中文摘要：

针对干旱区河流汇流计算的特点，选用了3种计算模型，即多元线性回归模型、系统模型、人工神经网络模型，并以塔里木河三源流汇流计算为例，建立了多源流汇流计算的三种模型，模拟了实测旬径流过程。通过实例分析。说明本文建立的三种计算模型模拟精度高。且适用性强，可以作为塔里木河汇流计算的预测模型。为塔里木河上游源流来水预测提供基础模型，为流域水资源分配和调度奠定基础。

英文摘要：

Located in the Tarim Basin, the Tarim River Watershed （34°55＇-43°08＇N, 73°10＇-94°05＇E） is the largest continental drainage basin in China, flows along the northern marginal zone of the The Takimakan Desert from west to east and passes through the Tarim Bain, and its drainage area is 4.355x10^5 km^2. There is no runoff formation in the mainstream area of the Tarim River, so the mainstream of the Tarim River is completely recharged by its tributaries. Most of the tributaries do not recharge the mainstream anymore due to the climate change and the impacts of human activities, and currently only 4 source streams recharge the mainstream. Based on the characteristics of calculating the runoff volume converged by the rivers in arid areas, in this paper three calculation models are selected, they are the multivariate linear regression model, system model and artificial neural network model respectively, a case study on calculating the runoff volume of the Tarim River, converged by three main source streams, is carried out by developing the models of multi-source convergence calculation, and the ten-day runoff process of the Tarim River is simulated. The results show that the three calculation models are effective and feasible, so they can be used to predict the runoff volume of the Tarim River. The three calculation models are the foundation in calculating the inflow of the Tarim River, converged by the three main source streams, and also can provide the scientific basis for redistributing and scheduling water resources in the Tarim River Watershed.