A Framework on Analyzing Long-Term Drought Changes and Its Influential Factors Based on the PDSI

Abstract

Drought is one of the most frequent and most widespread natural disasters worldwide, significantly impacting agricultural production and the ecological environment. An investigation of long-term drought changes and its influencing factors provides not only an understanding of historical droughts but also a scientific basis for the protection of future water resources. This study investigated the temporal characteristics of drought in a study site located in the center of Southwest China (SWC) over a 700-year period (AD 1300–2005) using the Palmer Drought Severity Index (PDSI). The linkage between drought and its influencing factors is discussed. An algorithm based on the random forest (RF) method was proposed to analyze the dynamic influence of the factors on drought. We also examined the linkages between the demise of two dynasties and historical drought events. The results showed that the study site was a drought-prone area in the study period and experienced a non-significant drying trend in all centuries, except for the 17th century; a total of 232 droughts were detected in the study site from AD 1300–2005. The wavelet spectrum of the PDSI series showed the existence of 4-, 8-, 16-, 32-, and 128-year-periods. A strong correlation existed between the sunspot numbers and the PDSI. The correlation of the period between the PDSI and El Niño-Southern Oscillation (ENSO) series in the same frequency domain was weak, while the ENSO exhibited a strong interaction with the PDSI in some time periods. The Pacific Decadal Oscillation (PDO) and PDSI had no resonance period in the low-frequency region, but there was a period of 80–130 years in the high-frequency region. The relative rates of influence of the ENSO, sunspot numbers, and PDO during AD 1700–1996 were 38.40%, 31.81%, and 29.8%, respectively. However, the mechanism of the interaction between droughts and the influential factors is complex, and the dominant factor changed over time. The analysis of long-term drought changes based on the PDSI series may provide clues to understand the development of historical events.