小波分解与Prophet框架融合的电离层VTEC预报模型

doi:10.12305/j.issn.1001-506X.2021.03.03

Abstract

Abstract:

In the application of global navigation satellite system (GNSS), the ionosphere vertical total electron content (VTEC) is an important parameter that directly determines the ionosphere delay error. In order to improve its short-term prediction accuracy, a time series prediction model fused with the wavelet decomposition and Prophet framework is proposed on the basis of considering the influence of geomagnetic disturbance. The comparative experiments are carried out based on the global ionosphere model (GIM) grid data. The indicator of root mean square error, mean absolute error and mean absolute percentage error are used to evaluate the prediction results and analyze the prediction residual. The results show that under different conditions (ionosphere quiet period and active period), the prediction accuracy of the model is higher, which is better than the unimproved Prophet frame, significantly better than the autoregressive integrated moving average (ARIMA) model, and has good applicability in middle and high latitudes.

Key words: ionosphere vertical total electron content, time series prediction, wavelet decomposition, Prophet framework, geomagnetic disturbance

CLC Number:

P288

Rui TIAN, Xurong DONG. Ionosphere VTEC prediction model fused with wavelet decomposition and Prophet framework[J]. Systems Engineering and Electronics, 2021, 43(3): 610-622.

Figures/Tables 15

Table 1

Fig.1

Table 2

Fig.2

Fig.3

Fig.4

Fig.5

Table 3

Table 4

Table 5

Fig.6

Fig.7

Table 6

Table 7

Fig.8

References 45

1	GOODMAN J M . Operational communication systems and relationships to the ionosphere and space weather[J]. Advances in Space Research, 2005, 36 (12): 2241- 2252. doi: 10.1016/j.asr.2003.05.063
2	GURNETT D A , HUFF R L , MORGAN D D , et al. An overview of radar soundings of the Martian ionosphere from the mars express spacecraft[J]. Advances in Space Research, 2008, 41 (9): 1335- 1346. doi: 10.1016/j.asr.2007.01.062
3	ZHAO Q L , WANG Y T , GU S F , et al. Refining ionospheric delay modeling for undifferenced and uncombined GNSS data processing[J]. Journal of Geodesy, 2019, 93 (4): 545- 560. doi: 10.1007/s00190-018-1180-9
4	KOUCKÁ K P , PODOLSKÁ K , POTUŽNÍKOVÁ K , et al. Evidence of vertical coupling: meteorological storm Fabienne on 23 September 2018 and its related effects observed up to the ionosphere[J]. Annales Geophysicae Discussions, 2020, 38 (1): 73- 93. doi: 10.5194/angeo-38-73-2020
5	FRON A, KRANKOWSKI A, HERNÁNDEZ-PAJARES M, et al. Reference total electron content: global 30-day average empirical VTEC maps for rapid assessment of weather-climate deviation[C]//Proc.of the EGU General Assembly Conference Abstracts, 2018: 17356.
6	LIU T , ZHANG B C , YUAN Y B , et al. Real-time precise point positioning (RTPPP) with raw observations and its application in real-time regional ionospheric VTEC modeling[J]. Journal of Geodesy, 2018, 92 (11): 1267- 1283. doi: 10.1007/s00190-018-1118-2
7	李志刚, 程宗颐, 冯初刚, 等. 电离层预报模型研究[J]. 地球物理学报, 2007, 50 (2): 327- 337. doi: 10.3321/j.issn:0001-5733.2007.02.001
	LI Z G , CHENG Z Y , FENG C G , et al. Research on ionospheric prediction models[J]. Chinese Journal of Geophysics, 2007, 50 (2): 327- 337. doi: 10.3321/j.issn:0001-5733.2007.02.001
8	范国清, 王威, 郗晓宁. 基于广义回归神经网络的电离层VTEC建模[J]. 测绘学报, 2010, 39 (1): 16- 21.
	FAN G Q , WANG W , XI X N . Modeling of ionospheric VTEC based on generalized regression neural network[J]. Journal of Surveying and Mapping, 2010, 39 (1): 16- 21.
9	RATNAM D V , DINESH B V , TEJASWI B , et al. TEC prediction model using neural networks over a low latitude GPS station[J]. International Journal of Soft Computing and Engineering, 2012, 2, 517- 521.
10	ACHARYA R , ROY B , SIVARAMAN M R , et al. Prediction of ionospheric total electron content using adaptive neural network with in-situ learning algorithm[J]. Advances in space research, 2011, 47 (1): 115- 123. doi: 10.1016/j.asr.2010.08.016
11	HABARULEMA J B , MCKINNELL L A , CILLIERS P J . Prediction of global positioning system total electron content using neural networks over South Africa[J]. Journal of Atmospheric and Solar-Terrestrial Physics, 2007, 69 (15): 1842- 1850. doi: 10.1016/j.jastp.2007.09.002
12	刘先冬, 宋力杰, 杨晓晖, 等. 基于小波神经网络的电离层TEC短期预报[J]. 海洋测绘, 2010, 30 (5): 49- 51. doi: 10.3969/j.issn.1671-3044.2010.05.014
	LIU X D , SONG L J , YANG X H , et al. Short-term prediction of ionosphere TEC based on wavelet neural network[J]. Ocean Surveying and Mapping, 2010, 30 (5): 49- 51. doi: 10.3969/j.issn.1671-3044.2010.05.014
13	ERDOGˇAN H , ARSLAN N . Identification of vertical total electron content by time series analysis[J]. Digital Signal Processing, 2009, 19 (4): 740- 749. doi: 10.1016/j.dsp.2008.07.002
14	LI X H, GUO D Z. Modeling and prediction of ionospheric total electron content by time series analysis[C]//Proc.of the 2nd International Conference on Advanced Computer Control, 2010: 375-379.
15	张小红, 任晓东, 吴风波, 等. 自回归移动平均模型的电离层总电子含量短期预报[J]. 测绘学报, 2013, 43 (2): 118- 124.
	ZHANG X H , REN X D , WU F B , et al. Short-term prediction of total ion content of ionosphere in autoregressive moving average model[J]. Journal of Surveying and Mapping, 2013, 43 (2): 118- 124.
16	李秀海, 郭达志. 应用半参数AR模型的电离层TEC建模与预测[J]. 测绘科学, 2011, 36 (2): 149- 151.
	LI X H , GUO D Z . Modeling and prediction of ionospheric TEC using semiparametric AR model[J]. Science of Surveying and Mapping, 2011, 36 (2): 149- 151.
17	汤俊, 姚宜斌, 陈鹏, 等. 利用EMD方法改进电离层TEC预报模型[J]. 武汉大学学报:信息科学版, 2013, 38 (4): 408- 411.
	TANG J , YAO Y B , CHEN P , et al. Improving ionospheric TEC prediction model using EMD method[J]. Journal of Wuhan University: Information Science Edition, 2013, 38 (4): 408- 411.
18	安家春, 宁新国, 王泽民, 等. 利用球冠谐函数和时间序列分析预报南极地区电离层[J]. 武汉大学学报:信息科学版, 2015, 40 (5): 677- 681.
	AN J C , NING X G , WANG Z M , et al. Prediction of the ionosphere in Antarctica using spherical crown harmonic function and time series analysis[J]. Wuhan University Journal: Information Science Edition, 2015, 40 (5): 677- 681.
19	田祥雨, 刘立龙, 杨可可, 等. BP神经网络和ARIMA模型的变权组合电离层TEC预报[J]. 桂林理工大学学报, 2019, 39 (4): 899- 904. doi: 10.3969/j.issn.1674-9057.2019.04.014
	TIAN X Y , LIU L L , YANG K K , et al. Variable weight combination ionospheric TEC prediction of BP neural network and ARIMA model[J]. Journal of Guilin University of Technology, 2019, 39 (4): 899- 904. doi: 10.3969/j.issn.1674-9057.2019.04.014
20	CIECKO A , GRUNWALD G . Klobuchar, NeQuick G, and EGNOS ionospheric models for GPS/EGNOS single-frequency positioning under 6-12 september 2017 space weather events[J]. Applied Sciences, 2020, 10 (5): 1553. doi: 10.3390/app10051553
21	陈亚楠, 徐继生. 顶部电离层离子密度经度结构的特征及其随季节、太阳活动和倾角的变化[J]. 地球物理学报, 2015, 58 (6): 1843- 1852.
	CHEN Y N , XU J S . Characteristics of the ionospheric longitude structure of the top ionosphere and its changes with season, solar activity and dip angle[J]. Chinese Journal of Geophysics, 2015, 58 (6): 1843- 1852.
22	翟笃林, 张学民, 熊攀, 等. Prophet时序预测模型在电离层TEC异常探测中的应用[J]. 地震, 2019, 39 (2): 46- 62.
	ZHAI D L , ZHANG X M , XIONG P , et al. Application of prophet time series prediction model in ionospheric TEC anomaly detection[J]. Earthquake, 2019, 39 (2): 46- 62.
23	TAYLOR S J , LETHAM B . Forecasting at scale[J]. The American Statistician, 2018, 72 (1): 37- 45. doi: 10.1080/00031305.2017.1380080
24	赵丹宁, 高蕊, 雷雨. 利用小波分解改进极移预报模型[J]. 武汉大学学报:信息科学版, 2019, 44 (12): 1797- 1801.
	ZHAO D N , GAO R , LEI Y . Improving polar motion forecasting model using wavelet decomposition[J]. Journal of Wuhan University: Information Science Edition, 2019, 44 (12): 1797- 1801.
25	李丽萍, 段桂华, 王建新. 基于Prophet框架的银行网点备付金预测方法[J]. 中南大学学报(自然科学版), 2019, 50 (1): 75- 82.
	LI L P , DUAN G H , WANG J X . Prediction method of bank branch reserve fund based on Prophet framework[J]. Journal of Central South University(Science and Technology), 2019, 50 (1): 75- 82.
26	DONG J C , DAI W , TANG L , et al. Why do EMD-based methods improve prediction? A multiscale complexity perspective[J]. Journal of Forecasting, 2019, 38 (7): 714- 731. doi: 10.1002/for.2593
27	BÜYÜKŞAHIN Ü Ç , ERTEKIN Ş . Improving forecasting accuracy of time series data using a new ARIMA-ANN hybrid method and empirical mode decomposition[J]. Neurocomputing, 2019, 361, 151- 163. doi: 10.1016/j.neucom.2019.05.099
28	AN X L , JIANG D X , LIU C , et al. Wind farm power prediction based on wavelet decomposition and chaotic time series[J]. Expert Systems with Applications, 2011, 38 (9): 11280- 11285. doi: 10.1016/j.eswa.2011.02.176
29	ZHANG K Y , GEN? AY R , YAZGAN M E . Application of wavelet decomposition in time-series forecasting[J]. Economics Letters, 2017, 158, 41- 46. doi: 10.1016/j.econlet.2017.06.010
30	刘立龙, 陈雨田, 黎峻宇, 等. 活跃期区域电离层总电子短期预报及适用性分析[J]. 武汉大学学报:信息科学版, 2019, 44 (12): 1757- 1764.
	LIU L L , CHEN Y T , LI J Y , et al. Short-term prediction and applicability analysis of total ionosphere in the active period[J]. Wuhan University Journal: Information Science Edition, 2019, 44 (12): 1757- 1764.
31	鲍亚东, 刘长建, 柴洪洲. 小波分解改进电离层VTEC时间序列预报模型[J]. 大地测量与地球动力学, 2015, 35 (5): 784- 787.
	BAO Y D , LIU C J , CHAI H Z . Wavelet decomposition improves ionospheric VTEC time series prediction model[J]. Geodesy and Geodynamics, 2015, 35 (5): 784- 787.
32	王秀杰, 封桂敏, 耿庆柱. 小波分析组合模型在日径流预测中的应用研究[J]. 自然资源学报, 2014, 29 (5): 885- 893.
	WANG X J , FENG G M , GENG Q Z . Application research of wavelet analysis combined model in daily runoff prediction[J]. Journal of Natural Resources, 2014, 29 (5): 885- 893.
33	周惠成, 彭勇. 基于小波分解的月径流预测校正模型研究[J]. 系统仿真学报, 2007, 19 (5): 1104- 1108. doi: 10.3969/j.issn.1004-731X.2007.05.042
	ZHOU H C , PENG Y . Research on prediction model of monthly runoff based on wavelet decomposition[J]. Journal of System Simulation, 2007, 19 (5): 1104- 1108. doi: 10.3969/j.issn.1004-731X.2007.05.042
34	张华, 陈小宏, 杨海燕. 地震信号去噪的最优小波基选取方法[J]. 石油地球物理勘探, 2011, 46 (1): 70- 171.70-75, 164, 170-171
	ZHANG H , CHEN X H , YANG H Y . Optimal wavelet base selection method for seismic signal denoising[J]. Petroleum Geophysical Prospecting, 2011, 46 (1): 70- 171.70-75, 164, 170-171
35	李彬, 夏虹. 基于最优小波基的主泵裂纹转子特征识别研究[J]. 振动与冲击, 2014, 33 (21): 207- 212.
	LI B , XIA H . Research on crack recognition of main pump crack rotor based on optimal wavelet basis[J]. Vibration and Shock, 2014, 33 (21): 207- 212.
36	李月琴, 栗苹, 闫晓鹏, 等. 无线电引信信号去噪的最优小波基选择[J]. 北京理工大学学报, 2008, 28 (8): 723- 726.
	LI Y Q , LI P , YAN X P , et al. Optimal wavelet basis selection for radio fuze signal denoising[J]. Journal of Beijing Institute of Technology, 2008, 28 (8): 723- 726.
37	TOUFIK B , MOKHTAR N . The wavelet transform for image processing applications[J]. Advances in Wavelet Theory and Their Applications in Engineering, Physics and Technology, 2012, (17): 395- 422.
38	孙延奎. 小波分析及其应用[M]. 北京: 机械工业出版社, 2005.
	SUN Y K . Wavelet analysis and its application[M]. Beijing: Machinery Industry Press, 2005.
39	AHUJA N , LERTRATTANAPANICH S , BOSE N K . Properties determining choice of mother wavelet[J]. IEEE Proceedings-vision, Image and Signal Processing, 2005, 152 (5): 659- 664. doi: 10.1049/ip-vis:20045034
40	SINGH V P, FIORENTINO M. A historical perspective of entropy applications in water resources[M]. Springer, Dordrecht, 1992.
41	桑燕芳, 王中根, 刘昌明. 水文时间序列分析方法研究进展[J]. 地理科学进展, 2013, 32 (1): 20- 30.
	SANG Y F , WANG Z G , LIU C M . Research progress of hydrological time series analysis methods[J]. Advances in Geography, 2013, 32 (1): 20- 30.
42	杨薛明, 边继飞, 朱霄珣, 等. 基于最大熵混沌时间序列的支持向量机短期风速预测模型研究[J]. 太阳能学报, 2016, 37 (9): 2173- 2179. doi: 10.3969/j.issn.0254-0096.2016.09.001
	YANG X M , BIAN J F , ZHU X X , et al. Research on support vector machine short-term wind speed prediction model based on maximum entropy chaotic time series[J]. Journal of Solar Energy, 2016, 37 (9): 2173- 2179. doi: 10.3969/j.issn.0254-0096.2016.09.001
43	SINGH V . Entropy-based parameter estimation in hydrology[M]. Berlin: Springer Science & Business Media, 1998.
44	孔豫龙.实时单点定位电离层延迟改正方法研究[D].郑州: 信息工程大学, 2018.
	KONG Y L. Research on ionospheric delay correction method for real-time single-point positioning[D]. Zhengzhou: Information Engineering University, 2018.
45	黄林峰, 黄江, 邓柏昌, 等. 利用单台地磁数据估算Ap和Kp指数的可行性分析[J]. 地震地磁观测与研究, 2011, 32 (3): 45- 51. doi: 10.3969/j.issn.1003-3246.2011.03.008
	HUANG L F , HUANG J , DENG B C , et al. Feasibility analysis of estimating Ap and Kp indices using single geomagnetic data[J]. Seismological and Geomagnetic Observation and Research, 2011, 32 (3): 45- 51. doi: 10.3969/j.issn.1003-3246.2011.03.008

小波基函数	正交性	双正交性	对称性	消失距	支撑长度
Haar	是	是	对称	1	1
Daubechies (db N)	是	是	近似对称	N	2N-1
Symlets (sym N)	是	是	近似对称	N	2N-1
Coiffets (Coif N)	是	是	近似对称	2N	6N-1
meyer	是	是	对称	无	无限
morlet	否	否	对称	无	无限

实验区域	纬度范围	经度范围
低纬度地区	2.5°N~17.5°N	75°E~125°E
中纬度地区	32.5°N~47.5°N	75°E~125°E
高纬度地区	62.5°N~77.5°N	75°E~125°E

参数	说明	设置值
τ	通过changepoint_prior_scale属性进行设置	活跃期低纬度	高频:0.50 低频:0.35
		活跃期中纬度	高频:0.50 低频:0.30
		活跃期高纬度	高频:0.05 低频:0.05
		平静期低纬度	高频:0.50 低频:0.05
		平静期中纬度	高频:0.05 低频:0.05
		平静期高纬度	高频:0.05 低频:0.03
增长模型	Linear(即分段线性模型)	通过Growth属性进行设置
N	3	通过add_seasonality方法中的fourier_order进行设置
σ	10	通过Seasonality_prior_scale属性进行设置
υ	10	通过holidays_prior_scale属性进行设置
参数估计方法	MAP	默认设置, 无需改动

预报模型	预报天数	低纬度地区			中纬度地区			高纬度地区
预报模型	预报天数	RMSE/TECu	MAE/TECu	MAPE/%	RMSE/TECu	MAE/TECu	MAPE/%	RMSE/TECu	MAE/TECu	MAPE/%
本文改进	1	2.047	1.575	11.32	0.673	0.566	7.75	0.620	0.523	16.97
	2	1.827	1.489	10.73	0.858	0.745	11.05	0.685	0.593	20.43
	3	2.272	1.789	13.38	1.095	0.978	13.70	1.069	0.953	40.44
Prophet	1	2.065	1.624	12.50	0.732	0.624	8.72	0.665	0.568	18.45
	2	1.827	1.510	11.92	0.973	0.854	12.71	0.804	0.711	24.81
	3	2.437	1.980	15.83	1.255	1.147	16.11	1.267	1.143	48.38
ARIMA	1	3.459	2.878	25.14	1.915	1.569	18.96	1.111	0.886	25.84
	2	3.927	3.328	28.64	2.120	1.839	23.45	1.446	1.233	36.73
	3	4.657	3.842	35.04	2.142	1.828	24.02	1.619	1.470	56.17

预报模型	预报天数	低纬度地区			中纬度地区			高纬度地区
预报模型	预报天数	RMSE/TECu	MAE/TECu	MAPE/%	RMSE/TECu	MAE/TECu	MAPE/%	RMSE/TECu	MAE/TECu	MAPE/%
本文改进	1	2.306	1.831	11.01	1.169	0.977	10.01	0.663	0.539	10.80
	2	3.232	2.620	15.58	1.086	0.919	8.65	0.782	0.608	11.08
	3	2.581	2.152	13.70	1.386	1.078	9.70	0.916	0.806	13.18
Prophet	1	2.342	1.864	11.43	1.310	1.121	11.62	0.678	0.554	11.17
	2	3.277	2.649	15.72	1.305	1.156	11.01	0.796	0.623	11.51
	3	2.600	2.136	13.63	1.651	1.359	12.84	0.921	0.809	13.38
ARIMA	1	5.581	4.700	30.55	2.628	2.278	23.39	1.782	1.522	29.75
	2	5.980	4.906	36.12	2.824	2.484	24.48	2.115	1.845	34.53
	3	5.898	5.004	34.79	3.826	3.296	31.67	2.345	1.979	32.24

Ionosphere VTEC prediction model fused with wavelet decomposition and Prophet framework

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 15

References 45

Related Articles 8

Recommended Articles

Metrics

Comments

预报模型	实验区域	Δ＜1 TECu			1 TECu≤Δ＜2 TECu			2 TECu≤Δ＜3 TECu			Δ≥3 TECu
预报模型	实验区域	第1日	第2日	第3日	第1日	第2日	第3日	第1日	第2日	第3日	第1日	第2日	第3日
本文改进模型	低纬度	42.59%	39.81%	43.06%	29.17%	32.41%	23.61%	16.21%	16.20%	19.43%	12.04%	11.58%	13.88%
	中纬度	84.26%	70.37%	51.38%	15.27%	28.23%	45.38%	0.46%	1.38%	3.23%	0.00%	0.00%	0.00%
	高纬度	88.20%	84.49%	58.57%	11.80%	15.51%	36.33%	0.00%	0.00%	5.09%	0.00%	0.00%	0.00%
Prophet	低纬度	40.27%	38.43%	35.19%	33.34%	35.64%	24.08%	13.89%	16.20%	24.52%	12.50%	9.72%	16.19%
	中纬度	79.17%	62.97%	39.83%	20.37%	32.87%	53.69%	0.46%	4.17%	6.02%	0.00%	0.00%	0.46%
	高纬度	86.58%	75.01%	45.84%	13.42%	24.99%	42.58%	0.00%	0.00%	11.57%	0.00%	0.00%	0.00%
ARIMA	低纬度	27.31%	19.45%	20.83%	19.92%	18.52%	14.81%	15.27%	16.20%	17.59%	37.49%	45.83%	46.75%
	中纬度	40.74%	25.92%	31.49%	29.63%	30.56%	26.39%	14.82%	27.77%	27.32%	14.82%	15.73%	14.81%
	高纬度	66.66%	48.16%	34.26%	25.01%	31.94%	39.36%	6.02%	14.36%	22.68%	2.31%	5.55%	3.70%

[1]	Chaowei SHI, Xiangru MENG, Qiaoyan KANG, Yuze SU. Virtual network topology reconfiguration approach based on hybrid traffic prediction [J]. Systems Engineering and Electronics, 2021, 43(5): 1382-1388.
[2]	Bohan WANG, Chao WU, Wenjun KE, Kaizhi ZHENG, Xiufeng FU, Shan JIANG. Intelligent planning method of combat mission based on probability graph [J]. Systems Engineering and Electronics, 2020, 42(12): 2795-2801.
[3]	LIU Lifang, YANG Haixia, QI Xiaogang. Time-frequency domain feature extraction algorithm based on linear discriminant analysis [J]. Systems Engineering and Electronics, 2019, 41(10): 2184-2190.
[4]	FAN Geng， MA Deng-wu， WU Ming-hui， MENG Shang. Condition time series prediction of electronic system based on optimized relevance vector machin [J]. Systems Engineering and Electronics, 2013, 35(9): 2011-2015.
[5]	LIU Yang, JIN Tian. Application of wavelet de noising in weak satellite signal acquisition [J]. Journal of Systems Engineering and Electronics, 2011, 33(4): 842-846.
[6]	LIU Ming-qian, LI Bing-bing, TANG Ning-jie. Blind ZP-OFDM time-parameters estimation in multipath channel [J]. Journal of Systems Engineering and Electronics, 2011, 33(11): 2531-2535.
[7]	ZHANG Lei， LI Xingshan， YU Jinsong， WAN Jiuqing. Fault prognostic algorithm based on relevance vector machine regression [J]. Journal of Systems Engineering and Electronics, 2010, 32(7): 1540-1543.
[8]	WANG Hai1，FAN Wen-jing1，XU Jian-hua2，ZHANG Chao2. Application of wavelet-ridge method to detect transient signals [J]. Journal of Systems Engineering and Electronics, 2010, 32(1): 5-8.

预报模型	实验区域	Δ＜1 TECu			1 TECu≤Δ＜2 TECu			2 TECu≤Δ＜3 TECu			Δ≥3 TECu
预报模型	实验区域	第1日	第2日	第3日	第1日	第2日	第3日	第1日	第2日	第3日	第1日	第2日	第3日
本文改进模型	低纬度	31.94%	23.62%	25.46%	36.11%	24.54%	29.62%	15.27%	18.06%	20.84%	16.67%	33.79%	24.08%
	中纬度	55.08%	59.26%	56.48%	34.72%	33.79%	34.26%	9.25%	6.93%	5.56%	0.92%	0.00%	3.70%
	高纬度	86.57%	87.04%	68.06%	13.43%	8.33%	28.71%	0.00%	4.63%	3.24%	0.00%	0.00%	0.00%
Prophet	低纬度	33.33%	24.53%	29.17%	32.88%	18.06%	23.14%	16.67%	22.68%	22.22%	17.13%	34.73%	25.47%
	中纬度	48.61%	48.14%	43.06%	39.34%	37.04%	39.81%	10.17%	13.42%	11.11%	1.84%	1.38%	6.02%
	高纬度	84.72%	86.11%	67.13%	15.28%	9.26%	29.63%	0.00%	4.63%	3.24%	0.00%	0.00%	0.00%
ARIMA	低纬度	20.38%	23.14%	14.35%	19.91%	11.11%	13.89%	10.64%	10.19%	16.67%	49.08%	55.55%	55.08%
	中纬度	22.22%	17.12%	17.59%	24.08%	22.23%	12.50%	25.47%	25.01%	16.19%	28.24%	35.64%	53.70%
	高纬度	40.28%	31.94%	32.42%	31.94%	29.16%	23.16%	14.82%	18.97%	22.68%	12.97%	19.91%	21.76%