Ground-based multi-channel microwave radiometers (MWRs) can continuously detect atmospheric profiles in the tropospheric atmosphere. This makes MWR an ideal tool to supplement radiosonde and satellite observations in monitoring the thermodynamic evolution of the atmosphere and improving numerical weather prediction (NWP) through data assimilation. The analysis of product characteristics of MWR is the basis for applying its data to real-time monitoring and assimilation. In this paper, observations from the latest generation of ground-based multi-channel MWR RPG-HATPRO-G5 installed in Shanghai, China, are compared with the radiosonde observations (RAOB) observed in the same location. The detection performance, characteristics of various channels, and the accuracy of the retrieval profile products of the MWR RPG are comprehensively evaluated during various weather conditions. The results show that the brightness temperatures (BTs) observed by the ground-based MWR RPG during precipitation conditions were high, which affected its detection performance. The bias and the standard deviation (SD) between the BT observed by MWR RPG and the simulated BT during clear and cloudy sky conditions were slight and large, respectively, and the coefficient of determination ($R^2$) was high and low, respectively. However, when the cloud liquid water (CLW) information was added when simulating BT, the bias and the SD of the observed BT and the simulated BT during cloudy days were reduced and the $R^2$ value improved, which indicated that CLW information should be taken into account when simulating BT during cloudy conditions. The temperature profiles of the MWR retrieval had the same accuracy of RMSEs (root-mean-square error) with heights during both clear-sky and cloudy sky conditions, where the RMSEs were below 2 K when the heights were below 4 km. In addition, the MWR RPG has the potential ability to retrieve the temperature inversion in the boundary layer, which has important application value for fog and air pollution monitoring.

To recover the actual responsivity for the Ultraviolet Multi-Filter Rotating Shadowband Radiometer (UV-MFRSR), the complex (e.g., unstable, noisy, and with gaps) time series of its in situ calibration factors ( V ₀ ) need to be smoothed. Many smoothing techniques require accurate input uncertainty of the time series. A new method is proposed to estimate the dynamic input uncertainty by examining overall variation and subgroup means within a moving time window. Using this calculated dynamic input uncertainty within Gaussian process (GP) regression provides the mean and uncertainty functions of the time series. This proposed GP solution was first applied to a synthetic signal and showed significantly smaller RMSEs than a Gaussian process regression performed with constant values of input uncertainty and the mean function. GP was then applied to three UV-MFRSR V ₀ time series at three ground sites. The method appropriately accounted for variation in slopes, noises, and gaps at all sites. The validation results at the three test sites (i.e., HI02 at Mauna Loa, Hawaii; IL02 at Bondville, Illinois; and OK02 at Billings, Oklahoma) demonstrated that the agreement among aerosol optical depths (AODs) at the 368 nm channel calculated using V ₀ determined by the GP mean function and the equivalent AERONET AODs were consistently better than those calculated using V ₀ from standard techniques (e.g., moving average). For example, the average AOD biases of the GP method (0.0036 and 0.0032) are much lower than those of the moving average method (0.0119 and 0.0119) at IL02 and OK02, respectively. The GP method's absolute differences between UV-MFRSR and AERONET AOD values are approximately 4.5 %, 21.6 %, and 16.0 % lower than those of the moving average method at HI02, IL02, and OK02, respectively. The improved accuracy of in situ UVMRP V ₀ values suggests the GP solution is a robust technique for accurate analysis of complex time series and may be applicable to other fields.

High-resolution three-dimensional (3D) wind field data are critical for a wide range of applications, including wind energy assessment, low-altitude aviation, air quality modeling, and extreme weather forecasting. Although ERA5 reanalysis remains widely used, its relatively coarse spatial resolution (∼ 31 km) limits its ability to capture local-scale atmospheric processes. To address this, this study develops an hourly 3D dynamic wind field dataset with 1 km horizontal resolution covering the Yangtze River Delta (YRD) region during the summer months (June–August) from 2021 to 2023, namely YRD1km, generated through advanced dynamical downscaling of ERA5 using a customized Weather Research and Forecasting (WRF) model configuration. The methodology integrates multi-source observational nudging with high-resolution land use parameterization to enhance near-surface wind accuracy and terrain-induced flow representation, particularly in urban clusters and mountainous areas. Validation against ground-based observations confirms the superior performance of YRD1km over ERA5 for hourly 10 m wind components, with Mean Absolute Error (MAE) reduced by 21.61 % for U and 26.04 % for V, Root Mean Square Error (RMSE) reduced by 18.30 % for U and 22.63 % for V, and Nash–Sutcliffe Efficiency (NSE) improved by 33.27 % and 40.13%, respectively. On a daily mean basis, both MAE and RMSE are reduced to below 0.5 m s⁻¹, and NSE reaches approximately 0.88. Spatially, YRD1km captures finer spatial wind speed gradients and localized terrain-induced circulations that are not captured by ERA5. Temporally, consistent accuracy improvements with approximately 20 % lower hourly error variability are seen when compared to ERA5. Vertically, 42.18 % accuracy gains are observed in the near-surface layer when compared with radiosonde profiles. Moreover, convective case analyses indicate that YRD1km captures vertically coherent wind structures across multiple tropospheric levels that are closely linked to the initiation and maintenance of deep convection, highlighting its diagnostic advantage in high-impact weather events. Overall, the YRD1km 3D wind field dataset and its integrated methodological framework provide a robust foundation for regional meteorological applications, including high-resolution AI-based forecasting, renewable energy planning, and weather risk management in rapidly developing regions such as the YRD. The YRD1km 3D wind field dataset is available at https://doi.org/10.57760/sciencedb.23752 (Zhang and Liu, 2025).

The observation error covariance (R) matrix is a key component in the data assimilation (DA) process for retrieval of atmospheric state parameters (ASPs), also impacting the subsequent numerical weather forecast. However, one commonly used type of R matrix depends on instrument noise, which contravenes reality because the retrieved ASPs would depend on the instrument used. Other types of R matrix rely on the observation operator (H), analyzed state ( $x_a$ ), background error covariance (B) matrix or the background state ( $x_b$ ), and the selected forecast ensemble. All these dependences reduce the representativeness of the R matrix, since the correctness of H needs verification and no true values exist for $x_a$ or $x_b$ . As such, a better method to correctly specify the R matrix is needed. Through the physical mechanism occurring between incident radiation and particles in the atmosphere, which complies with the phenomena of energy absorption and emission, correlations among bands or channels in a detected atmospheric radiance spectrum occur. This paper thus proposes a data-derived R matrix based on a large number (N) of detected atmospheric radiance spectra constructed from N real-time measurements, where N real-time measurements can be acquired by staring at some observation location of interest during a short amount of time. This data-derived R matrix for satellite radiance observations does not rely on any assumed quantities and is unambiguous. Technically, recording N real-time measurements is achievable by modifying the trigger configuration of data recording from ground.