In air traffic flow management (ATFM), flights delays are most often caused by weather events, which are stochastic in nature. In practice however, these stochastic obstacles are dealt with deterministic strategies, especially ground hold method. This means increasing delay or canceling flights. This paper presents a novel traffic assignment model considering uncertain weather in an air traffic network to improve the method. A generalized link travel time function is proposed to capture these effects, and base on this it derives concepts of reliability of link travel time to reflect flight delay range, in order to reach the goal of minimizing total delay and maximizing flights contained in the network, it proposed a multi-constrained nonlinear optimal scheduling model, and moreover, an modified gradient projection algorithm has been proposed to solve the mathematical model efficiently. Numerical experiments show the model has a significant improvement in comparison with multi-constrained ground holding strategy.