中国气象干旱时空特征与混合模型预测

doi:10.12118/j.issn.1000-6060.2025.445

干旱区地理 ›› 2026, Vol. 49 ›› Issue (5): 881-893.doi: 10.12118/j.issn.1000-6060.2025.445 cstr: 32274.14.ALG2025445

中国气象干旱时空特征与混合模型预测

刘洋洋¹(), 毛克彪²(), 郭中华¹, 袁紫晋²

¹ 宁夏大学电子与电气工程学院，宁夏银川 750021
² 中国农业科学院农业资源与农业区划研究所，北方干旱半干旱耕地高效利用全国重点实验室，北京 100081

收稿日期:2025-07-14 修回日期:2025-09-18 出版日期:2026-05-25 发布日期:2026-05-25
通讯作者: 毛克彪（1977-），男，博士，研究员，主要从事交叉学科研究和推动人工智能在地学和农学中的应用研究. E-mail: maokebiao@126.com
作者简介:刘洋洋（2001-），女，硕士研究生，主要从事气象干旱与深度学习的交叉应用研究. E-mail: 12023130610@stu.nxu.edu.cn
基金资助:
中央级公益性科研院所基本科研业务费专项(Y2025YC86);宁夏回族自治区科技厅自然科学基金重点项目(2024AC02032)

Spatiotemporal characteristics and hybrid model prediction of meteorological drought in China

LIU Yangyang¹(), MAO Kebiao²(), GUO Zhonghua¹, YUAN Zijin²

¹ School of Electronic and Electrical Engineering, Ningxia University, Yinchuan 750021, Ningxia, China
² Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, National Key Laboratory for Efficient Utilization of Arid and Semi-Arid Cultivated Land in Northern China, Beijing 100081, China

Received:2025-07-14 Revised:2025-09-18 Published:2026-05-25 Online:2026-05-25

摘要/Abstract

摘要：

为提升干旱预测精度，基于1980—2023年中国地面气象观测数据，选取标准化降水蒸散发指数（SPEI）作为干旱指标，通过相关性分析筛选预测因子，采用Theil-Sen Median趋势分析等方法，构建小波变换与长短期记忆神经网络结合的混合模型（WT-LSTM），设计单因子和多因子2种预测方案，分析中国气象干旱的时空演变特征。结果表明：（1）各因子的空间趋势分布不均匀，降水和潜在蒸散发整体为增加趋势，降水从东南向西北呈“增-减-增-减”的特征，潜在蒸散发在空间上从西北向东南递增，中国88.87%的区域年SPEI趋势系数小于0·a^-1，干旱化趋势普遍。（2）中国年均干旱持续时间大部分在1~2个月左右，干旱严重程度显著增加的区域主要在西北、华北、东北的北部，干旱特征趋势呈现出北部高、南部低的空间分布格局。（3）各季节干旱持续时间长的地区对应的干旱强度并不高，其中夏季干旱频率的高值区分布较为广泛，冬季的干旱频率最低。（4）相比LSTM，混合模型WT-LSTM的性能更优，而对于单因子，多因子增强了模型对复杂干旱模式的表征能力，显著提高了模型的预测效果。（5）混合模型下的单因子预测更适合气候干旱模式较稳定的区域，而多因子在新疆维吾尔自治区、青藏高原等气候复杂区对干旱趋势的捕捉能力更强。

关键词: 干旱预测, 标准化降水蒸散发指数, 离散小波变换, 长短期记忆神经网络

Abstract:

To enhance drought prediction accuracy, this study uses Chinese ground meteorological observations from 1980 to 2023 and selects the standardized precipitation evapotranspiration index (SPEI) as the drought indicator. Key predictors were identified through correlation analysis, and a hybrid wavelet transform-long short-term memory (WT-LSTM) model was developed using Theil-Sen Median trend analysis and related methods. Two prediction schemes—single-factor and multifactor—were designed to analyze the spatiotemporal evolution of meteorological drought in China. Results show (1) Spatial trends of factors are uneven; precipitation and potential evapotranspiration generally increase, with precipitation showing an “increase-decrease-increase-decrease” pattern from southeast to northwest, and potential evapotranspiration increasing from northwest to southeast. SPEI trends are negative in 88.87% of areas, indicating widespread drought intensification. (2) The average annual drought duration is mostly 1-2 months, with significant increases in average annual drought severity mainly in northwest, north, and northern northeast China. Trends in average annual drought characteristics exhibit a spatial pattern of higher values in the north and lower values in the south. (3) Regions with long seasonal drought durations in each season do not correspond to high drought intensity; high-value areas of summer drought frequency are widely distributed, while winter drought frequency is lowest. (4) Compared with LSTM, the WT-LSTM model performs better, and for single-factor predictions, the multi-factor approach enhances the ability of the model to represent complex drought patterns, significantly improving prediction performance. (5) Under the hybrid model, single-factor prediction is more suitable for regions with relatively stable climatic drought patterns, while multifactor prediction better captures drought trends in climatically complex areas such as Xinjiang Uygur Autonomous Region and the Qinghai-Xizang Plateau.

Key words: drought forecasting, standardized precipitation evapotranspiration index, discrete wavelet transform, long short-term memory neural network

刘洋洋, 毛克彪, 郭中华, 袁紫晋. 中国气象干旱时空特征与混合模型预测[J]. 干旱区地理, 2026, 49(5): 881-893.

LIU Yangyang, MAO Kebiao, GUO Zhonghua, YUAN Zijin. Spatiotemporal characteristics and hybrid model prediction of meteorological drought in China[J]. Arid Land Geography, 2026, 49(5): 881-893.

图/表 12

表1

表2

图1

图2

图3

图4

表3

表4

图5

图6

图7

图8

参考文献 36

[1]	管晓丹, 石瑞, 孔祥宁, 等. 全球变化背景下半干旱区陆气机制研究综述[J]. 地球科学进展, 2018, 33(10): 995-1004. doi: 10.11867/j.issn.1001-8166.2018.10.0995.
	[Guan Xiaodan, Shi Rui, Kong Xiangning, et al. An overview of researches on land-atmosphere interaction over semi-arid region under global changes[J]. Advances in Earth Science, 2018, 33(10): 995-1004.] doi: 10.11867/j.issn.1001-8166.2018.10.0995.
[2]	Yu M X, Li Q F, Hayes M J, et al. Are droughts becoming more frequent or severe in China based on the standardized precipitation evapotranspiration index: 1951—2010?[J]. International Journal of Climatology, 2014, 34(3): 545-558. doi: 10.1002/joc.2014.34.issue-3
[3]	Palmer W C. Meteorological drought[M]. Washington: US Department of Commerce, Weather Bureau, 1965.
[4]	Guttman N B, Wallis J R, Hosking J R M. Spatial comparability of the Palmer drought severity index[J]. Journal of the American Water Resources Association, 1992, 28(6): 1111-1119. doi: 10.1111/jawr.1992.28.issue-6
[5]	Yan G X, Lu G H, Wu Z Y, et al. Study on integrated meteorological drought index based on PDSI and SPI[J]. Water Resources and Hydropower Engineering, 2009, 40(4): 10-13.
[6]	McKee T B, Doesken N J, Kleist J. The relationship of drought frequency and duration to time scales[C]// American Meteorological Society. Proceedings of the 8^th Conference on Applied Climatology. Anaheim, CA: American Meteorological Society, 1993: 179-183.
[7]	Vicente-Serrano M S, Beguería S, Lorenzo-Lacruz J, et al. Performance of drought indices for ecological, agricultural, and hydrological applications[J]. Earth Interactions, 2012, 16(10): 1-27.
[8]	Chen H P, Sun J Q. Changes in drought characteristics over China using the standardized precipitation evapotranspiration index[J]. Journal of Climate, 2015, 28(13): 5430-5447. doi: 10.1175/JCLI-D-14-00707.1
[9]	Allen R G, Pereira L S, Raes D, et al. Crop evapotranspiration-guidelines for computing crop water requirements[M]. Rome: Food and Agriculture Organization of the United Nations, 1998: 15-86.
[10]	Gilbert H, Jay P, Ashutosh S, et al. Evaluation of hybrid wavelet models for regional drought forecasting[J]. Remote Sensing, 2022, 14(24): 6381, doi: 10.3390/RS14246381.
[11]	Lecun Y, Bengio Y, Hinton G. Deep learning[J]. Nature, 2015, 521(7553): 436-444. doi: 10.1038/nature14539
[12]	Reza R, Ani S. Integrating wavelet transform and support vector machine for improved drought forecasting based on standardized precipitation index[J]. Journal of Hydroinformatics, 2025, 27(2): 320-337. doi: 10.2166/hydro.2025.292
[13]	Khan M M H, Muhammad N S, El-Shafie A. Wavelet based hybrid ANN-ARIMA models for meteorological drought forecasting[J]. Journal of Hydrology, 2020, 590: 125380, doi: 10.1016/j.jhydrol.2020.125380.
[14]	Vicente-Serrano S M, Beguería S, López-Moreno J I. A multiscalar drought index sensitive to global warming: The standardized precipitation evapotranspiration index[J]. Journal of Climate, 2010, 23(7): 1696-1718. doi: 10.1175/2009JCLI2909.1
[15]	李峥嵘, 彭涛, 林青霞, 等. 三峡水库影响下长江中下游水文干旱演变及对气象干旱的响应[J]. 湖泊科学, 2022, 34(5): 1683-1696.
	[Li Rongzheng, Peng Tao, Lin Qingxia, et al. Hydrological drought evolution and its response to meteorological drought in the middle-lower reach of the Yangtze River triggered by Three Gorges Reservoir[J]. Journal of Lake Sciences, 2022, 34(5): 1683-1696.] doi: 10.18307/2022.0521
[16]	Yevjevich V. An objective approach to definitions and investigations of continental hydrologic droughts[J]. Journal of Hydrology, 1969, 7(3): 353, doi: 10.1016/0022-1694(69)90110-3.
[17]	Wu J F, Chen X H. Spatiotemporal trends of dryness/wetness duration and severity: The respective contribution of precipitation and temperature[J]. Atmospheric Research, 2019, 216: 176-185. doi: 10.1016/j.atmosres.2018.10.005
[18]	Qi Z, Chi Y M, Gou J J, et al. Spatiotemporal characteristics and forecasting of short-term meteorological drought in China[J]. Journal of Hydrology, 2023, 624: 129924, doi: 10.1016/j.jhydrol.2023.129924.
[19]	Xu K, Yang D W, Yang H B, et al. Spatio-temporal variation of drought in China during 1961—2012: A climatic perspective[J]. Journal of Hydrology, 2015, 526: 253-264. doi: 10.1016/j.jhydrol.2014.09.047
[20]	Cohen J, Cohen P, West S G, et al. Applied multiple regression/correlation analysis for the behavioral sciences[M]. New York: Routledge, 2013.
[21]	Rodgers J L, Nicewander W A. Thirteen ways to look at the correlation coefficient[J]. The American Statistician, 1988, 42(1): 59-66.
[22]	Theil H. A rank-invariant method of linear and polynomial regression analysis, Part 2[J]. Proceedings of the Royal Netherlands Academy of Sciences, 1950, 53: 597-604.
[23]	Sen P K. Estimates of the regression coefficient based on Kendall’s Tau[J]. Journal of the American Statistical Association, 1968, 63(324): 1379-1389. doi: 10.1080/01621459.1968.10480934
[24]	Gocic M, Trajkovic S. Analysis of precipitation and drought data in Serbia over the period 1980—2010[J]. Journal of Hydrology, 2013, 494: 32-42. doi: 10.1016/j.jhydrol.2013.04.044
[25]	Wu J W, Miao C Y, Zheng H Y, et al. Meteorological and hydrological drought on the Loess Plateau, China: Evolutionary characteristics, impact, and propagation[J]. Journal of Geophysical Research: Atmospheres, 2018, 123(20): 11569-11584.
[26]	Mallat S G. A theory for multiresolution signal decomposition: The wavelet representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 11(7): 674-693. doi: 10.1109/34.192463
[27]	Addison P S. The illustrated wavelet transform handbook: Introductory theory and applications in science, engineering, medicine and finance[M]. Boca Raton: CRC Press, 2017.
[28]	徐东坡, 李金明, 周祖昊, 等. 1956—2018年中国降水特征的时空分布规律研究[J]. 水利水电技术, 2020, 51(10): 20-27.
	[Xu Dongpo, Li Jinming, Zhou Zuhao, et al. Study on the spatial and temporal distribution of precipitation characteristics in China from 1956 to 2018[J]. Water Resources and Hydropower Engineering, 2020, 51(10): 20-27.]
[29]	魏涛, 王云权. 基于PML-V2数据集的中国蒸散发时空动态特征分析[J]. 干旱区地理, 2023, 46(6): 857-867. doi: 10.12118/j.issn.1000-6060.2022.476
	[Wei Tao, Wang Yunquan. Temporal and spatial dynamic analysis of terrestrial evapotranspiration in China based on PML-V2 product[J]. Arid Land Geography, 2023, 46(6): 857-867.] doi: 10.12118/j.issn.1000-6060.2022.476
[30]	史继清, 边多, 杨霏云, 等. 西藏地区潜在蒸散量变化特征及灰色模型预测初探[J]. 干旱区地理, 2021, 44(6): 1570-1579. doi: 10.12118/j.issn.1000–6060.2021.06.06
	[Shi Jiqing, Bian Duo, Yang Feiyun, et al. Variation characteristics of potential evapotranspiration and the forecast of grey model in Xizang[J]. Arid Land Geography, 2021, 44(6): 1570-1579.] doi: 10.12118/j.issn.1000–6060.2021.06.06
[31]	杨永红, 张展羽, 阮新建. 西藏参考作物蒸发蒸腾量的时空变异规律[J]. 水科学进展, 2009, 20(6): 775-781.
	[Yang Yonghong, Zhang Zhanyu, Ruan Xinjian. Temporal and spatial variation law of reference crop evapotranspiration in Xizang[J]. Advances in Water Science, 2009, 20(6): 775-781.]
[32]	宋文起, 王志恒. 近30年来中国气温、降水和水分盈亏的趋势及其变化[J]. 气候与环境研究, 2023, 28(1): 1-16.
	[Song Wenqi, Wang Zhiheng. Trends and changes in temperature, precipitation, and water surplus and deficit in China over the last 30 years[J]. Climatic and Environmental Research, 2023, 28(1): 1-16.]
[33]	易科帆, 吝海霞, 秦国鹏, 等. 基于像元化SPEI的新疆气象干旱时空变化特征[J]. 中国农业气象, 2025, 46(7): 1026-1038.
	[Yi Kefan, Lin Haixia, Qin Guopeng, et al. Spatiotemporal variation characteristics of meteorological drought in Xinjiang based on pixel-scale SPEI[J]. Chinese Journal of Agrometeorology, 2025, 46(7): 1026-1038.]
[34]	李敏, 王兴繁, 孙栋元, 等. 基于SPEI的甘肃省气象干旱时空演变特征分析[J]. 水土保持研究, 2025, 32(5): 191-200, 212.
	[Li Min, Wang Xingfan, Sun Dongyuan, et al. Analysis of spatiotemporal evolution characteristics of meteorological droughts in Gansu Province based on SPEI[J]. Research of Soil and Water Conservation, 2025, 32(5): 191-200, 212.]
[35]	Zhang G X, Su X L, Singh V P, et al. Appraising standardized moisture anomaly index (SZI) in drought projection across China under CMIP6 forcing scenarios[J]. Journal of Hydrology: Regional Studies, 2021, 37: 100898, doi: 10.1016/j.ejrh.2021.100898.
[36]	刘婷婷, 朱秀芳, 郭锐, 等. ERA5再分析降水数据在中国的适用性分析[J]. 干旱区地理, 2022, 45(1): 66-79. doi: 10.12118/j.issn.1000–6060.2021.132
	[Liu Tingting, Zhu Xiufang, Guo Rui, et al. Applicability of ERA5 reanalysis of precipitation data in China[J]. Arid Land Geography, 2022, 45(1): 66-79.] doi: 10.12118/j.issn.1000–6060.2021.132

气象站点	纬度/N	经度/E
北京首都国际机场站	40°08′	116°58′
成都双流区站	30°57′	103°94′
哈尔滨站	45°93′	126°56′
乌鲁木齐站	43°78′	87°61′
广州白云国际机场站	23°39′	113°29′
日喀则站	29°25′	88°88′

气象干旱等级	标准化降水蒸散发指数（SPEI）
无旱	SPEI>-0.5
轻旱	-1.0<SPEI≤-0.5
中旱	-1.5<SPEI≤-1.0
重旱	-2.0<SPEI≤-1.5
特旱	SPEI≤-2.0

时态	模型	MAE	MSE	RMSE	RSR	RSD	NSE	相关系数
方案1全部数据	LSTM	0.823	1.112	1.048	0.868	1.024	0.244	0.607
方案1全部数据	WT-LSTM	0.261	0.110	0.328	0.340	0.312	0.882	0.947
春季	LSTM	0.763	0.899	0.939	0.856	0.918	0.265	0.596
春季	WT-LSTM	0.305	0.152	0.378	0.412	0.354	0.822	0.923
夏季	LSTM	0.765	0.891	0.933	0.826	0.904	0.312	0.647
夏季	WT-LSTM	0.302	0.151	0.376	0.403	0.360	0.828	0.922
秋季	LSTM	0.819	1.103	1.035	0.874	0.999	0.230	0.575
秋季	WT-LSTM	0.285	0.134	0.357	0.363	0.338	0.864	0.945
冬季	LSTM	0.658	0.681	0.819	0.832	0.800	0.303	0.607
冬季	WT-LSTM	0.244	0.095	0.302	0.314	0.284	0.898	0.957

时态	模型	MAE	MSE	RMSE	RSR	RSD	NSE	相关系数
方案2全部数据	LSTM	0.647	0.759	0.866	0.710	0.860	0.494	0.724
方案2全部数据	WT-LSTM	0.280	0.150	0.384	0.322	0.376	0.895	0.960
春季	LSTM	0.657	0.760	0.865	0.766	0.855	0.408	0.665
春季	WT-LSTM	0.337	0.282	0.511	0.423	0.489	0.814	0.930
夏季	LSTM	0.599	0.676	0.813	0.682	0.794	0.531	0.754
夏季	WT-LSTM	0.371	0.341	0.570	0.447	0.537	0.793	0.929
秋季	LSTM	0.632	0.790	0.878	0.695	0.863	0.514	0.739
秋季	WT-LSTM	0.308	0.177	0.413	0.323	0.398	0.893	0.963
冬季	LSTM	0.663	0.761	0.863	0.807	0.845	0.342	0.616
冬季	WT-LSTM	0.266	0.137	0.358	0.331	0.342	0.886	0.956

中国气象干旱时空特征与混合模型预测

Spatiotemporal characteristics and hybrid model prediction of meteorological drought in China

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 36

相关文章 2

Metrics

本文评价

推荐阅读 0

[1]	蔡文静,陈伏龙,何朝飞,骆成彦,龙爱华. 基于时频分析的LSTM组合模型径流预测[J]. 干旱区地理, 2021, 44(6): 1696-1706.
[2]	张建海, 张棋, 许德合, 丁严. ARIMA-LSTM组合模型在基于SPI干旱预测中的应用——以青海省为例[J]. 干旱区地理, 2020, 43(4): 1004-1013.