The High-Spectral Infrared Atmospheric Sounder (HIRAS) is a Fourier transform spectrometer onboard the fourth polar-orbiting FengYun 3D satellite (FY-3D). The FY-3D HIRAS provides interferogram measurements of Earth view radiance spectra in three infrared spectral bands at 29 cross-track positions, each with a 2 x p 2 array of field of views (FOVs). The HIRAS level 1 radiance data cover the spectral bands from 650 to 1135 cm⁻¹ [long-wave (LW) band], 1210 to 1750 cm⁻¹ [mid-wave (MW) band], and 2155 to 2550 cm⁻¹ [short-wave (SW) band] with a spectral resolution of 0.625 cm⁻¹. The radiometric calibration algorithm and the methods of refining the nonlinearity (NL) and the polarization correction coefficients on orbit are summarized in this article. The NL correction coefficients are derived by minimizing the spread of the responsivity functions derived from the measurements of the internal calibration target with varying temperatures. The polarization correction coefficients are derived from the cold space observations and the routine Earth scene measurements. The radiometric accuracy is assessed by comparing the HIRAS measurements to the collocated Cross-track Infrared Sounder (CrIS) observations and radiance simulations. The results show that, compared to CrIS, the radiometric differences are about 0.3 and 0.7 K for the LW and MW bands, respectively, and 0.5 K for the CO absorption and window regions in the SW band. The consistency of the radiometric calibration among the four FOVs is estimated to be within 0.2 K for most of the spectral domain. Some remaining issues for the FY-3D HIRAS are also discussed.

HIRAS-II is the hyperspectral detector carried on FengYun-3E which is the world’s first meteorological satellite in dawn–dusk orbit. It fills the observation gaps during the dawn and dusk periods of polar orbit meteorological satellites, enabling a 100% global data coverage and assimilation of polar orbit satellite data within each 6 h window for numerical weather forecasting models. With 3053 vertical detection channels, it provides high-resolution vertical temperature and humidity information, thus playing an important role in improving the forecast skills of the global medium-range weather prediction models. This study assimilated data from 56 CO₂ channels of FY-3E HIRAS into the CMA-GFS 4DVAR system. Two sets of experiments, FY3EHIRAS and CTRL, were designed, conducting a one-month cycle assimilation test to evaluate the impact of assimilating FY-3E HIRAS data on CMA-GFS analysis and forecasting. Using the ECMWF reanalysis data ERA5 as a reference, the study demonstrated that after assimilating data from FY-3E HIRAS’s 56 CO₂ channels, there was a certain extent of improvement in the temperature field at almost all model levels. The RMSE notably reduced in the southern hemisphere’s temperature analysis field near the surface and at 500 hPa by 3.5% and 2%, respectively. The most significant improvement in the entire temperature analysis field was observed in the tropical region, followed by the southern and then the northern hemisphere. Additionally, there was a reduction in RMSE for the height and wind fields, showing considerable improvement compared to the CTRL experiment. Overall, assimilating the FY-3E HIRAS data led to positive improvements in the forecasting skills for temperature, wind fields, and height fields in both the southern and northern hemispheres. The forecasting effectiveness was slightly lower in the tropical region but displayed an overall neutral-to-positive effect.

Considering the characteristics of nonlinear problems, a new method based on the L-curve method and including the concept of entropy was designed to select the regularization parameter in the one-dimensional variational analysis-based sounding retrieval method. In the first iteration, this method uses an empirical regularization parameter derived by minimizing the entropy of variables. During subsequent iterations, it uses the L-curve method to select the regularization parameter in the vicinity of the regularization parameter selected in the last iteration. The new method was employed to select the regularization parameter in retrieving atmospheric temperature and moisture profiles from Atmospheric Infrared Sounder radiance measurements selected from the first day of each month in 2008. The results show that compared with the original L-curve method, the new method yields 5.5% and 2.5% improvements on temperature and relative humidity profiles, respectively. Compared with the discrepancy principle method, the improvements on temperature and relative humidity profiles are 1.6% and 2.0%, respectively.

Based on the community radiative transfer model, ensemble tree-based random forest algorithm, and extreme gradient boosting tree algorithm, this study established a random forest retrieval model (RF) and an extreme gradient boosting tree retrieval model (XGBoost) for the microwave land surface emissivity by using ERA5 reanalysis data and the observed brightness temperature of 10.65 GHz vertical polarization from FY3C Microwave Radiation Imager-I. In addition, an optimized Bayesian regularized neural network retrieval model (M2_30NN) has also been established on the basis of the original neural network land surface emissivity retrieval model (M1_20NN). The results show that compared with the simulated brightness temperature of the original land surface emissivity, the simulated brightness temperature of the land surface emissivity from each retrieval model is not only significantly improved in the correlation coefficient with the observed brightness temperature (5.92% (M1_20NN), 4.23% (M2_30NN), 14.21% (RF), 13.07% (XGBoost)), but also in the evaluation indexes of root mean square error, mean absolute error and explained variance regression score in the training datasets. Furthermore, in terms of testing datasets and spatiotemporal independence test datasets, the retrieval results of RF and XGBoost models can capture the spatial distribution patterns that are consistent with observations well, and also show great numerical improvement compared with the original model. In general, the XGBoost retrieval model is the best, followed by the RF retrieval model.

Cloud contamination is a critical source of errors in the data assimilation of hyperspectral infrared radiance (IR). Therefore, it is necessary to filter out cloudy observations. In this study, we review and summarize the principles and research progress of cloud detection methods for the hyperspectral IR in the past two decades. Based on the impact of IR data utilization on cloud detection results, cloud detection methods are categorized into five types, namely clear field-of-view (FOV) detection, clear channel detection, three-dimensional cloud detection, cloud-clearing and deep learning methods. Clear FOV methods and clear channel methods aim to identify the purely clear FOVs and spectral channels that are not affected by clouds, respectively. Cloud-clearing methods are used to reconstruct clear-column radiance for cloudy observations. Deep learning cloud detection methods can quickly learn the mapping relationship between infrared hyperspectral radiation characteristics and FOV cloud distribution from a large amount of infrared radiative information with known FOV cloud labels. In this paper, we discuss and provide an outlook on the key issues in current hyperspectral IR cloud detection. Specifically, we analyze and summarize the factors affecting cloud detection, such as surface background information, vertical cloud distribution, hyperspectral IR channel selection, improvements in cloud detection algorithms and model applicability. The results indicate the use of deep learning methods offer advantages in detection accuracy and algorithm efficiency of hyperspectral IR cloud detection.