Abstract: Abstract: In order to improve the efficiency of hybrid rice planting mechanization and expanding the row width ratio of the parents planting, rotary-wing UAV (unmanned aerial vehicle) is used to the supplementary pollination work. Rotor wind is driven by UAV rotor rotating, which propels the air flow in crop canopy and forms wind field. Cover width of wind field, wind speed in 3 directions and distribution of wind field will directly affect the agricultural UAV's field effect. In this paper, based on the UAV flight parameters, wind speed acquisition system was used to collect pollination's wind speed of 18-rotor UAV; for wind data, the significance of the row and column data of matrix data (100×60) was fully discussed, and the characteristics of row and column data were summarized and it was processed with the field test. The temporal change law of the wind speed data in three directions has the characteristics of consistency, and the average value of X direction is greater than Y and Z direction before the maximum moment; the difference of the wind speed value sequence curve between X and Y is less than the differences between X and Z or Y and Z. The space distribution of wind speed values in 3 directions suggests that the maximum average value of collected wind speed occurs in the intersection of UAV flight path and a sensor array (9#-11#); considering the error of measurement value, the farther the distance between the sample point and flight path, the more the attenuation of corresponding wind speed value of sampling points. Summarizing two-dimensional wind field data, it is found that the result of the wind field widths in 3 directions is Y>X>Z. On this basis, the method of Gaussian curve fitting is used to calculate the row data and column data; by comparing the statistical parameters, column data is fitted to establish the five-order exponent function model of the relationship between wind speed data and time, and row data is fitted to establish the six-order exponent function model of the relationship between wind speed data and sample point. The method of matrix transformation is used to eventually establish the ideal 2-dimensional wind field model in UAV rotor X direction in rice canopy based on row and column data models. And by the model diagram, it is found that "steep" effect exists in the distribution shape of wind field in X direction, which means the maximum wind speed is below the rotor drones, the increasing rate of the wind speed in forward direction is significantly higher than the reducing rate of backward direction, and the wind field "steep" presents bilateral symmetry along the UAV flight direction. "Steep" effect and the model parameters are used to clarify the shape of the distribution of UAV rotorcraft wind field in rice canopy plane. Then we can study how to use independent air source or auxiliary device to change the existing wind field distribution shape to improve pollination effect. The new method provides the theoretical foundation for the UAV pollination work. It must be noted that the model is only a single sample from a single-direction data, and only the ideal basic model of wind field distribution of UAV rotorcraft in the canopy, and the further researches are needed for one-direction model of the UAV rotorcraft wind field.