Based on the description of the concept of the low flow and low-flow discharges,the Probability Plot Correlation Coefficient(PPCC) method is used to test the probability distribution of monthly low flow at twenty stations in Guangdong Province.

Daily runoffs during low flow were simulated with the model for the four study basins in the Dongjiang watershed.

The 7-day low flow was regarded as statistical quantity,on the base of analysis of the discordancy measure and heterogeneity measure,the regional low flow frequency for Donjiang basin was carried out using L-moments based five frequency distributions with three parameters viz.

The importance of low flow forecast was briefly described in this paper.

This paper mainly expounds the hydrographical features of the river low water runoff and its extremum, including discharge variation, occurring time, water quality and water temperature of river low water runoff In the paper, discussion is emphsized on the mechanism of river low water runoff and its influencing factors such as: ① zonal elements, including climate, vegetation, etc.

Using the systemic and comparative methods, this paper reviews the low flow runoff from natural and anthropogenic factors, its forecast methods.

Prediction of monthly runoff based on BP neural network;

Influences of sea surface temperature and meteorological factors on monthly runoff of Hongjiadu Hydropower Station;

Relationship between sea-surface temperature (SST) and monthly runoff of upper Yellow River;

This model is used to forecast the monthly runoff volume of Hongjiadu Hydropower station with the method of stepwise regression considering sea temperature and meteorological factors.