Global wind field measurement is the basis of weather forecasting and climate research since wind speed is one of the basic parameters to describe the atmospheric state. Spaceborne lidar is an effective means to obtain the global wind field due to the characteristics of high accuracy, high vertical resolution, global coverage, etc. At present, spaceborne wind lidar technology is in development in China, and it needs a lot of theoretical and experimental support. In many research directions of this technology, the simulation technology of detection mode and observation results is very important. Previous research has simulated different detection modes of spaceborne lidar, but those simulations mostly came from standard ideal scenes, and have not involved the analysis of complex real scenes. In this work, we carry out a practical simulation by employing a real scene from the WRF model in which both clouds and aerosol exist at the same time. The results from the analysis on detection ability can improve the practicability of the simulation method, and provide support for the design and improvement of spaceborne wind lidar.

We build a forward model including six sub-modules (Fig. 1): atmospheric scene, calculation of atmospheric radiation transfer, satellite orbital platform parameters, instrument parameters, inversion analysis and comparison verification, and parameter sensitivity analysis. The parameters such as wind speed, aerosol mass density, mixture ratio of hydrometeor, and molecular number density are obtained from the atmospheric scene. The optical properties of particles in clouds and aerosol depend on the Optical Properties of Aerosols and Clouds (OPAC) database. The atmospheric absorption calculation comes from the line-by-line integration model of LBLTRM. Then the overall attenuation backscattering coefficients at different altitudes can be calculated through atmospheric radiation transfer. Substituting the parameters of the satellite orbit platform and lidar instruments into the forward model, the photon signal received before the discriminator can be obtained. By coupling the Fizeau interferometer, accumulation charge coupled device (ACCD) detector, and other models, the output signal on the detector is simulated. To verify the simulation results, the retrieved wind speed from the simulation signal is compared with the original wind speed provided by the scene. In addition, parameter sensitivity analysis is also used to discuss the impact on the simulation detection results.

The forward model in this paper can simulate complex real scenes, and the authenticity and practicability of the simulation have been greatly improved (Fig. 1). The results of wind speed profiles with regard to altitude show that clouds and aerosol can increase the signal-to-noise ratio (SNR) of the detection and improve the inversion accuracy. However, when the cloud layer is thick, due to the attenuation effect of the cloud layer on the signal, the SNR below the cloud layer will decrease, thus increasing the error of wind speed error (Fig. 6). The two-dimensional (2D) profiles of wind speed are then simulated with the horizontal resolutions of 1 km and 5 km, respectively. The results show that the wind speed error can be reduced by the decreasing horizontal resolution due to the increasing echo signal intensity through accumulating pulses. When the centroid method is used to retrieve the wind speed, the wind speed oscillation error can be reduced by increasing the number of channels of the ACCD detector (Fig. 10).

Based on the spaceborne wind lidar detection mode of the Fizeau interferometer and using the Atmospheric Laser Doppler Instrument (ALADIN) instrument parameters as input, we simulate the detection signals in the presence of clouds and aerosol. By comparing the inversion results with the input original value, we discuss the main factors affecting the wind speed error and analyze the parameter sensitivity. The results show that the clouds and aerosol in the atmosphere can increase the attenuation backscattering coefficient, improve the SNR and reduce the wind speed error. When the maximum likelihood function method is used to retrieve the wind speed, the wind speed error ranges within ±1.2 m/s. However, when the cloud layer tends to be thicker, the lidar cannot penetrate the cloud layer due to the strong attenuation effect of the cloud layer on the pulse energy, so the information below the cloud layer cannot be detected effectively. For the same vertical resolution, the wind speed error with the horizontal resolution of 5 km is significantly better than that with the horizontal resolution of 1 km, which means that reducing the horizontal resolution and increasing the number of cumulative pulses can improve the wind speed detection accuracy. Through the sensitivity analysis of typical parameters, the lower satellite orbit altitude or the greater laser emission power can enhance the SNR and reduce the wind speed error. Furthermore, the increasing number of the ACCD detector channels can reduce the system oscillation error caused by the principle of the centroid method.