|
|
- 2018
基于深度学习的短时强降水天气识别
|
Abstract:
气象预报人员面临的问题之一是如何准确有效地识别短时强降水天气．短时强降水是一种主要由强对流天气形成的气象灾害，产生原因与空气湿度、大气中的水分以及温湿等物理量参数有关，由此提出基于物理量参数和深度学习模型DBNs的短时强降水天气识别模型．首先，利用SMOTE算法人工合成短时强降水少数类(相对于非短时强降水天气类)样本，调整原始数据集不均衡分布问题；然后通过含有高斯玻耳兹曼机的深度学习模型对地面大气监测站逐小时加密的观测量，以及常用于天气预报分析的物理量等低层特征构造出抽象的高层特征，发现数据特征内在关系；最后实现了DBNs短时强降水的自动识别模型．结果表明，该方法能够较为准确地识别短时强降水，对于短时强降水的命中率、误警率和临界成功指数，都有着较好的表现．
One of the key studies for meteorological practitioners is how to recognize and predict short-time heavy rainfall accurately and effectively. The short-time heavy rainfall is a severe meteorological disaster that is mainly caused by strong convective weather, which is related to such physical parameters as air humidity, moisture in the atmosphere, temperature and humidity. In this paper, a recognition model of the short-time heavy rainfall based on physical parameters and deep learning model DBNs is constructed. Firstly, SMOTE algorithm is used to synthesize a few samples of the short-time heavy rainfall, which is much less than normal weather, to adjust the distribution of the original data set. Secondly, a deep learning model with a Gaussian Boltzmann machine is constructed based on the observed data from automatic monitoring stations on a local ground and the physical quantities commonly used in weather forecast analysis. Finally, the automatic recognition model of short-term heavy rainfall is obtained. Through the analysis of the experimental results, the model can accurately recognize the short-time heavy rainfall, and have a good performance in the POD, FAR and CSI of short-time heavy rainfall recognition
| [1] | Arel I, Rose D C, Karnowski T P. Deep machine learning―A new frontier in artificial intelligence research [research frontier][J]. <i>IEEE Computational Intelligence Magazine</i>, 2010, 5(4):13-18. |
| [2] | Yin Jian, Lu Chengmin, Yang Guijun. Combinations of discriminatory analysis and logistic regression for classification[J]. <i>Journal of Applied Statistics and Management</i>, 2014, 33(2):256-265(in Chinese). |
| [3] | 刘健文. 天气分析预报物理量计算基础[M]. 北京:气象出版社, 2005. |
| [4] | Liu Jianwen. <i>Calculation of Physical Quantities Based on Weather Forecast</i>[M]. Beijing:China Meteorological Press, 2005(in Chinese). |
| [5] | 杨景梅, 邱金桓. 用地面湿度参量计算我国整层大气可降水量及有效水汽含量方法的研究[J]. 大气科学, 2002, 26(1):9-22. |
| [6] | Yang Jingmei, Qiu Jinhuan. A method for estimating perceptible water and water vapor content from ground humidity parameters[J]. <i>Chinese Journal of Atmospheric Sciences</i>, 2002, 26(1):9-22(in Chinese). |
| [7] | Luino F. Sequence of instability processes triggered by heavy rainfall in the Northern Italy[J]. <i>Geomorphology</i>, 2005, 66(1/2/3/4):13-39. |
| [8] | You C H, Lee D I, Kang M Y, et al. Classification of rain types using drop size distributions and polarimetric radar:Case study of a 2014 flooding event in Korea[J]. <i>Atmospheric Research</i>, 2016, 181:211-219. |
| [9] | Root B, Yu T Y, Yeary M. Consistent clustering of radar reflectivities using strong point analysis:A prelude to storm tracking[J]. <i>Geoscience and Remote Sensing Letters IEEE</i>, 2011, 8:273-277. |
| [10] | 胡文东, 杨侃, 黄小玉, 等. 一次阵风锋触发强对流过程雷达资料特征分析[J]. 高原气象, 2015, 34(5):1452-1464. |
| [11] | Hu Wendong, Yang Kan, Huang Xiaoyu, et al. Analysis on a severe convection triggered by gust front in Yinchuan with radar data[J]. <i>Plateau Meteorology</i>, 2015, 34(5):1452-1464(in Chinese). |
| [12] | Hinton G, Osindero S, Teh Y. A fast learning algorithm for deep belief nets[J]. <i>Neural Computation</i>, 1989, 18(7):1527-1554. |
| [13] | Hinton G E. A practical guide to training restricted Boltzmann machines[J]. <i>Momentum</i>, 2012, 9(1):599-619. |
| [14] | Yamashita T, Tanaka M, Yoshida E, et al. To be Bernoulli or to be Gaussian, for a restricted Boltzmann machine[C]//<i>International Conference on Pattern Recognition</i>, <i>IEEE Computer Society</i>. Sweden, 2014:1520-1525. |
| [15] | Hinton G E. Training products of experts by minimizing contrastive divergence[J]. <i>Neural Computation</i>, 2002, 14(8):1771-1800. |
| [16] | Karakida R, Okada M, Amari S I. Dynamical analysis of contrastive divergence learning:Restricted Boltzmann machines with Gaussian visible units[J]. <i>Neural Networks</i>, 2016, 79(C):78-87. |
| [17] | Hinton G E, Dayan P, Frey B J, et al. The "wake-sleep" algorithm for unsupervised neural networks[J]. <i>Science</i>, 1995, 268(268):1158-1161. |
| [18] | 路志英, 刘海, 贾惠珍, 等. 基于雷达反射率图像特征的冰雹暴雨识别[J]. 物理学报, 2014, 63(18):485-496. |
| [19] | Lu Zhiying, Liu Hai, Jia Huizhen, et al. Recognition of hail and rainstorm based on the radar reflectivity image features[J]. <i>Acta Physica Sinica</i>, 2014, 63(18):485-496(in Chinese). |
| [20] | Jeon S S. Damage pattern recognition of spatially distributed slope damages and rainfall using optimal GIS mesh dimensions[J]. <i>Journal of Mountain Science</i>, 2014, 11(2):336-344. |
| [21] | Chawla N V, Bowyer K W, Hall L O, et al. SMOTE:Synthetic minority over-sampling technique [J]. <i>Journal of Artificial Intelligence Research</i>, 2002, 16(1):321-357. |
| [22] | Hinton G E, Salakhutdinov R R. Reducing the dimensionality of data with neural networks[J]. <i>Science</i>, 2006, 313(5786):504. |
| [23] | Chabiron O, Malgouyres F, Tourneret J Y, et al. Toward fast transform learning[J]. <i>International Journal of Computer Vision</i>, 2015, 114(2/3):195-216. |
| [24] | Roux N L, Bengio Y. Representational power of restricted Boltzmann machines and deep belief networks [J]. <i>Neural Computation</i>, 2008, 20(6):1631-1632. |
| [25] | Pelillo M, Refice M. Learning compatibility coefficients for relaxation labeling processes[J]. <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>, 1994, 16(9):933-945. |
| [26] | Najafabadi M M, Villanustre F, Khoshgoftaar T M, et al. Deep learning applications and challenges in big data analytics[J]. <i>Journal of Big Data</i>, 2015, 2(1):1-10. |
| [27] | 张家国, 王珏, 黄治勇, 等. 几类区域性暴雨雷达回波模型[J]. 气象, 2011, 37(3):285-290. |
| [28] | Zhang Jiaguo, Wang Jue, Huang Zhiyong, et al. Several kinds of region rainstorm radar echo models[J]. <i>Meteorological Monthly</i>, 2011, 37(3):285-290(in Chinese). |
| [29] | Kalal Z, Mikolajczyk K, Matas J. Tracking-learning-detection[J]. <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>, 2012, 34(7):1409-1422. |
| [30] | Wan J, Liu J, Ren G, et al. Day-ahead prediction of wind speed with deep feature learning[J]. <i>International Journal of Pattern Recognition and Artificial Intelligence</i>, 2016, 30(5):433-438. |
| [31] | Zhang C Y, Chen C L P, Gan M, et al. Predictive deep Boltzmann machine for multipored wind speed forecasting[J]. <i>IEEE Transactions on Sustainable Energy</i>, 2015, 6(4):1416-1425. |
| [32] | 尹剑, 陆程敏, 杨贵军. 判别分析与Logistic回归组合分类[J]. 数理统计与管理, 2014, 33(2):256-265. |
| [33] | 李超, 魏合理, 刘厚通, 等. 合肥整层大气可降水量与地面露点相关性分析[J]. 高原气象, 2009, 28(2):452-457. |
| [34] | Li Chao, Wei Heli, Liu Houtong, et al. Correlation analyses on total perceptible water and surface dew point temperature over Hefei[J]. <i>Plateau Meteorology</i>, 2009, 28(2):452-457(in Chinese). |
