At present, the influence of environmental field errors on the simulation effect is rarely considered in seeding models, and the conclusions are often uncertain. In view of this, in this paper, ensemble forecast model by initial field perturbation is one-way coupled with the columnar cloud model with AgI seeding scheme. The cloud model is driven in real time by using the multiple sets of thermal and micro physical profiles provided by the mesoscale model, which include the environmental field disturbance error. The multi-member, single / multi-grid AgI seeding numerical experiments were carried out to simulate the precipitation process of convective-stratiform mixed clouds in Zhejiang Province on January 23, 2022, to determine the optimal seeding scheme and the probability distribution of seeding effect. The following results are obtained. From the simulation effect of a single station (Hangzhou Station), all members of ensemble forecast can achieve positive precipitation enhancement when the amounts of AgI used at a height of 3.6 km(- 5.2 ℃) was 1.2×10^-7 ~1.2×10^-4g/kg at 1500 UTC on 23^rd, in which the precipitation enhancement is the largest when the amounts of AgI was 1.2×10^-5 g/kg, with the mean value of 4.67% and the 99% quantile of 7.77%. In the single point simulation, the initial field disturbance has a great influence on the judgment of whether excessive seeding results in precipitation reduction, for example, when the amounts of AgI is increased to 1.2 ×10^-2g/kg, more than 50% of the ensemble members showed the effect of precipitation reduction, but some members still showed the effect of enhancement. Multiple grid sensitive experiments show that it should be seeding in the northwest and north of Zhejiang, especially in the northeast of Jiaxing and near Lin " an area from the perspective of the optimal probability of seeding effect. These regions also tend to correspond to relatively high average supercooled water content and low average ice crystal number concentration.