干旱是一种复杂的自然灾害，制约着我国社会经济发展和生态文明建设。面对频发的 干旱灾害，传统的干旱应对策略亟待进一步深入研究。单一干旱类型的研究已经不能满足 干旱灾害全方面、系统性防治的需求。开展变化环境下多类型干旱的发展演变过程研究， 是增强防旱抗旱能力的重要环节。探明不同类型干旱间的传播演变机制，可为干旱的发生 提供预报预警，为减小干旱损失、保障流域水安全提供理论和技术支持。

本研究以京津冀地区重要水源涵养区滦河流域为研究区，基于观测、模拟和遥感数据， 研究变化环境下滦河流域不同类型干旱的演变特征及传播规律。首先，构建分布式水文模 型，对研究区各水文要素进行重构，进行多类型干旱评价。然后，运用游程理论识别各类 干旱事件，评价不同类型干旱的发展、恢复和三维时空演变特征。其次，基于不同时间尺 度干旱的时空关联度，评价气象干旱对水文、农业干旱的累积影响时间特征，评估不同干 旱的联合风险分布。再次，根据不同干旱发生的时间顺序，识别不同类型干旱事件的传播 关系；在全面讨论干旱传播事件的滞后关系和发展、恢复特征的基础上，运用 Copula 函数 构建干旱特征传播关系及评价模型，通过偏相关分析系统评价环境要素对干旱传播规律的 影响。最后，基于未来气候和人类活动变化情景，预估未来流域干旱传播特征，并为流域 应对气候变化、加强水资源和干旱灾害管理提供建议。

本文得到的主要研究结论如下：

（1）基于 SWAT 模型重构径流、土壤水等水文要素，分别选取不同时间尺度的标准化 降水蒸散发指数（SPEI）、标准化径流指数（SRI）和标准化土壤水含量指数（SSI），作为 气象、水文、农业干旱的评价指标。基于 1973-2016 年的气象要素趋势分析，滦河流域的 气象、水文干旱加剧。气象干旱的历时、强度低于水文和农业干旱，其中干旱发展演化过 程中，发展阶段历时、强度高于恢复阶段。流域三维干旱特征研究结果显示，近 20 年来流 域极端气象、水文干旱事件发生频繁，农业干旱时间分布没有明显变化。极端气象干旱事 件更易从流域下游地区向上游迁移，水文干旱从支流向干流迁移，极端农业干旱多点并发 向全流域扩散。

（2）气象干旱与农业干旱的关系较水文干旱相关性高、累积影响时间短、联合风险高。 季节上，干旱相关性在夏季最为紧密，干旱的累积影响时间经历了从春季到夏季迅速缩短， 再经秋、冬季逐渐增高的过程。气象干旱对水文干旱的累积影响时间相对较长，气象干旱 对水文干旱累积影响时间在冬季最长（6-8 个月）；对农业干旱累积影响时间在春季最长（4- 8 个月）。干旱联合风险在夏秋季更高，下游地区干旱联合风险更高。

（3）气象干旱向水文、农业干旱的传播中，干旱历时、强度均有不同程度的增强，且在干旱恢复阶段增强趋势更明显。流域干旱传播关系可划分为一对一（S1）、一对多（S2）、 多对一（S3）及多对多（S4）4 种类别。干旱传播事件的平均历时、强度特征在 S4 最大， S1 最小。干旱发生时间上，春、冬季节气象干旱传播到水文干旱时间早于农业干旱 1-2 个 月，夏季传播时间相近，秋季气象干旱传播到农业干旱的时间明显短于传播到水文干旱的 时间。

（4）基于干旱特征关系建立 Copula 函数，应用最大可能权函数方法构建干旱播模型。 模型模拟结果表明，引发水文干旱的临界条件低于引发农业干旱的临界条件。干旱历时、 强度主要受地形坡度、海拔等地形地貌条件影响，峰值强度受降水、气温和相对湿度等气 象条件影响。气象干旱向农业干旱传播模型历时特征还受人口数量等社会经济因素影响。

（5）校正后的未来气候模式中，低碳排放的情景（SSP126）下降水、气温变化小，而 中、高碳排放情景（SSP245、SSP585）模式下降水显著增加，升温超过 1.5℃。土地利用 变化主要集中在下游西南部。未来干旱传播事件中，S3、S4 传播类别占比增加，干旱传播 关系变得更加复杂。未来碳排放强度增高将使干旱发展和恢复阶段的差距缩短，干旱传播 风险升高，干旱形成的临界值降低。建议通过有效的措施减少温室气体排放，建立完善的 干旱预报预警体系，优化土地利用格局，多部门联合调控干旱风险。

Drought is a complex natural disaster and adversely affects China's socio-economic development and ecological civilization. Due to the frequent occurrences of droughts, traditional drought response strategies need to be improved. Research on a single type of drought can no longer meet the needs of comprehensive and systematic prevention and control of drought disasters. It is essential to enhance the ability of drought prevention and prediction, through the improved analysis of the evolution of multi-types of droughts under the changing environment, which can provide theoretical and technical support for reducing the losses of drought and ensuring the safety of water supply and usage.

This study takes the Luanhe River Basin, an important water conservation area in the Beijing- Tianjin-Hebei region, as the research area. Observation, simulation, and remote sensing data were used to analyze the evolution of multi-types of droughts. Firstly, a distributed hydrological model is constructed to simulate the hydrological components of the study area. The meteorological drought (MD), hydrological drought (HD), and agricultural drought (AD) in the study area were reconstructed, and multi-scale drought indices were built respectively. Based on the run theory, the development and recovery characteristics and three-dimensional spatio-temporal evolution of droughts are evaluated. Secondly, based on the correlation between MD and HD, AD at different time scales, the accumulated time of influence from MD to HD and AD and their joint risk are evaluated. Then, taking the time sequence of drought occurrence as the index, the development and recovery characteristics and lag characteristics from MD to HD and AD are classified and discussed. Based on the Copula function, the propagation model of drought characteristics is constructed. The influencing factors of the drought propagation are systematically evaluated by the partial correlation analysis method. Finally, under the changing climate and human activities in the future, drought propagation characteristics were projected. The projection results may provide useful insights for climate change adaptation, water resources regulation, and drought disaster management.

The main conclusions of this paper are as follows:

（1）Based on the reconstructed runoff, soil water, and other hydrological elements by the SWAT model, the standardized precipitation evapotranspiration index (SPEI), standardized runoff index (SRI) and standardized soil moisture index (SSI) of different time scales were selected as the MD, HD, and AD evaluation index. The trend analysis of drought from 1973 to 2016 showed that MD and HD in the Luan River Basin have intensified. The duration and severity of MD are lower than HD and AD. In the process of drought development and recovery, the duration and severity of the development stage are higher than the recovery stage. The results of the three- dimensional drought characteristics of the basin show that extreme MD and HD events have occurred frequently in the past 20 years, and the frequency of AD has not changed significantly. Extreme MD events are more likely to migrate from the downstream of the basin to the upstream, HD migrating from tributaries to main streams. And extreme AD spreads to the whole basin.

（2）The closer correlation, shorter drought influences accumulated time, and higher joint drought risk are shown in the relation between MD and AD than MD and HD. The seasonal fluctuation of the correlation coefficient between MD and HD in the upstream is significantly higher than in the downstream. The drought influence accumulated time for both HD and AD has experienced the process of rapid shortening for one month from spring to summer, and gradually increasing through autumn and winter. The HD in winter is affected by the longer-scale MD for about 6-8 months, while the longest-scale MD with the greatest impact on AD occurs in spring (4- 8 months). The joint risk of MD and HD is lower in winter and spring than in summer and autumn.

（3）Drought duration was extended when propagated from MD to HD and AD, and the amplification will be aggregated on the drought recovery stage. The propagation from MD to HD and AD can be divided into the one-to-one scenario (S1), multiple HD (AD) caused by one MD (S2), one HD (AD) caused by multiple MDs (S3), and multiple HD (AD) and ADs caused by multiple MDs (S4). The average duration and intensity characteristics of drought propagation events are the largest in S4 and the smallest in S1. The beginning time of HD in spring and winter is 1-2 months earlier than AD. The lag time is the same between HD and AD in summer, and the beginning time of HD in autumn is slightly later than AD.

（4）The Copula function was established based on the relationship of drought characteristics, and the drought propagation model was constructed using the maximum possible weight function method. The simulation results of the model show that the critical conditions for hydrological drought are lower than those for agricultural drought. The duration and severity of drought are mainly affected by topographic and geomorphic conditions, such as topographic slope and altitude. Furthermore, the peak intensity of the model is affected by meteorological conditions, such as precipitation, air temperature, and relative humidity. The propagation characteristics of the MD to AD propagation model are also affected by socio-economic factors such as population.

The corrected future climate model presents little change in precipitation and temperature under  the  low emission  scenario (SSP126).  Under  the  medium and high  emission   scenarios (SSP245 and SSP585), the precipitation increases significantly, and the temperature rise exceeds 1.5 ℃. Land use change is mainly concentrated in the southwest of the study area. In the future, the proportion of S3 and S4 type events increases. The characteristics of drought duration and severity in the SSP585 scenario will be higher than the others. With the increase of emissions: the gap between the drought development stage and recovery stage is shortened. The critical value of drought formation is reduced with increased drought propagation risk. Effective measures should be carried out to reduce greenhouse gas emissions, establish a complete drought forecast and early warning system, optimize land use patterns, and multi-sectoral joint control of drought risks.