人工神经网络和决策树模型在滑坡易发性分析中的性能对比   总被引：1，自引：0，他引：1  

田乃满  兰恒星  伍宇明  李郎平 《地球信息科学学报》2020,22(12):2304-2316

机器学习模型广泛应用于区域性滑坡易发性分析。模型的选择关系到评价结果的可信度、准确率和稳定性。现有滑坡易发性分析模型对比研究侧重模型的预测精度。模型的稳定性和数据量敏感性对机器学习模型的性能评估同样非常重要。本文以福建省南平市蔡源流域为研究区,以四川省绵阳市北川县为验证区,从预测精度、稳定性和数据量敏感性3个方面深入对比BP（Back Propagation）人工神经网络模型和CART（Classification and Regression Tree）决策树模型在滑坡易发性分析中的效果,主要结论如下:① 在逐渐增加一定数量训练样本的过程中,BP人工神经网络模型预测精度的增长率更高。在蔡源流域内,当训练样本数量增加10 000时,BP人工神经网络模型的预测精度上升5.22%,CART决策树模型的预测精度上升2.11%。② BP人工神经网络的预测精度高于CART决策树模型,且较为稳定。在100组数据集上,BP人工神经网络模型验证集预测精度的均值和验证集滑坡样本预测精度的均值分别为81.60%和84.86%,高于CART决策树模型的72.97%和76.59%。与此同时,BP人工神经网络模型对应预测精度的标准差分别是0.32%和0.37%,小于CART决策树模型的0.35%和0.67%。③ BP人工神经网络模型分析的滑坡易发区相比CART决策树模型,更接近实际滑坡的空间分布。最后,北川县的验证实验也出现了相同的现象。  相似文献   

与地震有关的水文及地球化学变化   总被引：1，自引：1，他引：1  

金继宇 《大地测量与地球动力学》2004,24(1):19-28

简要回顾了几十年来对地震发生前、地震过程中和震后地下流体和地球化学变化的研究和成果，这些研究一般都是以探索地震预报可能性为目的的。论述了与地震有关的地下水文及地球化学变化的机理，这些地下流体(包括地下水和气体诸如氢、氧和惰性气体)的起源和迁移流动现象以及详细介绍了早期和近代对有关地震的地下流体和地球化学变化的观测成果。同时指出了对地下流体和地球化学作为地震前兆来观测研究的困难所在以及为了克服这些困难而应该采取的地震前兆观测研究的方向，例如多种手段和多种原理方法，开发有效的地球物理和地球化学模型以及适当的数据分析统计方法等。  相似文献   

��ͬģ���º������ɶԸ߾���GPS������Ӱ��     

方川  高井祥  王坚  胡洪 《大地测量与地球动力学》2014,34(4):135-139

???й?????????IGS????,??????FES2004????????NAO99b?????????????????????????????????λ?????????????????????????????????λ????????cm??????????????U????????????????????????3~4????????????????????????????????mm????????????????????????С????????????????GPS????????????????10^-8??????????????????????????????????С??mm?????????????????????????????1 cm??  相似文献   

Floodplain hydrology of the Mekong Delta,Vietnam     

Nguyen Nghia Hung  José Miguel Delgado  Vo Khac Tri  Le Manh Hung  Bruno Merz  András Bárdossy  Heiko Apel 《水文研究》2012,26(5):674-686

The Mekong Delta is one of the largest and most intensively used estuaries in the world. Each year it witnesses widespread flooding which is both the basis of the livelihood for more than 17 million people but also the major hazard. Therefore, a thorough understanding of the hydrologic and hydraulic features is urgently required for various planning purposes. While the general causes and characteristics of the annual floods are understood, the inundation dynamics in the floodplains in Vietnam which are highly controlled by dikes and other control structures have not been investigated in depth. Especially, quantitative analyses are lacking, mainly due to scarce data about the inundation processes in the floodplains. Therefore, a comprehensive monitoring scheme for channel and floodplain inundation was established in a study area in the Plain of Reeds in the northeastern part of the Vietnamese Delta. This in situ data collection was complemented by a series of high‐resolution inundation maps derived from the TerraSAR‐X satellite for the flood seasons 2008 and 2009. Hence, the inundation dynamics in the channels and floodplains, and the interaction between channels and floodplains, could be quantified for the first time. The study identifies the strong human interference which is governed by flood protection levels, cropping patterns and communal water management. In addition, we examine the tidal influence on the inundation in various parts of the Delta, since it is expected that climate change‐induced sea level rise will increase the tidal contribution to floodplain inundation. Copyright © 2011 John Wiley & Sons, Ltd.  相似文献   

Ensemble rainfall forecasting with numerical weather prediction and radar‐based nowcasting models     

Shan He  Srivatsan V. Raghavan  Ngoc Son Nguyen  Shie‐Yui Liong 《水文研究》2013,27(11):1560-1571

The overall objective of this study is to improve the forecasting accuracy of the precipitation in the Singapore region by means of both rainfall forecasting and nowcasting. Numerical Weather Predication (NWP) and radar‐based rainfall nowcasting are two important sources for quantitative precipitation forecast. In this paper, an attempt to combine rainfall prediction from a high‐resolution mesoscale weather model and a radar‐based rainfall model was performed. Two rainfall forecasting methods were selected and examined: (i) the weather research and forecasting model (WRF); and (ii) a translation model (TM). The WRF model, at a high spatial resolution, was run over the domain of interest using the Global Forecast System data as initializing fields. Some heavy rainfall events were selected from data record and used to test the forecast capability of WRF and TM. Results obtained from TM and WRF were then combined together to form an ensemble rainfall forecasting model, by assigning weights of 0.7 and 0.3 weights to TM and WRF, respectively. This paper presented results from WRF and TM, and the resulting ensemble rainfall forecasting; comparisons with station data were conducted as well. It was shown that results from WRF are very useful as advisory of anticipated heavy rainfall events, whereas those from TM, which used information of rain cells already appearing on the radar screen, were more accurate for rainfall nowcasting as expected. The ensemble rainfall forecasting compares reasonably well with the station observation data. Copyright © 2012 John Wiley & Sons, Ltd.  相似文献   

Oort Cloud Comet Perihelion Asymmetries: Galactic Tide, Shower or Observational Bias?     

John J. Matese  Patrick G. Whitman  Daniel P. Whitmire 《Celestial Mechanics and Dynamical Astronomy》1997,69(1-2):77-87

We investigate the distribution of Oort cloud comet perihelia. The data considered includes comets having orbital elements of the two highest quality classes with original energies designated as new or young. Perihelion directions are determined in galactic, ecliptic and geocentric equatorial coordinates. Asymmetries are detected in the scatter and are studied statistically for evidence of adiabatic galactic tidal dynamics, an impulse-induced shower and observational bias. The only bias detected is the well-known deficiency of observations with perihelion distances q > 2.5 AU. There is no significant evidence of a seasonal dependence. Nor is there a substantive hemispherical bias in either ecliptic or equatorial coordinates. There is evidence for a weak stellar shower previously detected by Biermann which accounts for ≈ 10% of the total observations. Both the q bias and the Biermann star track serve to weaken the evidence for a galactic tidal imprint. Nevertheless, statistically significant asymmetries in galactic latitude and longitude of perihelia remain. A latitude asymmetry is produced by a dominant tidal component perpendicular to the galactic disk. The longitude signal implies that ≈ 20% of new comets need an additional dynamical mechanism. Known disk non-uniformities and an hypothetical bound perturber are discussed as potential explanations. We conclude that the detected dynamical signature of the galactic tide is real and is not an artifact of observational bias, impulsive showers or poor data. This revised version was published online in July 2006 with corrections to the Cover Date.  相似文献   

Are sea-level-rise trends along the coasts of the north Indian Ocean consistent with global estimates?   总被引：2，自引：0，他引：2  

A.S. Unnikrishnan  D. Shankar 《Global and Planetary Change》2007,57(3-4):301-307

Mean-sea-level data from coastal tide gauges in the north Indian Ocean were used to show that low-frequency variability is consistent among the stations in the basin. Statistically significant trends obtained from records longer than 40 years yielded sea-level-rise estimates between 1.06–1.75 mm yr^− 1, with a regional average of 1.29 mm yr^− 1, when corrected for global isostatic adjustment (GIA) using model data. These estimates are consistent with the 1–2 mm yr^− 1 global sea-level-rise estimates reported by the Intergovernmental Panel on Climate Change.  相似文献   

The R Coronae Borealis Connection     

Geoffrey C. Clayton 《Astrophysics and Space Science》2002,279(1-2):167-170

The R Coronae Borealis (RCB) stars are rare hydrogen-deficient carbon-rich supergiants which undergo spectaculardeclines in brightness of up to 8 magnitudes at irregular intervals as dust forms along the line of sight.Understanding the RCB stars is a key test for any theory whichaims to explain hydrogen deficiency in post-Asymptotic Giant Branch (AGB) stars. There are two major evolutionary models for the origin of RCB stars: the Double Degenerate and the Final Helium ShellFlash. In the final flashmodel, there is a close relationship between RCB stars and Planetary Nebulae (PNe). The connection between RCB stars and PNe has recentlybecome stronger, since the central stars of three old PNe (Sakurai's Object, V605 Aql and FG Sge) have had observedoutbursts that transformed them from hot evolved central stars into cool giants with the spectral properties of an RCB star.  相似文献   

水污染物浓度的神经网络预测模式及效果检验     

于扬  薛丽梅  聂伊辰 《成都信息工程学院学报》2004,19(1):100-103

污染物浓度预测是环境保护的重要内容,将神经网络用于水中有机污染物浓度的预测并对效果进行检验.结果表明,预测值与观测值符合较好.  相似文献   

成矿定量预测与深部找矿   总被引：19，自引：1，他引：18  

赵鹏大 《地学前缘》2007,14(5):1-10

成矿预测是在不确定条件下制定最优决策的工作。成矿预测作为一种地质系统,与其他技术、经济系统存在重要区别。由于矿床类型的多样性,矿床成因的复杂性,控矿因素的隐蔽性和找矿信息的多解性,成矿预测结果具有不确定性并常常因人而异。探索成矿预测过程客观化、定量化和精确化一直是成矿预测学的前沿课题。文中以个旧锡矿为例展示致矿地质异常与矿体空间产出及分布的密切关系,强调以"求异"准则为指导的成矿定量预测的重要性。当今地质勘查工作面临深部找矿问题,论文从深部找矿的概念、类型、目标、效益等方面简要介绍了国外理论研究和找矿实践概况。强调深部找矿中要加强地壳深部结构的研究,要重视深部找矿的经济"回报率"和勘查项目的"转化率"的重要性。  相似文献