Temporal clustering of the Tarim River floods and its relationship to climate teleconnection patterns

Temporal clustering of magnitudes and times of occurrence of floods in the Tarim River basin recorded at 8 hydrological stations located in 5 states was analyzed using the Cox regression model and monthly frequencies of flood occurrences and dispersion methods under the impact of low-frequency climate change. Results indicated that:under the effect of climate change, there was a significant temporal clustering in the flood process in Tarim River basin. Periods of high flood frequencies trended to be the periods of high flood mitigation seriously influenced by regional flood control which was the main reason for large flood-induced losses in the Tarim River basin. The positive phase of climate indices resulted in the early occurrences of floods for the same exceedance probabilities and decreasing exceedance probabilities for the same time of flood occurrence at the hydrological stations and in the region, which both had positive coefficient values fitted by the Cox regression model with climate indices as covariates, while the negative phase of climate indices had an opposite effect. The exceedance probability of flood occurrence time preferably responded to low-frequency change, which was advantageous for flood control and flood risk management in the Tarim River basin. The majority of hydrological stations did not reveal any inner-annual clustering of flood occurrences when obtained by POT sampling, while regional flood occurrences had significant clustering.