后DEM的精度评估提供了科学的参考依据和评估模型。     

全球虚拟地形环境中Mipmap纹理技术研究

Mipmap纹理技术是目前解决纹理分辨率与视点距离关系的最有效途径.将Mipmap纹理引入全球虚拟地形环境的金字塔模型中,在分析现有Mipmap纹理生成技术不足的基础上,提出了一种新的Mipmap纹理生成技术,并针对Mipmap纹理以多级形式存储会使数据量增加的问题,提出了Mipmap纹理压缩算法,以提高实时渲染中Mipmap纹理的装载速度和创建速度,并用实验结果验证了算法的可行性.最后给出了全球金字塔模型中Mipmap压缩纹理的存储结构.     

WindSim软件在复杂地形风电场风能资源评估中的应用

复杂地形的风能资源评估研究对于风电事业的发展具有极为重要的意义。作者利用基于计算流体力学(CFD)方法的风能资源评估系统软件WindSim对我国复杂地形新疆托克逊县境内某风电场2007年的风资源情况进行了模拟,并将模拟结果与测风塔观测结果进行对比分析。模拟结果表明,WindSim软件对我国复杂地形地区平均风速的模拟能力比较好,10m、40 m、50 m、70 m高度的月平均风速的平均相对误差分别是12.91%、10.21%、9.68%、12.91%。同时,WindSim软件不仅可以较为准确地模拟出该区域2007年的主导风向,而且对该年的有效风速频率的模拟效果也很好。此风电场风能资源模拟试验说明,该地区具有十分丰富的可供开发利用的风能资源,进一步说明WindSim软件对我国复杂地形地区的风能资源评估工作具有很好的参考和应用价值。     

复杂山地近地层强风特性分析

利用在贵州省西南部复杂山地上获取的近地层梯度风观测资料和三维超声测风仪观测资料,从中筛选出具有该地气候特征的强风样本,利用数量统计和谱分析等方法,计算分析了由于复杂地形影响而导致的局地低层强风的平均和脉动特征.平均风场主要表现在受当地主要特征地形(西北-东南走向的深切峡谷)影响,其全年的主导风向和最大风速出现的方向几乎完全转为沿峡谷走向,即使在符合中性大气层结条件下,风的垂直廓线也完全不满足幂指数分布,风攻角远远大于规范推荐的值,并且不同风向的强风攻角因地形影响其差异可达20°;强风条件下脉动风场的主要特征是:不同风向强风的湍流强度有所不同,在纵、横和垂直方向的湍流强度比值与现行设计规范给出的三维湍流强度比值有明显差异,其中垂直方向的湍流强度显著偏大是突出特征;湍流积分尺度偏大,其中纵向值偏大20%-60%,横向值在某些风向上可偏大3倍以上,垂直方向则普遍较平坦地形偏大一个量级左右;在桥梁结构较为敏感的频域范围内,各风向的湍流谱密度值有显著差异,其中不同风向在纵向上最大差值可达8倍,横向和垂直向可相差6倍,但无论哪个方向的湍流谱密度值均比台风中心要小1-2个量级.