Search

An improved microwave gaseous absorption scheme based on statistical regression is proposed in this study. In the new scheme, Monochromatic Radiative Transfer Model (MonoRTM) replaces the Millimeter‐wave Propagation Model (MPM) to train the new scheme and the effect of real Spectral Response Functions is included in the training process. After the replacement, results of the new scheme are closer to observations from Advanced Technology Microwave Sounder (ATMS) onboard Suomi National Polar‐orbiting Partnership satellite in low level channels while MPM has some advantages in upper level channels. Introducing ozone absorption can cause a systematic bias but results in small standard deviations in channels with frequency 183 ± 1.8 GHz and 183 ± 1 GHz. In addition, the new scheme updates the vertical interpolation of water vapor and optimizes the vertical distribution of Planck function. These updates can reduce biases caused by vertical interpolation, especially for water vapor absorption channels. The bias associated to vertical interpolation can reach 0.25 K whereas the new scheme can decrease it to 0.003 K. To further validate the accuracy of the new scheme, we apply the new scheme to Advanced Radiative transfer Modeling System and compare simulated results to RTTOV 13.2 under 37 and 137 level (L) atmosphere (atm). Observations from ATMS onboard NOAA‐20 are used as true values. Results show that the new scheme agrees with RTTOV 13.2 well in accuracy and performs even better in upper level channels and water vapor absorption channels. Plain Language Summary: Fast radiative transfer (RT) models play a crucial role in satellite data assimilation as well as remote sensing. Gaseous absorption calculation is a fundamental aspect of fast RT models. With the capability of new‐generation spaceborne microwave instruments to observe spectra beyond 200 GHz, Monochromatic Radiative Transfer Model (MonoRTM) is utilized to train the microwave gaseous absorption scheme. The absorption of oxygen, ozone, water vapor and nitrogen is considered. In addition, the mismatch of vertical coordinates between fast RT models and input profiles requires fast RT models to do vertical interpolation during simulations. Vertical interpolation may introduce notable biases in microwave water vapor absorption channels. In this study, we enhance the vertical interpolation of water vapor and optimize the vertical distribution of the Planck function to mitigate these biases. The accuracy of the new scheme is assessed by comparing it with MonoRTM and satellite observations. Key Points: A new microwave gaseous absorption scheme, trained using Monochromatic Radiative Transfer Model (MonoRTM), is developed and the effect of ozone and nitrogen absorption is evaluatedThe accuracy of the new scheme is enhanced through updates to water vapor interpolation and the vertical distribution of Planck functionAfter application in Advanced Radiative transfer Modeling System (ARMS), the new scheme aligns with RTTOV and gets greater accuracy in upper‐level and water vapor absorption channels [ABSTRACT FROM AUTHOR]

A polarized bidirectional reflectance distribution function (pBRDF) matrix was developed based on the two-scale roughness theory to provide consistent simulations of fully polarized microwave emission and scattering, required for the ocean–atmosphere-coupled radiative transfer model. In this study, the potential of the two-scale pBRDF matrix was explored for simulating ocean full-polarization backscattering and bistatic-scattering normalized radar cross sections (NRCSs). Comprehensive numerical simulations of the two-scale pBRDF matrix across the L-, C-, X-, and Ku-bands were carried out, and the simulations were compared with experimental data, classical electromagnetic, and GMFs. The results show that the two-scale pBRDF matrix demonstrates reasonable dependencies on ocean surface wind speeds, relative wind direction (RWD), geometries, and frequencies and has a reliable accuracy in general. In addition, the two-scale pBRDF matrix simulations were compared with the observations from the advanced scatterometer (ASCAT) onboard MetOP-C satellites, with a correlation coefficient of 0.9634 and a root mean square error (RMSE) of 2.5083 dB. In the bistatic case, the two-scale pBRDF matrix simulations were compared with Cyclone Global Navigation Satellite System (CYGNSS) observations, demonstrating a good correlation coefficient of 0.8480 and an RMSE of 1.2859 dB. In both cases, the two-scale pBRDF matrix produced fairly good simulations at medium-to-high wind speeds. The relatively large differences at low wind speeds (<5 m/s) were due probably to the swell effects. This study proves that the two-scale pBRDF matrix is suitable for the applications of multiple types of active instruments and can consistently simulate the ocean surface passive and active signals. [ABSTRACT FROM AUTHOR]

Global surface snowfall rate estimation is crucial for hydrological and meteorological applications but is still a challenging task. A novel approach is developed to comprehensively use passive microwave, infrared data and physical constraints in deep‐learning neural networks with an attention module for retrieving surface snowfall rate (PCSSR‐DNNWA). The PCSSR‐DNNWA consistently outperforms traditional approaches in predicting surface snowfall rates with a correlation coefficient of ∼0.76, mean error of ∼−0.02 mm/hr, and root mean squared error of ∼0.21 mm/hr. It is found that graupel water path is of vital importance with largest contributions in retrieving surface snowfall rate. By integrating the physical constraints, the algorithm of PCSSR‐DNNWA opens a new avenue for retrieving the surface snowfall rate from satellites since some predictors are intelligently considered, resulting in an increased accuracy, interpretability, and computational efficiency. Plain Language Summary: Comprehensively monitoring surface snowfall on Earth can effectively be achieved through space‐borne instruments. However, estimating surface snowfall from space is a challenging task as the signals measured by space sensors are indirectly related to surface snowfall rate. In this study, a novel deep learning algorithm is developed based on deep neural networks, which is more accurate, interpretable and computationally efficient, compared with traditional approaches, in estimating surface snowfall rate using observations from various space‐borne sensors and physically relevant parameters. Key Points: Physical constraints greatly improve the ability of surface snowfall rate retrievalAttention module in deep neural networks could intelligently adjust the weights of predictors [ABSTRACT FROM AUTHOR]