In order to correct the land surface temperature (LST) errors caused by angular effects to normalize the multisource data, various models have been proposed, especially the kernel-driven model shows good prospects for development. However, there are fewer studies on the validation of thermal radiation directionality (TRD) characteristics of features based on real measurements. In this article, a method based on unmanned aerial vehicle (UAV) observation of TRD characteristics of different surfaces is proposed to validate the TRD, which uses the circling flight mode to obtain the thermal radiation characteristics of features in multiple directions. Observations show that there is a significant hotspot effect in the three vegetation scenes, and the TRD and dispersion degree are different due to the differences in vegetation structure. For the realization of further applications of UAV-based methods for measuring TRD, the UAV-based angular observations are evaluated against the simulation results of the current kernel-driven model with high accuracy, and the results show that the correlation coefficients are all greater than 0.78, the coefficients of determination are greater than 0.60, and the root-mean-square error (RMSE) is less than 1. However, the ability of the model to extrapolate forward varies considerably in different vegetation scenes. The present study provides an effective and potential way to observe the TRD, which can lay a practical foundation for the theoretical study and model simulation of the TRD.

The accurate acquisition of land surface temperature (LST) on complex mountainous surfaces has always been a difficult problem and hot topic in thermal infrared remote sensing inversion, and the uncertainty caused by the radiation angle effect is one of the important factors hindering the accurate inversion of LST. Researchers have proposed a variety of models to simulate and eliminate the influence of the angle effect, among which, the kernel-driven model has a bright prospect of development. However, fewer studies have been conducted to observe the effects of terrain and land cover on thermal radiation directionality (TRD) properties based on measured data. This paper intends to carry out observations on a small spatial scale of a complex mountainous area using an unmanned aerial vehicle (UAV) to investigate the specific effects of different slopes, aspects, and land cover on the TRD characteristics. The measured results show that the intensity of thermal radiation anisotropy is actively correlated with the complexity of surface structure, and the influence of slope on TRD bias is in the form of a staged “S”, where its intensity increases slowly and then rapidly before slowing down again; the dispersion of thermal radiation in TRDs of different aspects is affected by the duration and intensity of solar radiation, and there is a time lag effect. Meanwhile, in order to evaluate the accuracy of different radiation directionality models, this paper further evaluates the currently more recognized kernel-drive model named LSF-Chen and the thermal equivalent slope kernel-driven (TESKD) model based on the TRD measurements from UAVs. The results show that the correlation coefficients of simulation results and measurements are all greater than 0.6, and the RMSEs are all less than 2K, and that the two methods both have a better simulation of the TRD effect, but the TESKD is better overall in terms of accuracy and methodological details. Through this study, a new method of applying UAVs to capture the thermal direction of the complex surface in mountainous areas is proposed, which provides methodological support for the extraction and accurate simulation of the TRD characteristics on the complex mountainous areas.

Spatiotemporally continuous Land Surface Temperature (LST) is crucial for monitoring extreme weather and providing disaster warnings. It captures abnormal temperature fluctuations, offering timely early warning and response for sudden climate events and natural disasters. However, cloud cover and satellite observation gaps often limit the spatial completeness of LST, while previous reconstruction methods seldom consider the effects of solar radiation and cloud cover on land surface temperature. To address these challenges, this study proposed the All-Weather Real Estimation (AWRE) method, which integrated thermal infrared and passive microwave data with environmental factors to estimate the LST under all-weather conditions. By incorporating deep learning and land surface energy balance models, and analyzing the impact of clouds on temperature fluctuations, the proposed method retrieves all-weather LST. Applied to the 2022 data of China, the AWRE method demonstrated high accuracy in estimating LST. The overall average RMSE and Bias were 2.90 K and 0.56 K, respectively, with daytime and nighttime RMSEs of 2.97 K and 2.83 K, respectively. Specifically, for daytime (nighttime) conditions, the RMSEs under clear sky were 2.94 K (2.58 K), partially cloudy 3.08 K (2.76 K), and fully cloudy 2.9 K (3.14 K). The estimated all-weather LST effectively captured diurnal and seasonal variations, with accuracy comparable to in-situ LST measurements, maintaining temporal continuity. This approach improves the detection of extreme heat events and addresses spatiotemporal coverage gaps, providing more accurate data for climate models, weather monitoring, and public health decisions.

Land surface temperature (LST) is crucial for the energy balance between the Earth’s surface and the atmosphere. Thermal infrared (TIR) and passive microwave (PMW) remote sensing are key methods for acquiring surface temperature globally and regionally. TIR observations have certain limitations due to their inability to penetrate cloud cover. Conversely, PMW measurements partially overcome this drawback to some extent, but their lower retrieval accuracy and coarse resolution limit its wider application. This study developed an artificial intelligence (AI) framework for precise and high-resolution LST estimation from PMW measurements, comprising PMW LST retrieval and downscaling components. Within this framework, high-resolution LST products have been obtained from Advanced Microwave Scanning Radiometer 2 (AMSR2), and the station-based validations and sensitivity analysis have also been conducted on the algorithm. The results were given as follows. First, the GeoFusionNet algorithm achieved higher LST retrieval accuracy than empirical or physical models. The mean absolute error (MAE) was 2.37 K (1.60 K) during daytime (nighttime). Second, the downscaled PMW LST retained high accuracy, with a daytime (nighttime) MAE increase of 0.28 K (0.14 K) compared to the Moderate Resolution Imaging Spectroradiometer (MODIS) 1-km product. Station-based validations showed that the coefficient of determination$R^{2}$was above 0.9, with an average root-mean-squared error (RMSE) of 3.4 K (2.4 K) for daytime (nighttime) and an MAE of 2.80 K (1.98 K). Third, sensitivity analysis demonstrated the algorithm’s stable performance, especially in summer and autumn. Spatially, the accuracy remained within 3 K for various land types, including cropland, evergreen forests, and deciduous forests. These results indicate that PMW LST retrieved by this framework has sufficient accuracy and fine-spatial resolution for monitoring dynamic changes in large-scale hydrological, climatic, and agricultural fields.

Land surface component temperature (LSCT) is a vital parameter in many remote sensing applied fields such as drought monitoring and evapotranspiration. Previous large-scale two-source (vegetation and soil) LSCT retrieval method seldom considered the hot-spot effect of soil, thereby failing to account for the observed higher temperatures when the view direction is closer to the sun. Thus, an algorithm using both Sea and Land Surface Temperature Radiometer (SLSTR) and the Advanced Himawari Imager (AHI) dataset was performed and validated. The validation dataset was collected from Daman, China with underlying of forest and crop. The validation results demonstrated that the proposed method, which considers the hot-spot effect, significantly improves the accuracy of LSCT estimation, as indicated by root mean square error (RMSE) values of 2.58K and 3.34K for crops and trees, respectively. This study paving the way for improved understanding and applications of LSCT estimation account for the hot-spot effect in real-world scenarios.

The validation of remote sensing land surface temperature (LST) data necessitates a comparison between satellite retrieval outcomes andin situobservations. The efficiency ofin situobservations can be ameliorated via analysis and modeling, whereby the heterogeneity ofin situobservations on temporal and spatial scales is central to the analysis. A fresh algorithm has been developed to optimize deployment by relying on the standard deviation of spatial heterogeneity. The validation outcomes indicated that the coefficient of determination (R2) of the five typical surface features at three time points was 0.66, with a root mean square error (RMSE) of 1.99 °C and a mean absolute error (MAE) of 1.62 °C. Moreover, the spatiotemporal heterogeneity character of typical surface features displayed different features, and the LST variation curves of each typical surface feature displayed a similar pattern under sunny conditions. The application of the Savitzky–Golay filtering method reduced errors by 4% of the total errors caused by random errors inin situobservations. With the analysis of the spatiotemporal characteristics of in-situ observation. First, the number of required sites algorithm computed a minimum sampling number of 4. Second, the analysis of the means algorithm computed the 5 optimal points. Additionally, the multipointin situobservations were regularized by standard scores. The optimization of the selected points could be executed to improve the results by eliminating the "distance" points, which are located further away from the multipointin situobserved LST statistical mean. Our outcomes will deepen the comprehension of the spatiotemporal character ofin situobserved LST and enhance the efficiency of equipment with equivalent accuracy.

Accurate estimates of downward surface shortwave radiation (DSSR) are critical for hydrological, biogeochemical, and ecological studies and remote sensing-based estimation of DSSR is an important way to derive DSSR at different spatio-temporal ranges. However, current estimation algorithms usually somewhat rely on atmospheric parameters or in-situ measurements, further blocking the application of these methods. Inspired by the emerging DSSR reanalysis data from the model simulation, this study proposed an integrated method by initializing the estimation model with ERA5 reanalysis data and further refining the estimation through iterative training. The random forest regression method was applied in the estimation model to build the connection between DSSR with the MODIS top-of-atmosphere reflectance, cloud flag, geometry information, elevation, latitude, and coefficient of Sun-Earth distance as input features. To separately consider the impact from cloud cover, the estimation model was established for clear-sky and cloudy-sky conditions, respectively. The proposed method was applied to estimate instantaneous DSSR of MODIS daytime overpasses in the Southwest part of China in 2020. Comparison between the estimates of the initialized model and the finalized model shows that the iterative process improves the DSSR estimates on both spatial distribution and accuracy. Validated by the measurements from nine sites in the study area, the DSSR estimates of the finalized model show a 0.02 higher correlation coefficient (CC) and 7.35 W m-2lower root mean squared error (RMSE) than that of the initialized model. To better evaluate the performance of the proposed method, three popular DSSR products including ERA5, MCD18A1, and Himawari-8 were introduced to make an inter-comparison with the estimation of this study. The validation results showed that the all-sky DSSR estimated in this study had the best accuracy, with a CC of 0.90, a mean bias error of 37.80 W m-2, a RMSE of 125.30 W m-2, and a relative root mean squared error of 42.73%. Obvious improvements can be observed under cloudy-sky and clear-sky conditions, respectively. Because of the simplicity and reliable performance of the proposed method, it shows good potential for DSSR estimation.

Land surface temperature (LST) is a crucial parameter in the study of Land Surface processes. Currently, there are great progresses in LST retrieval based on thermal infrared (TIR) remote sensing. However, TIR-based LST suffers from serious spatial discontinuities due to clouds. Although there are methods developed to address this issue, the methods show high uncertainty in days with extremely clouds. Therefore, this study proposed an integrated method to reconstruct cloudy LSTs using Terra Moderate Resolution Imaging Spectroradiometer (MODIS) and the China Land Data Assimilation System (CLDAS) LST. This method was separated into two parts according to the ratio of clear-sky pixels (RCP). On days with RCP more than 30%, a random forest reconstruction method was used to establish the complicated relationship between LST and its predicting variables, including solar radiation factor, vegetation index, water index, topographic information and latitude, and then applied to cloudy pixels to derive LSTs. For the rest days, the CLDAS LST was selected to assist the reconstruction via downscaling it to 1 km and then merged with clear-sky data to generate spatially continuous results. The proposed method was applied to the Southwest China and generate daily LST product in 2019. Validation with ground measurements demonstrated a high accuracy with the correlation coefficient changing from 0.73 to 0.88. Additionally, the reconstructed LST dataset exhibits similar temporal variability as existing all-weather satellite-based and reanalysis LST products. The findings reveal that this method shows good potential in generating gap-free LST dataset, especially for the mountain regions with heavy clouds.

Many achievements have been made by studying how to visualize large-scale and high-dimensional data in typically 2D or 3D space. Normally, such a process is performed through a non-parametric (unsupervised) approach which is limited in handling the unseen data. In this work, we study the parametric (supervised) model which is capable to learn a mapping between high-dimensional data space Rd and low-dimensional latent space Rs with similarity structure in Rd preserved where s l d. The GNNVis is proposed, a framework that applies the idea of Graph Neural Networks (GNNs) to the parametric learning process and the learned mapping serves as a Visualizer (Vis) to compute the low-dimensional embeddings of unseen data online. In our framework, the features of data nodes, as well as the (hidden) information of their neighbors are fused to conduct Dimension Reduction. To the best of our knowledge, none of the existing visualization works have studied how to combine such information into the learning representation. Moreover, the learning process of GNNVis is designed as an end-to-end manner and can easily be extended to arbitrary Dimension Reduction methods if the corresponding objective function is given. Based on GNNVis, several typical dimension reduction methods t-SNE, LargeVis, and UMAP are investigated. As a parametric framework, GNNVis is an inherently efficient Visualizer capable of computing the embeddings of large-scale unseen data. To guarantee its scalability in the Training Stage, a novel training strategy with Subgraph Negative Sampling (SNS) is conducted to reduce the corresponding cost. Experimental results in real datasets demonstrate the advantages of GNNVis. The visualization quality of GNNVis outperforms the state-of-the-art parametric models, and is comparable to that of the non-parametric models.

Data change the game in terms of how we respond to pandemics. Global data on disease trajectories and the effectiveness and economic impact of different social distancing measures are essential to facilitate effective local responses to pandemics. COVID-19 data flowing across geographic borders are extremely useful to public health professionals for many purposes such as accelerating the pharmaceutical development pipeline, and for making vital decisions about intensive care unit rooms, where to build temporary hospitals, or where to boost supplies of personal protection equipment, ventilators, or diagnostic tests. Sharing data enables quicker dissemination and validation of pharmaceutical innovations, as well as improved knowledge of what prevention and mitigation measures work. Even if physical borders around the globe are closed, it is crucial that data continues to transparently flow across borders to enable a data economy to thrive, which will promote global public health through global cooperation and solidarity.