基于主成分分析的广西省干旱时空格局

广西省地处喀斯特地貌区,土壤保水效率低,且年降水时空分布不均,研究干旱时空分布尤为重要。基于广西省1981—2010年20个气象站实测和2011—2100年HadGEM2-ES模型模拟数据,利用标准化降水指数(SPI, Standardized Precipitation Index)和标准化降水蒸散指数(SPEI,Standardized Precipitation Evapotranspiration Index),分析广西省干旱的时空变化。对不同时间尺度的SPI和SPEI应用主成分分析(PCA,Principal Component Analysis)确定干旱的空间模式,结果揭示了3个空间分布明确的区域:桂东北地区(PC1),桂西北地区(PC2)和桂南地区(PC3)。各区域干旱的时空变化和频率分布差异显著。PC1和PC3的SPEI-12呈负增长趋势,PC2的SPEI-12呈正增长趋势,且PC1、PC2和PC3的SPI-12均大于SPEI-12。年尺度(SPEI-12)上PC1和PC3的干旱频率大于PC2,其干旱频率分别为34.24%和35.83%。SPI和SPEI在空间和时间尺度上相关,且具有高度可比性。SPEI因子载荷的空间模式优于SPI,严重干旱年份(1988、1996和2003年)的SPI值明显小于SPEI,且SPEI检测到的干旱频率较高。研究结果可作为广西省干旱预测、评估及其风险管理和应用决策的重要科学基础,也可为广西省区域社会经济发展提供重要的科学参考。

Spatio-temporal drought patterns in Guangxi Province based on principal component analysis

Guangxi Province is located in the karst landform where the soil water conservation efficiency is low and the annual precipitation is unevenly distributed, so it is especially important to study the spatial and temporal distribution of drought. This paper analyzed the spatial and temporal variability of drought in Guangxi using the Standardized Precipitation Index (SPI) and the Standardized Precipitation Evapotranspiration Index (SPEI) based on the historical measured (1981-2010) and HadGEM2-ES model simulated data (2011-2100) of 20 meteorological stations in Guangxi Province. The spatial patterns of drought were identified by applying Principal Component Analysis (PCA) to the SPI and SPEI values at different time scales. The results revealed the existence of three regions: Northeastern (PC1), Northwestern (PC1), and Southern (PC1). The spatial and temporal variation and frequency distribution of drought in different regions were significantly different. SPEI-12 of PC1 and PC3 showed a negative growth trend while that of PC2 showed a positive growth trend, and SPI-12 of PC1, PC2 and PC3 were larger than SPEI-12. The drought frequency of PC1 and PC3 (34.24% and 35.83%) was higher than that of PC2 on the annual scale (SPEI-12), which indicated PC1 and PC3 would have more drought than PC2. SPI and SPEI were highly comparable on the spatial and temporal scale. The spatial mode of the SPEI factor load is better than SPI, and the SPEI detects a higher drought frequency. The results can be used as an important scientific basis for drought prediction and assessment in Guangxi Province and its risk management and application decision-making. It can also provide an important scientific reference for the socio-economic development of Guangxi Province.