混沌理论是进行水文时间序列分析的重要手段。为了保证分析结果的可靠，主张充分利用现有资料，但目前缺乏时间序列长度对混沌特性识别影响的研究。以长江上游武隆站和北碚站日径流序列（1951年－2012年）为例，通过对二者进行混沌分析，研究了最大Lyapunov指数对序列长度的响应。结果表明，日径流时间序列长度过小时会影响混沌识别结果，使结果缺乏可靠性；并不是样本序列长度越长混沌识别结果越好；当序列长度达到3000左右时，序列的混沌特性达到稳定，结果可靠并缩短了计算时间。

Chaotic theory is an important means of hydrology time series analysis. In order to get reliable analysis results, it is recommended to make a full use of time series. But the research about how the length of time series affects the identification of chaotic characteristics is rare. In this paper, we carried out a study about the responding effect of the maximum Lyapunov exponent to the length of time series with the use of daily runoff time series of gauged stations named Wulong and Beibei in Yangtze River. The result suggested that short daily runoff time series would affect the result of chaotic identification and make the result unreliable; besides, when the length of daily runoff time series reached 3000, the chaotic characteristics became stable and reliable, and it saved a lot of computing time at the same time.

国家自然科学基金面上项目（51479061）.