排序方式： 共有8条查询结果，搜索用时 73 毫秒
1
1.
确定森林火烧迹地的准确时间点以及空间范围对于森林的受损评价、管理、碳核算以及森林恢复的管理有重要意义。由于森林火烧迹地在空间分布上具有一定的连续性,现有的森林火烧迹地提取方法大都采用先分类再后处理的两步处理策略来抑制虚警像素的影响。本文提出将时空检测方法Stacked ConvLSTM用于时间序列森林火烧迹地的检测,在保持结果具有较好空间连续性的基础上避免了具有主观性的后处理操作,实现端到端提取森林火烧迹地信息,提升了森林火烧迹地的提取精度。采用MODIS时间序列数据,基于2001年—2008年以及2001年—2016年的黑龙江沾河林业局伊南河林场和内蒙古自治区毕拉河林业局北大河林场两个区域的历史时间序列,分别对这两个区域2009年以及2017年发生的特大火灾区域进行火烧迹地检测,利用Stacked ConvLSTM、Stacked LSTM以及bfast算法在两个区域的MODIS时间序列中提取森林火烧迹地,并将火烧迹地检测结果与ESA发布的Fire_CCI 5.1火烧迹地产品进行对比分析。结果表明：首先,从目视效果来看,在研究区域Ⅰ,Stacked ConvLSTM检测的结果比Stacked LSTM和bfast算法错误检测点少,并且在空间分布也保持较高连续性;在研究区域Ⅱ,Stacked ConvLSTM检测到了较完整的火烧迹地区域。其次,在定量的精度评价指标上,在研究区域Ⅰ,Stacked ConvLSTM的精确度比Stacked LSTM和bfast算法分别高出0.120和0.405,并且召回率、准确度和F1-score也更高,Fire_CCI 5.1召回率虽更高,由于错检区域较大,其他精度指标远低于Stacked ConvLSTM;在研究区域Ⅱ,Stacked ConvLSTM精确度达0.924,召回率、准确度和F1-score相比Stacked LSTM和bfast算法以及Fire_CCI 5.1更高。 相似文献
2.
海面高度异常是反映海洋环境状况的主要变量之一。本文使用1993—2019年的融合月均海面高度异常数据,建立了基于深度学习的海面高度异常预测神经网络模型,提出了基于融合U型网络(U-Net)和卷积长短记忆网络(ConvLSTM)的中长期海面高度异常预报模型。在研究海域0.25°×0.25°的空间分辨率下,模型测试集预报结果的均方根误差和平均绝对误差分别为0.039 m和0.027 m,均优于全连接LSTM预报模型和ConvLSTM+CNN预报模型,为大中尺度的海面高度异常预报提供了新的方法。 相似文献
3.
Wetland ecosystems have experienced dramatic challenges in the past few decades due to natural and human factors. Wetland maps are essential for the conservation and management of terrestrial ecosystems. This study is to obtain an accurate wetland map using an object-based stacked generalization (Stacking) method on the basis of multi-temporal Sentinel-1 and Sentinel-2 data. Firstly, the Robust Adaptive Spatial Temporal Fusion Model (RASTFM) is used to get time series Sentinel-2 NDVI, from which the vegetation phenology variables are derived by the threshold method. Subsequently, both vertical transmit-vertical receive (VV) and vertical transmit-horizontal receive (VH) polarization backscatters (σ0 VV, σ0 VH) are obtained using the time series Sentinel-1 images. Speckle noise inherent in SAR data, resulting in over-segmentation or under-segmentation, can affect image segmentation and degrade the accuracies of wetland classification. Therefore, we segment Sentinel-2 multispectral images to delineate meaningful objects in this study. Then, in order to reduce data redundancy and computation time, we analyze the optimal feature combination using the Sentinel-2 multispectral images, Sentinel-2 NDVI time series, phenological variables and other vegetation index derived from Sentinel-2 multispectral images, as well as time series Sentinel-1 backscatters at the object level. Finally, the stacked generalization algorithm is utilized to extract the wetland information based on the optimal feature combination in the Dongting Lake wetland. The overall accuracy and Kappa coefficient of the object-based stacked generalization method are 92.46% and 0.92, which are 3.88% and 0.04 higher than that using the pixel-based method. Moreover, the object-based stacked generalization algorithm is superior to single classifiers in classifying vegetation of high heterogeneity areas. 相似文献
4.
风暴潮是指由强烈的大气扰动所导致的海面异常升高现象,由热带气旋引起的风暴潮常对沿海地区造成巨大的社会经济、人类活动和生命财产危害。依靠数据驱动的强非线性映射能力的机器学习方法较传统数值模式预报在耗费研究资源和计算时间上更具优势。本文选取广东省珠江口为研究区域,基于卷积长短时记忆网络（Convolutional LSTM network,ConvLSTM）机器学习算法展开风暴潮漫滩预报研究,利用由再分析资料驱动的数值模式产品构建了历史台风漫滩数据集,用于机器学习模型训练、验证和测试。研究了两种预报方式,一种是基于海表面高度场的自回归预报,另一种是依赖预报风场和初始海表面高度场进行的预报;它们可以实现基于数据驱动的风暴潮漫滩预报,其中自回归预报模型表现更优。相较于传统动力学数值预报,基于数据驱动的ConvLSTM预报模型结构更为轻便,所需驱动数据更少,在缺少边界条件、地形、径流等信号时,在短临预报中仍能基本复现数值模式模拟的结果。 相似文献
5.
渔场资源与位置的变动由空间与环境因子共同驱动,远洋渔场时空演变信息的精准预测是远洋捕捞的关键支撑。该研究考虑渔业生产统计数据,并兼顾同期海洋环境数据包括海表面温度（Sea surface temperature, SST）、海表面盐度（Sea surface salinity, SSS）、初级生产力（primary productivity, PP）和溶解氧浓度（dissolved oxygen concentration, O2）,提出了一种融合卷积长短期记忆网络（ConvLSTM）和卷积神经网络（CNN）的渔场时空分布预测模型。首先对时空因子进行编码,提取高层时空特征;其次采用CNN提取海洋环境变量的抽象特征,并基于ConvLSTM提取渔业数据的时空特征,最后融合高层时空关联信息对渔场时空演变趋势进行预测。以1995－2018年太平洋海域的延绳钓生产数据对模型进行验证,模型的根均方误差为0.1036,实验对比发现较传统渔场预报模型的预测误差降低15%~40%,预测的高产渔区与实际作业的高渔获量区匹配度高。该研究构建的渔场时空预测模型能够准确地预测出太平洋长鳍金枪鱼的时空分布,为太平洋长鳍金枪鱼的延绳钓渔业提供科学参考依据。 相似文献
6.
Dynamic analysis of stacked rigid blocks 总被引：1，自引：0，他引：1
Pol D. Spanos Panayiotis C. Roussis Nikolaos P. A. Politis 《Soil Dynamics and Earthquake Engineering》2001,21(7):472
The dynamic behavior of a structural model of two stacked rigid blocks subjected to ground excitation is examined. Assuming no sliding, the rocking response of the system standing free on a rigid foundation is investigated. The derivation of the equations of motion accounts for the consecutive transition from one pattern of motion to another, each being governed by a set of highly nonlinear differential equations. The system behavior is described in terms of four possible patterns of response and impact between either the two blocks or the base block and the ground. The equations governing the rocking response of the system to horizontal and vertical ground accelerations are derived for each pattern, and an impact model is developed by conservation of angular momentum considerations. Numerical results are obtained by developing an ad hoc computational scheme that is capable of determining the response of the system under an arbitrary base excitation. This feature is demonstrated by using accelerograms from the Northridge, CA, 1994, earthquake. It is hoped that the two-blocks model used herein can facilitate the development of more sophisticated multi-block structural models. 相似文献
7.
During March–April 2014 a series of earthquakes occurred around the Iquique city located in the northern Chile region. The two largest events of this sequence are the Mw8.2, April 1, 2014 and Mw7.7, April 4, 2014 quakes. Here we computed the nodal planes of eight of the large and well teleseismically recorded events of this series based on grid search, teleseismic moment tensors inversion, empirical Green's function deconvolution and its stack to average the deconvolutions for the Mw = 8.2, April 1, 2014, synthetic Green's function deconvolution and its stack to average the deconvolutions for the same event and 3D static deformation analysis of the above mentioned events based on the AK135 model. Grid search nodal planes and moment tensors suggest the dominance of reverse faulting. Almost all of the calculated teleseismic moment tensors represent a considerable amount of DC (usually more than 90%) and lower amount of CLVD for this sequence of events. Empirical and synthetic Green's function deconvolution showing down dip rupture propagation and 3D static deformation representing higher amount of vertical deformation in comparison with horizontal deformation components plus the existence of uplift and subsidence. According to the aftershocks distribution there is a bilateral distribution of the aftershocks around the first large event of this sequence that occurred March 16, 2014 (Mw6.7) so that they are approximately limited between the Mw8.2 (at north) and Mw7.7 (at south) quakes. Moreover there exist two bands of regional seismicity during early-mid 2014: a shallow off-shore band between the trench and coast and a deeper inland band under the active volcanic chain (both nearly parallel to the trench). 相似文献
8.
The correction of model forecast is an important step in evaluating weather forecast results. In recent years, post-processing models based on deep learning have become prominent. In this paper, a deep learning model named ED-ConvLSTM based on encoder-decoder structure and ConvLSTM is developed, which appears to be able to effectively correct numerical weather forecasts. Compared with traditional post-processing methods and convolutional neural networks, ED-ConvLSTM has strong collaborative extraction ability to effectively extract the temporal and spatial features of numerical weather forecasts and fit the complex nonlinear relationship between forecast field and observation field. In this paper, the post-processing method of ED-ConvLSTM for 2 m temperature prediction is tested using The International Grand Global Ensemble dataset and ERA5-Land data from the European Centre for Medium-Range Weather Forecasts (ECMWF). Root mean square error and temperature prediction accuracy are used as evaluation indexes to compare ED-ConvLSTM with the method of model output statistics, convolutional neural network postprocessing methods, and the original prediction by the ECMWF. The results show that the correction effect of ED-ConvLSTM is better than that of the other two postprocessing methods in terms of the two indexes, especially in the long forecast time. 相似文献
1