为了研究机载测风激光雷达对机下空域3维风场分布进行探测的问题，基于陆基测风激光雷达建立风场模型，采用Levenber-Marquardt(LM)最小二乘法优化径向速率，与传统反演算法进行了仿真对比。结果表明，机载测风激光雷达的工作原理和扫描方式与陆基测风激光雷达类似，不同的是，机载测风激光雷达不仅受飞机速度干扰，而且存在飞机飞行状态不稳定性影响、地球曲率影响、地面强散射回波干扰、地面杂波干扰等特有误差来源，使得风场反演难度更大；而LM最小二乘法具有Newton-Gaussian最小二乘法的快速收敛特性和梯度下降法的保证收敛特性，能满足风场快速和准确的反演。相关结果对测风激光雷达的优化具有参考价值。

In order to study how to detect three-dimension wind field under an airplane, a model for wind field was built based on ground-based wind lidars, and the radial velocity was optimized by means of Levenber-Marquardt(LM) least square method. It was compared with the traditional algorithms. The working mechanism and scanning modes for airborne lidars are similar with ground-based lidars. On the contrary, it is more difficult to retrieve the wind field because of its own flying speed, instability of flight status，the curvature of the earth，the strong ground scattering echo and ground clutter; the LM least square method is able to converge rapidly and accurately because it has the advantages of fast convergence features of the Newton-Gaussian least square method and the ensuring convergence features of the gradient descent method. The results provide references for the optimization of lidars.