基于图像三维连通性识别方法的长江流域干旱事件特征

全球变化背景下,干旱事件频发给区域水资源管理和社会经济可持续发展带来了巨大挑战,科学合理地辨识干旱发生发展过程一直是干旱研究领域的前沿与难点。该研究将图像三维连通性识别方法应用于气象干旱事件的识别和提取。在长江流域的应用表明此方法识别出的干旱事件与历史记录的吻合度较高,能有效识别气象干旱事件。基于1960—2015年标准化降水蒸散发指数(Standardized Precipitation Evapotranspiration Index, SPEI),长江流域共发生281场气象干旱事件,其中长历时干旱事件(历时大于4个月)64场。长历时干旱事件在流域中部发生次数多于东部和西部,2000年之后的发生频次、干旱面积和干旱强度大于2000年之前,单场事件发展方向以西北和东南为主。整体上,干旱事件发生频次和干旱强度在长江流域中部呈增加趋势,在流域东部和西部呈减小趋势。该研究有助于探索长江流域干旱事件发展演变规律。

Characteristics of drought events using three-dimensional graph connectedness recognition method in the Yangtze River Basin, China

Drought events occur more frequently against global warming in recent years. There is a great threat to the regional water resource management, ecosystem, and sustainable socio-economic development in China. Identification of drought events is essential to further understand characteristics of drought onset, evolution, and recovery. In this study, taking the Yangtze River Basin (YRB) as the study area, a new recognition method was proposed to identify the regional drought events using three-dimensional graph connectedness. By contrast, the specific procedure was first to identify the drought patches in each month using the threshold of occurring areas, and then to merge the drought patches in adjacent months using the threshold of overlapping areas, particularly in the conventional identification of drought events, such as clustering. This new identification was first to recognize the three-dimensional continuum with no limitation, and then to split the continuum using the threshold of occurring and overlapping areas, finally to extract the drought events with three-dimensional characteristics. The number of times was reduced significantly to calculate the overlaid area in space. The spatial-temporal patterns of droughts were obtained to verify the efficiency using the three-month YRB Standardized Precipitation Evapotranspiration Index (SPEI3) from 1960-2015. The results showed that: 1) 1.5% of the study region was considered as the threshold of occurring and overlapping areas, compared with the previous official reports. The drought events identified by the new method in YRB matched very well with the historical occurred ones, covering the onset, termination, and propagation. There was a significant positive correlation between the drought intensity of single drought events and the moisture deficit of surface soil in the affected region. The moisture deficit of surface soil was larger, as the drought intensity increased. It proved the efficiency of the new method in identifying and extracting three-dimensional drought events; 2) 281 drought events are identified in YRB from 1960-2015 and the drought duration of 64 events are larger than 4 months; 3) The duration of long duration drought events (which duration is larger than 4 months) decreased, while the drought area and drought frequency increased. Long duration droughts occurred 43 times during 1960-1999, the average drought duration and drought area were 8.1 months and 2.74×104 km2, respectively. And long duration droughts occurred 21 times during 2000-2015, the average drought duration and drought area were 7.6 months and 3.50×104 km2 respectively. The frequency of long duration drought events in the middle region of YRB is higher than that in the eastern and western regions. 4) The drought events occurred more frequently in Shannxi, Hubei, and Sichuan provinces, while the north Yunnan and Guizhou provinces presented a relatively lower frequency. Accumulative drought intensity was large in north Guizhou, northwest and east Sichuan, and Hubei provinces and Chongqing City, whereas, relatively small in the north Yunnan, Jiangxi and south Jiangsu provinces. The frequency and severity of drought events presented an increasing trend in the middle region of YRB, while a decreasing trend in the western and the eastern. The center of about 50% drought events moved to the northwest, due mainly to the water vapor transference and the migration of precipitation belt. The identification of drought events can effectively recognize and extract drought events in three-dimension images. The finding can provide greatly promising guidance for drought and regional water resource management.