To support efficient and accurate multi-task inference in edge environments, we propose a task-oriented broadcast communication system that enables an edge transmitter to serve multiple edge devices with heterogeneous inference tasks. The proposed system adopts a two-phase design inspired by Marton’s channel coding with rate splitting and the information bottleneck principle. In the first phase, a common feature vector is extracted to capture the shared information across tasks. In the second phase, task-specific private feature vectors are generated conditioned on the common feature to preserve unique task-relevant information. To facilitate interference-robust task execution, our scheme leverages the intrinsic structural alignment between the task correlations and broadcast channel properties; specifically, the common and private features are mapped directly to Marton’s common and private codewords. A variational approximation method is introduced to optimize the feature extraction process in both phases, allowing for compact and informative representations while reducing redundant data transmission. Extensive experiments on a real-world multi-label dataset demonstrate that the proposed method achieves superior inference accuracy and robustness over wireless networks, compared to traditional digital compression and deep learning-based joint source-channel coding schemes. These results confirm the potential of task-oriented design for scalable and reliable edge intelligence.

Satellite microwave remote sensing is acknowledged as the primary method for obtaining global-scale surface soil moisture (SSM) data. Typically, spaceborne microwave sensors aboard polar-orbiting satellites yield SSM estimates with a native spatial resolution of several tens of kilometers and a temporal resolution of about once or twice daily. This indicates considerable potential for enhancing both of their spatial and temporal resolutions. Although spatial downscaling of spaceborne microwave SSM has garnered significant attention recently, the enhancement of their temporal resolution has received less focus. This study pioneers a methodology for generating hourly-scale satellite SSM estimates. The developed approach integrates a novel blending module that combines geostationary satellite observations with a spatially downscaled SSM dataset derived from traditional fusion among microwave and optical observations on polar-orbiting platforms. This blending module leverages land surface temperature (LST) data from geostationary satellites, which effectively quantify SSM variations on an hourly interval upon the thermal inertia theory. Consequently, a comprehensive spatio-temporal integrated framework for SSM downscaling is established to produce hourly-scale SSM at a resolution of 6 km, enhancing upon the daily and 36-km resolutions of existing microwave SSM datasets. Validation of the downscaled hourly SSM estimates was conducted through an established ground soil moisture observatory network in North China, revealing an unbiased root mean square error (ubRMSE) of no higher than 0.04 cm³/cm³. This result confirms preservation of the fundamental accuracy of original microwave SSM retrievals and demonstrates the effectiveness of the developed framework in improving both spatial and temporal representativeness of SSM data.

Deploying intelligent service and executing inference tasks in the proximity of the edge enable models to access enormous real-time data generated by the edge devices. However, the dilemma of fulfilling service demands with limited resources at edge devices impairs the efficacy of conventional data-oriented communication systems. To achieve a better trade-off between inference accuracy and communication overhead, in this paper, we propose a dynamic unmanned aerial vehicle (UAV)-assisted cooperative edge inference system, where a UAV acts as an edge server to aggregate the wide-view features from mobile sensors through Over-the-Air computation (AirComp) to complete the inference task cooperatively. Discriminant gain, an effective indicator for the inference accuracy, is adopted to realize task-oriented design. To exploit channel diversity and data diversity in the multi-device cooperative edge inference system, we maximize the discriminant gain of the AirComp feature aggregation by jointly optimizing the UAV trajectory and the power allocation policy with respect to the different important levels of feature dimensions. An alternating algorithm and a successive convex approximation (SCA)-based method are then proposed to solve the optimization problem. Numerical simulations further validate the efficacy of the proposed design compared to the baselines.

While fusion of hyperspectral images (HSIs) with low spatial resolution and multispectral images (MSIs) with high spatial resolution has achieved significant success, high-quality fusion between noisy images has always been challenging. In this article, we propose a domain transform model driven by deep learning for anti-noise hyperspectral and multispectral image fusion (DTAFN). This marks the first time that wavelet decomposition theory is combined with deep learning for noise reduction in hyperspectral and MSI fusion. DTAFN initially decomposes hyperspectral and MSIs into frequency components and constructs a novel feature interaction fusion module (FIFM). This module, while using MSIs to guide the removal of noise from HSIs, also achieves the fusion of spatial and spectral information. Furthermore, it maps the fused features to a lower dimensional subspace to enhance computational efficiency. Additionally, we introduce a spatial-spectral self-attention mechanism to optimize the reconstructed frequency components using the subspace features. In the end, the wavelet inverse transform is used to reconstruct the clean fused image. It is worth noting that the extraction of the subspace is considered a process of nonlinear low-rank component extraction, which, to a certain extent, suppresses noise signals. Numerous experiments of mixed noise image fusion are carried out, and the experimental results show that DTAFN can obtain high-quality fusion results, is robust, and superior to the state-of-the-art methods.

Over ten-years, the Integrated Calibration and Validation System (ICVS) Long-Term Monitoring (LTM) System has provided near-real time (NRT) monitoring for Joint Polar Satellite System (JPSS) spacecraft and instruments including their on-orbit status and performance and science data product quality [1] - [4]. The ICVS also harnesses JPSS Sensor Data Record (SDR) data to rapidly (with little latency) visualize radiometric features of severe weather events such as hurricanes and volcanos [5] [6]. This study presents two case studies, one depicting the 3-dimensional (3D) atmospheric warm core structure inside Hurricane Ian from the 2022 North Atlantic Hurricane Season and another showing the 3D temperature structures present during the 2021 Heat Dome event by using JPSS ATMS (and VIIRS for hurricane events) SDR and TDR data. More details and images/animations for hurricane events can be found at https://www.star.nesdis.noaa.gov/smcd/sew/index.php.

Image registration aims to eliminate the geometric deviation between multi-source data with the same range, and to promote the collaborative application of data. In recent years, spaceborne hyperspectral (HS) and multispectral (MS) data have been widely used in Earth observation. However, the difference in the number of bands, spatial resolution, and spectral resolution puts forward higher requirements on the registration algorithm. The key to HS and MS image registration is to extract more common key points, weaken and eliminate the difference of radiation and spatial texture information to build superior descriptors, and achieve high-precision matching of key points. This paper introduces a new robust HS and MS registration method based on common deep feature subspaces. We first construct the common deep feature subspaces extraction network to extract consistent edge features and common subspace images of the image pair. Then, Harris algorithm is used to extract key points from consistent edge features between images, which reduces the impact of spatial resolution differences between images. Besides, the SIFT descriptor and subspace images are used to describe key points, which reduces the impact of radiation differences between images. Finally, Euclidean distance is used for the initial matching of key points, and the affine matrix is calculated after the outliers are eliminated, and image registration is performed. We perform experiments on spaceborne HS and MS datasets of different spatial resolutions and comparisons with state-of-the-art methods. Experimental results show that our method can obtain satisfactory registration results and is robust.

The maintenance of spectral variability between subclass objects and the relationship between hyperspectral (HS) bands have been a fundamental but challenging problem for fusing low spatial resolution (LR) HS and high spatial resolution (HR) multispectral (MS) images. This article presents a locally optimized image segmentation fusion (LOISF) framework for HS super-resolution reconstruction. First, LR HS and HR MS are clustered and segmented, and the label attributes of the segmented objects are identified by the prior information. Then, a novel joint fusion model for different typical ground objects is constructed based on spectral unmixing. The fusion problem is formulated mathematically as a convex optimization of a Frobenius norm, which includes spatial, spectral, and index constraints, with an alternating-directions’ optimization featuring linearization providing the solution. Experimental results demonstrate that the proposed LOISF preserves both spatial details and texture, achieving high spectral fidelity, and yielding significantly improved image quality compared to other state-of-the-art fusion methods.

The capacity of combined gauge-satellite precipitation estimates largely depends on the characteristics of the input data such as the number, location and reliability of rain gauges, and satellite-derived precipitation quality. The objective of this study is to examine the influence of rain gauge network configuration including density and spatial distribution on the performance of the gauge-satellite merging estimation at monthly and ten-day temporal scales. Dense rain gauge observations and satellite-derived precipitation data (i.e., TMPA 3B42 Version 7 and Version 06 IMERG Final Run) in two provinces of China are used. A two-stage downscaling-integration approach is applied in the gauge-satellite precipitation estimation. Various scenarios of rain gauge density and combination are designed and their corresponding merged precipitation estimates are evaluated using statistical indices. The merged results using the TMPA and IMERG precipitation product, respectively, are compared. The results show that: 1) the influence of rain gauge network configuration on the gauge-satellite merged precipitation estimates gradually decreases with the increase in rain gauge density, and the gauge-satellite merged precipitation estimates are more sensitive to the rain gauge network density in wet season and ten-day temporal scale than in dry season and monthly scale, respectively and 2) the merged precipitation estimation using the IMERG precipitation data generally outperforms the estimation using TMPA precipitation data in the low gauge density scenarios, and the gap decreases with the increase in the rain gauge network density. In the areas with sparse rain gauges, improving the quality of satellite precipitation data would significantly improve the performance of the gauge-satellite merging estimation.

Fusing low spatial resolution (LR) hyperspectral (HS) data and high spatial resolution (HR) multispectral (MS) data aims to obtain HR HS data. However, due to bad weather and the aging of sensor equipment, HS images usually contain a lot of noise, e.g., Gaussian noise, strip noise, and mixed noise, which would make the fused image have low quality. To solve this problem, we propose the multiscale low-rank deep back projection fusion network (MLR-DBPFN). First, HS and MS are superimposed, and multiscale spectral features of the stacked image are extracted through multiscale low-rank decomposition and convolution operation, which effectively removes noisy spectral features. Second, the upsampling and downsampling network mechanisms are used to extract the multiscale spatial features from each layer of spectral features. Finally, the multiscale spectral features and multiscale spatial features are combined for network training, and the weight of the noisy spectrum features is reduced through the network feedback mechanism, which suppresses the noisy spectrum and improves the noisy HS fusion performance. Experimental results on datasets of different noise demonstrate that MLR-DBPFN has superior spatial and spectral fidelity, comparative fusion quality, and robust antinoise performance compared with state-of-the-art methods.

Synthetic aperture radar (SAR) can be used to obtain remote sensing images of different growth stages of crops under all weather conditions. Such time-series SAR images can provide an abundance of temporal and spatial features for use in large-scale crop mapping and analysis. In this study, we propose a temporal feature-based segmentation (TFBS) model for accurate crop mapping using time-series SAR images. This model first extracts deep-seated temporal features and then learns the spatial context of the extracted temporal features for crop mapping. The results indicate that the TFBS model significantly outperforms traditional long short-term memory (LSTM), U-network, and convolutional LSTM models in crop mapping based on time-series SAR images. TFBS demonstrates better generalizability than other models in the study area, which makes it more transferable, and the results show that data augmentation can significantly improve this generalizability. The visualization of the temporal features extracted by the TFBS shows that there is a high degree of intraclass homogeneity among rice fields and interclass heterogeneity between rice fields and other features. TFBS also achieved the highest accuracy of the four deep learning models for multicrop classification in the study area. This study presents a feasible way of producing high-accuracy large-scale crop maps based on the proposed model.

The NOAA-20 Ozone Mapping and Profiler Suite (OMPS) is the second OMPS flight unit in the US Joint Polar Satellite System (JPSS) program [1]. Flying on the NOAA-20 satellite, the OMPS extends the 40+-year total column ozone and ozone profile records to monitor ozone concentration in the Earth atmosphere. Recent studies have been conducted over a wide range of radiometric calibration and geolocation calibration areas to improve the quality of OMPS sensor data records. This study covers several topics, including the removal of a solar intrusion signal during science data collection, reduction of solar activity impact resulting from solar activities, update of geolocation accuracy via refinement of the instrument CCD spatial registration, and minimization of a misalignment at the edges of the sensor spatial field of view during Earth view measurements. The objective of this study is to improve OMPS instrument performance to OMPS users' expectations, and to advance the OMPS calibration accuracy of sensor data records above current product requirements [2], [3]. The solar intrusion correction improved local radiance retrieval accuracy 4% for the Nadir Profiler sensor, during science observations in the Northern hemisphere where solar zenith angle ranges from$60^{\circ}$to$88^{\circ}$. Solar model improvements achieve up to 1% better fidelity in irradiance measurements during solar observations. The calibrated CCD spatial registration minimizes the misalignment in the Nadir Mapper sensor spatial field of view in both cross track and along track measurements, and refines the geolocation accuracy to less than 5 km for the sensor's geolocation products that have a nominal spatial resolution of 50 km x 17 km.

To provide near-real time monitoring of Suomi-National Polar-orbiting Partnership (S-NPP) and NOAA-20 data inter-sensor biases, this study extends the Geosynchronous Equatorial Orbit - Low-Earth Orbit (GEO-LEO) intercalibration method established in [1] to the Visible Infrared Imaging Radiometer Suite (VIIRS) Sensor Data Record (SDR) data at six reflective solar bands (RSBs) and three thermal emissive bands (TEBs) bands via the STAR Integrated Calibration and Validation System (ICVS) framework. The Geostationary Operational Environmental Satellite (GOES) Advanced Baseline Imager (ABI) is used as a transfer to calculate double difference (DD) of VIIRS-ABI Simultaneous Nadir Overpass (SNO) pairs for the S-NPP and NOAA-20 VIIRS SDR cross-calibration biases. A series of sensitivity analyses on the dependence of the results to view geometries, spectral difference corrections, latitudinal variation, and cloud screening are conducted for more accurate bias estimations. The findings are further verified by the other independent approaches, namely the 32-day average difference method (32Day-AD) [2] and DD with radiative transfer model as a transfer (RTM-DD) method.

Specular reflectance is an important error source in the remote retrieval of leaf pigment content. However, removing this disturbance is challenging because the specular component cannot be directly separated from the leaf reflectance. In this paper, we removed the specular reflectance through polarization measurements at single leaf scale. First, polarization measurements were taken on leaves in the nadir and an oblique viewing direction. Specular reflectance in these two directions was estimated. Second, leaf specular reflectance was subtracted from the total reflectance. Based on the consequent diffuse reflectance, three common types of vegetation indices (VIs), simple ratio (SR), normalized difference (ND), and red edge position (REP) were built to retrieve pigment content. The results show that: 1) specular reflectance interfered with the retrieval. The greater this component, the more serious it caused the disturbance. The retrieval accuracy in forward 35° was much lower than that in the nadir. 2) After specular reflectance removal, the retrieval in both directions was generally improved for the SR and the ND VIs. The improvement in forward 35° was remarkable, with accuracy reaching or approaching the corresponding nadir level. On the two pigments, improvement for chlorophylls was more apparent than for carotenoids. 3) Of all VIs investigated, the REP was insensitive to specular reflectance and had the best resistance capability. Overall, the proposed methods can effectively eliminate specular interference and improve pigment content retrieval. Even under disadvantageous circumstances with strong specular reflectance, the improvement was still reliable.

The land parameter retrieval model (LPRM) has been widely used for the retrieval of microwave-based global surface soil moisture (SSM). In the original LPRM algorithm, soil surface effective temperature (SSET) was retrieved from the Ka-band (36.5 GHz) brightness temperature (BT) and then used as input data for the estimation of SSM. In this study, SSET retrieved from the Advanced Microwave Scanning Radiometer of the Earth Observing System Radiometer (AMSR-E) Ka-band BT was validated against MODIS land surface temperature (LST) data, and the results show that when the fraction of water surface (FWS) within an AMSR-E pixel increases in the range from 0-0.01 to 0.15-0.4, the root-mean-square error (RMSE) increases drastically from 1.98 to 11.42 Kelvin (K). When validation was conducted on SSET data retrieved with the K-band (18.7 and 23.8 GHz) BT, the RMSEs were maintained below 2.0 K for FWS ranging from 0 to 0.4. Therefore, in the original LPRM, we substituted the Ka-band BT with the K-band BT to produce more accurate estimations of SSET and hence improved SSM retrievals. Our results show that for FWS ranges of 0-0.01, 0.01-0.05, 0.05-0.15, and 0.15-0.4, the correlation coefficients (R -values) between SSM retrieved with the improved LPRM and Global Land Data Assimilation System (GLDAS) measurements are 0.78, 0.75, 0.77, and 0.46, respectively, compared with the corresponding R -values of 0.78, 0.65, 0.56, and 0.34 for SSM retrieved with the original LPRM. This demonstrates that against the original LPRM, our improved LPRM can produce more accurate SSM retrievals under water-land mixed pixels.

This paper evaluates the capability of Sentinel-1 VH-polarized synthetic aperture radar signals, involving 738 scenes in the interferometric wide swath (IW) mode, for ocean surface wind speed retrieval using a novel subswath-based C-band cross-polarized ocean model. When compared with in situ measurements, it is observed that wind speed retrieval accuracy varies progressively along swath, with the most accurate wind speed retrievals being derived from subswath 3 [root-mean-square error (RMSE) of 1.82 m · s-1], followed by subswath 2 (RMSE of 1.92 m · s-1), while subswath 1 showed the lowest retrieval accuracy (RMSE of 2.37 m · s-1). The average RMSE of wind speeds retrieved from all the three subswaths is 2.08 m · s-1under low-to-high wind speed regimes (wind speeds-1). We further observed that the dependence of VH-polarized normalized radar cross section (NRCS) on incidence angle is attributable to the high and changing noise equivalent sigma zero (NESZ) with incidence angle under low-to-moderate wind speed regimes. And that strong VH-polarized radar signals could overcome the NESZ effect, thereby eliminating the dependence of VH-polarized NRCS on incidence angle under strong wind conditions. For Sentinel-1 IW mode VH-polarized data, the effect of NESZ could be ignored when wind speeds are greater than 15 m · s-1, as a better wind speed retrieval performance of these data has been recorded in this paper at wind speeds greater than 10 m · s-1, owing to an RMSE below 1.6 m · s-1and biases ranging from -0.5 to 0.5 m · s-1.

High-spatial-resolution wind fields derived from synthetic aperture radar (SAR) instruments are crucial to a wide range of applications. However, the spatial scale effect on wind speed retrieval accuracy has seldom been reported. For the purpose of understanding this issue, this letter makes a quality assessment of wind speed retrieval accuracy based on four commonly used C-band geophysical model functions (CMOD4, CMOD-IFR2, CMOD5, and CMOD5.N) at spatial resolutions ranging from 50 m to 50 km using Sentinel-1 interferometric wide (IW) swath mode images. Our results show that the CMOD5 function is the most effective among these functions, owing to a low root-mean-square error (RMSE) of 1.17 m/s and a bias of -0.28 m/s for wind speed retrieval at a spatial resolution of 500 m. It is further observed that the variance of wind speeds retrieved from copolarized SAR images decreases exponentially with the decrease of spatial resolutions. For Sentinel-1 IW mode images, the variance of wind speeds retrieved with CMOD5 decreases rapidly from 50 to 500 m, with a drop in RMSE of 40%, and thereafter levels off. Thus, a spatial resolution of 500 m, with the CMOD5 function, is recommended optimal in this letter, for wind speed retrieval using Sentinnel-1 IW mode data.

The Suomi National Polar-orbiting Partnership (SNPP) Vis ible Infrared Imaging Radiometer (VIIRS) instrument was launched on October 28, 2011 to provide operational environmental monitoring which includes exceptional events such as dust storms, smoke, urban haze, and volcanic ash. The VIIRS aerosol products include pixel-level (~750m) aerosol optical thickness (AOT) and aerosol detection which includes dust and smoke mask/aerosol index. The VIIRS aerosol product resolutions and its large swath width (~3000 km) provide an unprecedented capability to observe small-to-large scale exceptional events that impact human health and economy. The VIIRS aerosol products have a demonstrated accuracy, precision, and measurement range to reliably observe and track exceptional events and aid operational air quality forecasters in providing warnings and alerts to the public. This presentation will illustrate VIIRS aerosol product capabilities and limitations using examples of exceptional events such as a multi-day smog blanketing China (November-December 2015), long range transport of smoke from forest fires in Alaska and Canada impacting regional air quality in the northeast United States (June 2015), and long range transport of dust from Africa impacting air quality in the south eastern United States (June 2015). While the aerosol imagery is very valuable to forecasters in providing the regional extent of an exceptional event (e.g., upwind conditions and their impact downwind), drawing quantitative estimates of surface concentrations of particles smaller than 2.5 μm in diameter (PM2.5, μg/m3) from AOT often times requires additional information or assumptions. The Chinese smog case where the observed VIIRS AOT values did not necessarily reflect the high surface PM2.5 values (~600 μg/m3) observed by ground stations and the identification of aerosol type in the smog as smoke due to the presence of highly absorbing brown carbon will be used as an example to demonstrate the strengths and limitations of applying remotely sensed aerosol products to air quality monitoring and forecasting.

This paper presents a comparison strategy for investigating the influence of spatial resolutions on sea surface wind speed retrieval accuracy with cross-polarized synthetic aperture radar images. First, for wind speeds retrieved from vertical transmitting-vertical receiving (VV)-polarized images, the optimal geophysical C-band model (CMOD) function was selected among four CMOD functions. Second, the most suitable C-band cross-polarized ocean (C-2PO) model was selected between two C-2POs for the VH-polarized image data set. Then, the VH-wind speeds retrieved by the selected C-2PO were compared with the VV-polarized sea surface wind speeds retrieved using the optimal CMOD, which served as a reference, at different spatial resolutions. Results show that the VH-polarized wind speed retrieval accuracy increases rapidly with the decrease in spatial resolutions from 100 to 1000 m, with a drop in root-mean-square error of 42%. However, the improvement in wind speed retrieval accuracy levels off with spatial resolutions decreasing from 1000 to 5000 m. This demonstrates that the pixel spacing of 1 km may be the compromising choice for the tradeoff between the spatial resolution and wind speed retrieval accuracy with cross-polarized images obtained from RADASAT-2 fine quad-polarization mode.

The Visible Infrared Imaging Radiometer Suite (VIIRS) is an instrument on board S-NPP that was launched on October 28, 2011. A second VIIRS instrument will be on the next Joint Polar Satellite System (JPSS) satellite (J-1) that is planned to be launched in early 2017. VIIRS is a scanning radiometer that consists of 22 spectral channels with band centers from 412 nm to 12,050 nm. It collects imagery and radiometric measurements of the land, atmosphere, cryosphere, and oceans. The VIIRS data processing is done at NOAA Interface Data Processing Segment (IDPS) which is also responsible for creating all other atmospheric, land and ocean products from the other sensors on board S-NPP. The testing and integration of algorithm and Look-Up-Table (LUT) updates are done by using Algorithm Development Library, a test system that mimics the operational system. In this work we will present the testing and integration work conducted for Aerosol Environmental Data Record (EDR) Algorithm improvements. As Aerosol EDR is used in many downstream products, we will present the effect of Aerosol EDR updates on those products. We will also discuss the effect of changes to upstream VIIRS Sensor Data Record (SDR) and Cloud Mask algorithms on the Aerosol EDR products.

In this study, we plot a 3D aerosol map over China mainland based on CALIPSO and MODIS aerosol product combined, which is height-resolved. We can see there are several hot spots in terms of aerosol loadings across China, such as Pearl River Delta, Yangtze River Delta. Also, the maximum height that aerosol can reach were figured out based on the aerosol vertical profiles. Generally, this height is consistent with the planetary boundary layer height, less than 2km. Meanwhile, we attempted to sort out the aerosol indirect effect on cloud, by plotting Contoured Frequency by Altitude Diagram (CFAD) of cloud reflectivity from Cloudsat. It demonstrated that cloud tended to be restrained under heavy aerosol conditions.