Ecological flow provides reference values for maintaining the stability of river ecosystems and rich biodiversity. Therefore, formulating ecological flow in the river channel downstream of hydropower stations as an ecological constraint condition for the scheduling of cascade reservoir groups in the upper Yangtze River is crucial for protecting river ecosystems. Existing methods for determining ecological flow often only consider either intra-annual or inter-annual runoff distribution characteristics. However, an analysis of runoff variation patterns in the upper Yangtze River shows that inter-annual and intra-annual runoff distributions are highly uneven, with unique characteristics of wet-dry co-occurrences at annual and monthly scales. To address the limitation of existing methods in describing intra- and inter-annual variations in river runoff, this paper proposes an innovative ecological flow determination method based on the probability distributions of annual and monthly runoff. First, the marginal distributions of annual and monthly runoff are fitted, and a Copula function is used to construct the joint probability distribution of annual and monthly runoff. Then, based on Bayesian inference, the conditional probabilities of different monthly runoff magnitudes under various annual runoff conditions are calculated. Finally, these conditional probabilities are combined with the Flow-Duration Curve (FDC)-based method to determine an ecological flow process that accounts for both intra-annual and inter-annual runoff variations. This study provides a regulatory red line for the development of water resources and hydropower energy in the upper Yangtze River, while facilitating better protection of local wildlife species.