基于SPEI指数的1961-2014年东北地区气象干旱时空特征研究

干旱导致生态系统生产力降低，已成为全球观测的事实，并将在未来气候情景下继续加剧。揭示干旱的时空分布特征是抗旱防灾、保障农业生产安全、维持生态系统健康的迫切要求。应用标准化降水蒸散指数（SPEI）和经验正交函数（EOF）分解方法，分析东北地区干旱特征，揭示干旱发生的时空变化规律。基于东北地区90个气象台站，计算1961-2014年的SPEI指数，从干旱频率、干旱范围和干旱强度等方面研究其特征，并利用EOF分解方法解构干旱空间模态和时间系数。研究结果表明：1）东北地区干旱的时间差异明显。1961-2014年，东北地区以1983年、1995年和2008年为转折点经历了"干-湿-干-湿"的波动变化；夏季发生的干旱范围最大、强度最强，冬季的范围最小、强度最弱；2）东北地区干旱的空间分布差异大，西部干旱发生频繁、发生次数多、持续时间长、旱灾强度大；3）在1、3、6和12个月的多时间尺度下，东北地区年均SPEI变量场EOF分解的前3个主要空间模态均表现为全区一致型、南北相反型和东西相反型，其方差累积贡献率约为58%；4）随着研究时间尺度的增大，干旱的空间分布规律和时间变化趋势逐渐明晰，表明东北地区干旱具有明显的尺度特征。综上所述，基于SPEI指数对东北地区干旱进行多尺度时空分解，刻画了干旱的基本特征，并解构了干旱的时空分异规律，研究结果可为该区的干旱预警研究及生态系统灾害管理提供科学依据。

Spatiotemporal analysis of meteorological drought (1961-2014) in Northeast China using a standardized precipitation evapotranspiration index

Drought-induced reduction in ecosystem production prevails globally and is expected to increase with future climate change scenarios. Understanding spatio-temporal distributions of meteorological drought is fundamental for drought disaster prevention and management, food security, and ecological balance. A standardized precipitation evapotranspiration index (SPEI), derived from the difference between the precipitation and reference evapotranspiration, was used to quantify drought events and consequently reveal the spatio-temporal variation of drought across Northeast China. SPEI at four lags (1, 3, 6, 12 months) was calculated from daily records of mean temperature, daily precipitation, atmospheric pressure, wind speed, relative humidity, and sunshine duration. These climate variables were measured at 90 meteorological stations in Northeast China during 1961-2014. Based on the monthly SPEI values (SPEI01) of each climate station, drought events and disaster were investigated and grouped. Temporal variations of drought events in moderate, severe, and extreme grades were investigated by annual occurrence times, seasonal occurrence ranges, and intensities. Spatial characteristics of meteorological drought in each class were quantified by frequencies of drought events, duration times, and intensities of drought disasters. To analyze the effects of time scale on the drought spatial distribution, the principal modes of variability of the SPEI at four time scales were identified using the empirical orthogonal function (EOF) method. Our results indicated that both interannual and seasonal droughts showed obvious variations across the study area during the period. Interannual drought fluctuated, called "dry-wet-dry-wet" from 1961 to 2014, and was associated with three turning year points of 1983, 1995, and 2008. Seasonal drought appeared more intensive and widely distributed in summer, whereas it was weaker with a smaller distribution range in winter. Drought frequency ranged from 15.2% to 19.2% over the study area. The average drought frequency for moderate, severe, and extreme drought grades was 10.61, 5.2, and 1.2%, respectively. The duration of continuous drought disaster ranged from 2 to 9 months, and the intensity of drought disaster ranged from -8.8 to -3.7. Spatial heterogeneity and complexity of drought were observed in Northeast China. The western region was the most seriously affected area, with the highest drought frequency. The EOF results showed that there were three main spatial modes for drought in Northeast China. The explained variance of the three leading EOFs ranged between 57 and 58.9% as the time scale for calculating the SPEI increased from 1 to 12 months. The explained variance of EOF1 ranged between 37.4 and 39.4%, and the explained variance of EOF2 and EOF3 ranged between 10.6 and 13.1%, and 6.5 and 8.8%, respectively. The spatial coefficients of the EOF1 for all SPEI time scales showed a similar pattern, with spatially uniform variation over Northeast China. Based on the spatial distribution of the spatial coefficients of EOF2 at all SPEI time scales, we identified two homogenous regions:the southern and northern halves of Northeast China. Eastern and western parts of Northeast China were delimited by the spatial coefficients of EOF3. The spatial heterogeneity and temporal variation was more obvious as the time scale increased from 1 to 12 months. The results of this study provided valuable insights for early drought prediction and establishing ecosystem disaster management systems in Northeast China.