Search

Soil salinization is a significant global ecological issue that leads to soil degradation and is recognized as one of the primary factors hindering the sustainable development of irrigated farmlands and deserts. The integration of remote sensing (RS) and machine learning algorithms is increasingly employed to deliver cost-effective, time-efficient, spatially resolved, accurately mapped, and uncertainty-quantified soil salinity information. We reviewed articles published between January 2016 and December 2023 on remote sensing-based soil salinity prediction and synthesized the latest research advancements in terms of innovation points, data, methodologies, variable importance, global soil salinity trends, current challenges, and potential future research directions. Our observations indicate that the innovations in this field focus on detection depth, iterations of data conversion methods, and the application of newly developed sensors. Statistical analysis reveals that Landsat is the most frequently utilized sensor in these studies. Furthermore, the application of deep learning algorithms remains underexplored. The ranking of soil salinity prediction accuracy across the various study areas is as follows: lake wetland (R2 = 0.81) > oasis (R2 = 0.76) > coastal zone (R2 = 0.74) > farmland (R2 = 0.71). We also examined the relationship between metadata and prediction accuracy: (1) Validation accuracy, sample size, number of variables, and mean sample salinity exhibited some correlation with modeling accuracy, while sampling depth, variable type, sampling time, and maximum salinity did not influence modeling accuracy. (2) Across a broad range of scales, large sample sizes may lead to error accumulation, which is associated with the geographic diversity of the study area. (3) The inclusion of additional environmental variables does not necessarily enhance modeling accuracy. (4) Modeling accuracy improves when the mean salinity of the study area exceeds 30 dS/m. Topography, vegetation, and temperature are relatively significant environmental covariates. Over the past 30 years, the global area affected by soil salinity has been increasing. To further enhance prediction accuracy, we provide several suggestions for the challenges and directions for future research. While remote sensing is not the sole solution, it provides unique advantages for soil salinity-related studies at both regional and global scales. [ABSTRACT FROM AUTHOR]

Significant Wave Height (SWH) is a crucial parameter in oceanographic research, essential for understanding various marine and atmospheric processes. Traditional methods for obtaining SWH, such as ship-based and buoy measurements, face limitations like limited spatial coverage and high operational costs. With the advancement of Global Navigation Satellite Systems reflectometry (GNSS-R) technology, a new method for retrieving SWH has emerged, demonstrating promising results. This study utilizes Radio occultation sounder (GNOS) data from the FY-3E satellite and incorporates the latest Vision Transformer (ViT) technology to investigate GNSS-R-based SWH retrieval. We designed and evaluated various deep learning models, including ANN-Wave, CNN-Wave, Hybrid-Wave, Trans-Wave, and ViT-Wave. Through comparative training using ERA5 data, the ViT-Wave model was identified as the optimal retrieval model. The ViT-Wave model achieved a Root Mean Square Error (RMSE) accuracy of 0.4052 m and Mean Absolute Error (MAE) accuracy of 0.2700 m, significantly outperforming both traditional methods and newer deep learning approaches utilizing Cyclone Global Navigation Satellite Systems (CYGNSS) data. These results underscore the potential of integrating GNSS-R technology with advanced deep-learning models to enhance SWH retrieval accuracy and reliability in oceanographic research. [ABSTRACT FROM AUTHOR]

Landsat data correction using the Land Surface Reflectance Code (LaSRC) has been proposed as the basis for the atmospheric correction of smallsats. While atmospheric correction can enhance smallsat data, the Landsat/LaSRC pathway delays output and may constrain accuracy and utility. The alternative, the Closed-form Method for Atmospheric Correction (CMAC), developed for smallsat application, provides surface reflectance derived solely from scene statistics. In a prior paper, CMAC closely agreed with LaSRC software for correction of the four VNIR bands of Landsat-8/9 images for conditions of low to moderate atmospheric effect over quasi-invariant warehouse-industrial targets. Those results were accepted as surrogate surface reflectance to support analysis of CMAC and LaSRC reliability for surface reflectance retrieval in two contrasting environments: shortgrass prairie and barren desert. Reliability was defined and tested through a null hypothesis: the same top-of-atmosphere reflectance under the same atmospheric condition will provide the same estimate of surface reflectance. Evaluated against the prior surrogate surface reflectance, the results found decreasing error with increasing wavelength for both methods. From 58 comparisons across the four bands, the LaSRC average absolute error ranged from 0.59% (NIR) to 50.30% (blue). CMAC provided reliable results: error was well constrained from 0.01% (NIR) to 0.98% (blue). [ABSTRACT FROM AUTHOR]

The traditional method for retrieving cloud optical thickness (COT) is carried out through a Look-Up Table (LUT). Researchers must make a series of idealized assumptions and conduct extensive observations and record features in this scenario, consuming considerable resources. The emergence of deep learning effectively addresses the shortcomings of the traditional approach. In this paper, we first propose a daytime (SOZA < 70°) COT retrieval algorithm based on FY-4A AGRI. We establish and train a Convolutional Neural Network (CNN) model for COT retrieval, CM4CR, with the CALIPSO's COT product spatially and temporally synchronized as the ground truth. Then, a deep learning method extended from video prediction models is adopted to predict COT values based on the retrieval results obtained from CM4CR. The COT prediction model (CPM) consists of an encoder, a predictor, and a decoder. On this basis, we further incorporated a time embedding module to enhance the model's ability to learn from irregular time intervals in the input COT sequence. During the training phase, we employed Charbonnier Loss and Edge Loss to enhance the model's capability to represent COT details. Experiments indicate that our CM4CR outperforms existing COT retrieval methods, with predictions showing better performance across several metrics than other benchmark prediction models. Additionally, this paper also investigates the impact of different lengths of COT input sequences and the time intervals between adjacent frames of COT on prediction performance. [ABSTRACT FROM AUTHOR]

In this study, the global land surface albedo namely GAC43 was retrieved for the years 1979 to 2020 using Advanced Very High Resolution Radiometer (AVHRR) global area coverage (GAC) data onboard National Oceanic and Atmospheric Administration (NOAA) and Meteorological Operational (MetOp) satellites. We provide a comprehensive retrieval process of the GAC43 albedo, followed by a comprehensive assessment against in situ measurements and three widely used satellite-based albedo products, the third edition of the CM SAF cLoud, Albedo and surface RAdiation (CLARA-A3), the Copernicus Climate Change Service (C3S) albedo product, and MODIS BRDF/albedo product (MCD43). Our quantitative evaluations indicate that GAC43 demonstrates the best stability, with a linear trend of ±0.002 per decade at nearly all pseudo invariant calibration sites (PICS) from 1982 to 2020. In contrast, CLARA-A3 exhibits significant noise before the 2000s due to the limited availability of observations, while C3S shows substantial biases during the same period due to imperfect sensors intercalibrations. Extensive validation at globally distributed homogeneous sites shows that GAC43 has comparable accuracy to C3S, with an overall RMSE of approximately 0.03, but a smaller positive bias of 0.012. Comparatively, MCD43C3 shows the lowest RMSE (~0.023) and minimal bias, while CLARA-A3 displays the highest RMSE (~0.042) and bias (0.02). Furthermore, GAC43, CLARA-A3, and C3S exhibit overestimation in forests, with positive biases exceeding 0.023 and RMSEs of at least 0.028. In contrast, MCD43C3 shows negligible bias and a smaller RMSE of 0.015. For grasslands and shrublands, GAC43 and MCD43C3 demonstrate comparable estimation uncertainties of approximately 0.023, with close positive biases near 0.09, whereas C3S and CLARA-A3 exhibit higher RMSEs and biases exceeding 0.032 and 0.022, respectively. All four albedo products show significant RMSEs around 0.035 over croplands but achieve the highest estimation accuracy better than 0.020 over deserts. It is worth noting that significant biases are typically attributed to insufficient spatial representativeness of the measurement sites. Globally, GAC43 and C3S exhibit similar spatial distribution patterns across most land surface conditions, including an overestimation compared to MCD43C3 and an underestimation compared to CLARA-A3 in forested areas. In addition, GAC43, C3S, and CLARA-A3 estimate higher albedo values than MCD43C3 in low-vegetation regions, such as croplands, grasslands, savannas, and woody savannas. Besides the fact that the new GAC43 product shows the best stability covering the last 40 years, one has to consider the higher proportion of backup inversions before 2000. Overall, GAC43 offers a promising long-term and consistent albedo with good accuracy for future studies such as global climate change, energy balance, and land management policy.

Soil salinization is a significant global ecological issue that leads to soil degradation and is recognized as one of the primary factors hindering the sustainable development of irrigated farmlands and deserts. The integration of remote sensing (RS) and machine learning algorithms is increasingly employed to deliver cost-effective, time-efficient, spatially resolved, accurately mapped, and uncertainty-quantified soil salinity information. We reviewed articles published between January 2016 and December 2023 on remote sensing-based soil salinity prediction and synthesized the latest research advancements in terms of innovation points, data, methodologies, variable importance, global soil salinity trends, current challenges, and potential future research directions. Our observations indicate that the innovations in this field focus on detection depth, iterations of data conversion methods, and the application of newly developed sensors. Statistical analysis reveals that Landsat is the most frequently utilized sensor in these studies. Furthermore, the application of deep learning algorithms remains underexplored. The ranking of soil salinity prediction accuracy across the various study areas is as follows: lake wetland (R2 = 0.81) > oasis (R2 = 0.76) > coastal zone (R2 = 0.74) > farmland (R2 = 0.71). We also examined the relationship between metadata and prediction accuracy: (1) Validation accuracy, sample size, number of variables, and mean sample salinity exhibited some correlation with modeling accuracy, while sampling depth, variable type, sampling time, and maximum salinity did not influence modeling accuracy. (2) Across a broad range of scales, large sample sizes may lead to error accumulation, which is associated with the geographic diversity of the study area. (3) The inclusion of additional environmental variables does not necessarily enhance modeling accuracy. (4) Modeling accuracy improves when the mean salinity of the study area exceeds 30 dS/m. Topography, vegetation, and temperature are relatively significant environmental covariates. Over the past 30 years, the global area affected by soil salinity has been increasing. To further enhance prediction accuracy, we provide several suggestions for the challenges and directions for future research. While remote sensing is not the sole solution, it provides unique advantages for soil salinity-related studies at both regional and global scales.

Microwave hyperspectral instruments represent one of the main atmospheric sounders of China's next-generation Fengyun meteorological satellites. In order to better apply microwave hyperspectral observations in the fields of atmospheric parameter retrieval and data assimilation, this paper analyzes the sensitivity of trace gases to five selected bandwidth channels using a radiative transfer model based on the simulated data of microwave hyperspectral radiances at 50–60 GHz. This method uses information entropy and a weighting function to select channels and analyze the impact of this on the retrieval accuracy of atmospheric profiles before and after channel selection. The experimental results show that channel selection can reduce the number of channels by approximately 74.05% while maintaining a large amount of information content, and this retrieval effect is significantly better than that of MWTS-III. After channel selection, the 10 MHz, 30 MHz, and 50 MHz bandwidths have the best retrieval results in the stratosphere, whole atmosphere, and troposphere, respectively. When considering the number of channels, computational scale, and retrieval results comprehensively, the channel selection method is effective. [ABSTRACT FROM AUTHOR]

Owing to advancements in satellite remote sensing technology, the acquisition of global land surface parameters, notably, the leaf area index (LAI), has become increasingly accessible. The Sentinel-2 (S2) satellite plays an important role in the monitoring of ecological environments and resource management. The prevalent use of the 20 m spatial resolution band in S2-based inversion models imposes significant limitations on the applicability of S2 data in applications requiring finer spatial resolution. Furthermore, although a substantial body of research on LAI retrieval using S2 data concentrates on agricultural landscapes, studies dedicated to forest ecosystems, although increasing, remain relatively less prevalent. This study aims to establish a viable methodology for retrieving 10 m resolution LAI data in forested regions. The empirical model of the soil adjusted vegetation index (SAVI), the backpack neural network based on simulated annealing (SA-BP) algorithm, and the variational heteroscedastic Gaussian process regression (VHGPR) model are established in this experiment based on the LAI data measured and the corresponding 10 m spatial resolution S2 satellite surface reflectance data in the Saihanba Forestry Center (SFC). The LAI retrieval performance of the three models is then validated using field data, and the error sources of the best performing VHGPR models (R2 of 0.8696 and RMSE of 0.5078) are further analyzed. Moreover, the VHGPR model stands out for its capacity to quantify the uncertainty in LAI estimation, presenting a notable advantage in assessing the significance of input data, eliminating redundant bands, and being well suited for uncertainty estimation. This feature is particularly valuable in generating accurate LAI products, especially in regions characterized by diverse forest compositions. [ABSTRACT FROM AUTHOR]

Satellite-based hyper-spectral infrared (IR) sensors such as the Atmospheric Infrared Sounder (AIRS), the Cross-track Infrared Sounder (CrIS), and the Infrared Atmospheric Sounding Interferometer (IASI) cover many methane (CH4) spectral features, including the ν1 vibrational band near 1300 cm−1 (7.7 μm); therefore, they can be used to monitor CH4 concentrations in the atmosphere. However, retrieving CH4 remains a challenge due to the limited spectral information provided by IR sounder measurements. The information required to resolve the weak absorption lines of CH4 is often obscured by interferences from signals originating from other trace gases, clouds, and surface emissions within the overlapping spectral region. Consequently, currently available CH4 data product derived from IR sounder measurements still have large errors and uncertainties that limit their application scope for high-accuracy climate and environment monitoring applications. In this paper, we describe the retrieval of atmospheric CH4 profiles using a novel spectral fingerprinting methodology and our evaluation of performance using measurements from the CrIS sensor aboard the Suomi National Polar-orbiting Partnership (SNPP) satellite. The spectral fingerprinting methodology uses optimized CrIS radiances to enhance the CH4 signal and a machine learning classifier to constrain the physical inversion scheme. We validated our results using the atmospheric composition reanalysis results and data from airborne in situ measurements. An inter-comparison study revealed that the spectral fingerprinting results can capture the vertical variation characteristics of CH4 profiles that operational sounder products may not provide. The latitudinal variations in CH4 concentration in these results appear more realistic than those shown in existing sounder products. The methodology presented herein could enhance the utilization of satellite data to comprehend methane's role as a greenhouse gas and facilitate the tracking of methane sources and sinks with increased reliability. [ABSTRACT FROM AUTHOR]

Significant Wave Height (SWH) is a crucial parameter in oceanographic research, essential for understanding various marine and atmospheric processes. Traditional methods for obtaining SWH, such as ship-based and buoy measurements, face limitations like limited spatial coverage and high operational costs. With the advancement of Global Navigation Satellite Systems reflectometry (GNSS-R) technology, a new method for retrieving SWH has emerged, demonstrating promising results. This study utilizes Radio occultation sounder (GNOS) data from the FY-3E satellite and incorporates the latest Vision Transformer (ViT) technology to investigate GNSS-R-based SWH retrieval. We designed and evaluated various deep learning models, including ANN-Wave, CNN-Wave, Hybrid-Wave, Trans-Wave, and ViT-Wave. Through comparative training using ERA5 data, the ViT-Wave model was identified as the optimal retrieval model. The ViT-Wave model achieved a Root Mean Square Error (RMSE) accuracy of 0.4052 m and Mean Absolute Error (MAE) accuracy of 0.2700 m, significantly outperforming both traditional methods and newer deep learning approaches utilizing Cyclone Global Navigation Satellite Systems (CYGNSS) data. These results underscore the potential of integrating GNSS-R technology with advanced deep-learning models to enhance SWH retrieval accuracy and reliability in oceanographic research.

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. [ABSTRACT FROM AUTHOR]

The traditional method for retrieving cloud optical thickness (COT) is carried out through a Look-Up Table (LUT). Researchers must make a series of idealized assumptions and conduct extensive observations and record features in this scenario, consuming considerable resources. The emergence of deep learning effectively addresses the shortcomings of the traditional approach. In this paper, we first propose a daytime (SOZA < 70°) COT retrieval algorithm based on FY-4A AGRI. We establish and train a Convolutional Neural Network (CNN) model for COT retrieval, CM4CR, with the CALIPSO’s COT product spatially and temporally synchronized as the ground truth. Then, a deep learning method extended from video prediction models is adopted to predict COT values based on the retrieval results obtained from CM4CR. The COT prediction model (CPM) consists of an encoder, a predictor, and a decoder. On this basis, we further incorporated a time embedding module to enhance the model’s ability to learn from irregular time intervals in the input COT sequence. During the training phase, we employed Charbonnier Loss and Edge Loss to enhance the model’s capability to represent COT details. Experiments indicate that our CM4CR outperforms existing COT retrieval methods, with predictions showing better performance across several metrics than other benchmark prediction models. Additionally, this paper also investigates the impact of different lengths of COT input sequences and the time intervals between adjacent frames of COT on prediction performance.

Landsat data correction using the Land Surface Reflectance Code (LaSRC) has been proposed as the basis for the atmospheric correction of smallsats. While atmospheric correction can enhance smallsat data, the Landsat/LaSRC pathway delays output and may constrain accuracy and utility. The alternative, the Closed-form Method for Atmospheric Correction (CMAC), developed for smallsat application, provides surface reflectance derived solely from scene statistics. In a prior paper, CMAC closely agreed with LaSRC software for correction of the four VNIR bands of Landsat-8/9 images for conditions of low to moderate atmospheric effect over quasi-invariant warehouse-industrial targets. Those results were accepted as surrogate surface reflectance to support analysis of CMAC and LaSRC reliability for surface reflectance retrieval in two contrasting environments: shortgrass prairie and barren desert. Reliability was defined and tested through a null hypothesis: the same top-of-atmosphere reflectance under the same atmospheric condition will provide the same estimate of surface reflectance. Evaluated against the prior surrogate surface reflectance, the results found decreasing error with increasing wavelength for both methods. From 58 comparisons across the four bands, the LaSRC average absolute error ranged from 0.59% (NIR) to 50.30% (blue). CMAC provided reliable results: error was well constrained from 0.01% (NIR) to 0.98% (blue).

The atmospheric temperature profiles (ATPs) retrieved through the geostationary Interferometric Infrared Sounder (GIIRS) onboard the FY-4A satellite (GIIRS/FY-4A) can effectively fill the gap of the scarce conventional sounding data in the Taklimakan Desert (TD), the second largest desert in the world, with an area of 330,000 square kilometers. In this study, we take the experimental radiosonde observations (RAOB) from one RAOB station in the hinterland of TD and seven conventional radiosondes in the oasis region around the desert as the true values and analyze the bias distribution characteristics of GIIRS/FY-4A ATPs with quality control (QC) flags 0 or 1 for this region. In addition, a bias comparison is made with GIIRS/FY-4A ATPs, and the fifth generation ECMWF atmospheric reanalysis of the global climate (ERA5) ATPs. The results show that (1) Missing measurements in GIIRS/FY-4A ATPs are the most frequent in the near-surface layer, accounting for more than 80% of all the retrieved grid points. The averaged total proportion of GIIRS/FY-4A ATPs with QC marks 0 or 1 is about 33.06%. (2) The root mean square error (RMSE) of GIIRS/FY-4A ATPs is less than 3 K, smaller than that of ERA5 ATPs. The RMSE of ERA5 ATPs can exceed 10 K in the desert hinterland. The absolute mean biases of GIIRS/FY-4A ATPs and ERA5 ATPs are, respectively, smaller than 3 K and 2 K, the former being slightly larger. The correlation coefficients of GIIRS/FY-4A ATPs with ERA5 ATPs and RAOB ATPs are higher than 0.98 and 0.99, respectively, and the correlation between GIIRS/FY-4A ATPs and RAOB ATPs is inferior to the latter. (3) The overall atmospheric temperature retrieved by GIIRS/FY-4A is 0.08 K higher than the temperature of RAOB, on average, while the overall temperature from ERA5 is 0.13 K lower than that of RAOB, indicating that the temperature profile obtained by integrating GIIRS/FY-4A ATPs and ERA5 ATPs may be much closer to RAOB ATPs. (4) The probability density of the GIIRS/FY-4A ATP biases in the TD region generally follows the Gaussian distribution so that it can be effectively assimilated in the 3-D variational assimilation modules. The probability density distribution characteristics of the GIIRS/FY-4A ATP biases in the desert hinterland and oasis are not much different. However, due to the fusion analysis of the relatively rich multi-source conventional observation data from the oasis stations, the probability density of ERA5 ATPs biases at the oasis stations is nearer to Gaussian distribution than that of the GIIRS/FY-4A ATPs. In the desert hinterland, where conventional observation is not enough, the probability density distributions of the ATPs biases from ERA5 and GIIRS/FY-4A are alike. Therefore, the GIIRS FY4A can contribute to a more accurate estimation of ERA5 ATPs in the TD region. [ABSTRACT FROM AUTHOR]

The atmospheric temperature profile is an important parameter to describe the state of the atmosphere, and it is crucial to climate change research, weather forecasting, and atmospheric parameter retrieval. A machine learning algorithm that incorporates historical observations and ground-based measurements was developed in this study to retrieve the atmospheric temperature profile. Specifically, the deep learning network considered historical observations for the same period and temporally correlated temperature profiles. It combined multi-layer perceptron (MLP) and the convolutional neural network (CNN). MLP derived the features from the ground factors, and CNN captured the essential features associated with the temperature profiles at the current time from latent historical data. Then, the features of the two parts were concatenated to obtain the final network. The construction and parameters of the model were optimized to determine the best model configuration and retrieval performance. The results of the model were evaluated against those of other methods on the same dataset. The model showed a good retrieval precision, which was equivalent to a small retrieval bias, root-mean-square error, and mean absolute error at all altitudes. The analysis of the application of this model to the retrieval of atmospheric temperature profiles indicates that the method is feasible. [ABSTRACT FROM AUTHOR]

Microwave hyperspectral instruments represent one of the main atmospheric sounders of China’s next-generation Fengyun meteorological satellites. In order to better apply microwave hyperspectral observations in the fields of atmospheric parameter retrieval and data assimilation, this paper analyzes the sensitivity of trace gases to five selected bandwidth channels using a radiative transfer model based on the simulated data of microwave hyperspectral radiances at 50–60 GHz. This method uses information entropy and a weighting function to select channels and analyze the impact of this on the retrieval accuracy of atmospheric profiles before and after channel selection. The experimental results show that channel selection can reduce the number of channels by approximately 74.05% while maintaining a large amount of information content, and this retrieval effect is significantly better than that of MWTS-III. After channel selection, the 10 MHz, 30 MHz, and 50 MHz bandwidths have the best retrieval results in the stratosphere, whole atmosphere, and troposphere, respectively. When considering the number of channels, computational scale, and retrieval results comprehensively, the channel selection method is effective.

Satellite-based hyper-spectral infrared (IR) sensors such as the Atmospheric Infrared Sounder (AIRS), the Cross-track Infrared Sounder (CrIS), and the Infrared Atmospheric Sounding Interferometer (IASI) cover many methane (CH4) spectral features, including the ν1 vibrational band near 1300 cm−1 (7.7 μm); therefore, they can be used to monitor CH4 concentrations in the atmosphere. However, retrieving CH4 remains a challenge due to the limited spectral information provided by IR sounder measurements. The information required to resolve the weak absorption lines of CH4 is often obscured by interferences from signals originating from other trace gases, clouds, and surface emissions within the overlapping spectral region. Consequently, currently available CH4 data product derived from IR sounder measurements still have large errors and uncertainties that limit their application scope for high-accuracy climate and environment monitoring applications. In this paper, we describe the retrieval of atmospheric CH4 profiles using a novel spectral fingerprinting methodology and our evaluation of performance using measurements from the CrIS sensor aboard the Suomi National Polar-orbiting Partnership (SNPP) satellite. The spectral fingerprinting methodology uses optimized CrIS radiances to enhance the CH4 signal and a machine learning classifier to constrain the physical inversion scheme. We validated our results using the atmospheric composition reanalysis results and data from airborne in situ measurements. An inter-comparison study revealed that the spectral fingerprinting results can capture the vertical variation characteristics of CH4 profiles that operational sounder products may not provide. The latitudinal variations in CH4 concentration in these results appear more realistic than those shown in existing sounder products. The methodology presented herein could enhance the utilization of satellite data to comprehend methane’s role as a greenhouse gas and facilitate the tracking of methane sources and sinks with increased reliability.

Owing to advancements in satellite remote sensing technology, the acquisition of global land surface parameters, notably, the leaf area index (LAI), has become increasingly accessible. The Sentinel-2 (S2) satellite plays an important role in the monitoring of ecological environments and resource management. The prevalent use of the 20 m spatial resolution band in S2-based inversion models imposes significant limitations on the applicability of S2 data in applications requiring finer spatial resolution. Furthermore, although a substantial body of research on LAI retrieval using S2 data concentrates on agricultural landscapes, studies dedicated to forest ecosystems, although increasing, remain relatively less prevalent. This study aims to establish a viable methodology for retrieving 10 m resolution LAI data in forested regions. The empirical model of the soil adjusted vegetation index (SAVI), the backpack neural network based on simulated annealing (SA-BP) algorithm, and the variational heteroscedastic Gaussian process regression (VHGPR) model are established in this experiment based on the LAI data measured and the corresponding 10 m spatial resolution S2 satellite surface reflectance data in the Saihanba Forestry Center (SFC). The LAI retrieval performance of the three models is then validated using field data, and the error sources of the best performing VHGPR models (R2 of 0.8696 and RMSE of 0.5078) are further analyzed. Moreover, the VHGPR model stands out for its capacity to quantify the uncertainty in LAI estimation, presenting a notable advantage in assessing the significance of input data, eliminating redundant bands, and being well suited for uncertainty estimation. This feature is particularly valuable in generating accurate LAI products, especially in regions characterized by diverse forest compositions.

Spaceborne snow water retrievals over oceans are assessed using a multiyear coincident dataset of CloudSat Cloud Profiling Radar (CPR) and Global Precipitation Mission (GPM) Dual-frequency Precipitation Radar (DPR). Various factors contributing to differences in snow water retrievals between CPR and DPR are carefully considered. A set of relationships between radar reflectivity (Ze) and snow water content (SWC) at Ku- and W-bands is developed using the same microphysical assumptions. It is found that surface snow water contents from CPR are much larger than those from DPR at latitudes above 60°, while surface snow water contents from DPR slightly exceed those from CPR at latitudes below 50°. Coincident snow water content profiles between CPR and DPR are further divided into two conditions. One is that only CPR detects the falling snow. Another is that both CPR and DPR detect the falling snow. The results indicate that about 88% of all snow water content profiles are under the first condition and usually associated with light snowfall events. The remaining snow water content profiles are generally associated with moderate and heavy snowfall events. Moreover, CPR surface snow water contents are larger than DPR ones at high latitudes because most light snowfall events are misdetected by DPR due to its low sensitivity. DPR surface snow water contents exceed CPR ones at low latitudes because CPR may experience a significant reduction in backscattering efficiency of large particles and attenuation in heavy snowfall events. The low sensitivity of DPR also causes a noticeable decrease in detected snow layer depth. The results presented here can help in developing global snowfall retrieval algorithms using multi-radars. [ABSTRACT FROM AUTHOR]

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.

Different cloud-top heights (CTHs) have different degrees of atmospheric heating, which is an important factor for weather forecasting and aviation safety. AHIs (Advanced Himawari Imagers) on the Himawari-8 satellite are a new generation of visible and infrared imaging spectrometers characterized by a wide observation range and a high temporal resolution. In this paper, a cloud-top height retrieval algorithm based on XGBoost is proposed. The algorithm comprehensively utilizes AHI L1 multi-channel radiance data and calculates the input parameters of the generated model according to the characteristics of the cloud phase, texture, and the local brightness temperature change of the cloud. In addition, the latitude, longitude, solar zenith angle and satellite zenith angle are input into the model to further constrain the influence of the geographical and spatial factors such as the sea and land location, on CTH. Compared with the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) cloud-top height data (CTHCAL), the results show that: the algorithm retrieved the cloud-top height (CTHXGB) with a mean error (ME) of 0.3 km, a standard deviation (Std) of 1.72 km, and a root mean square error (RMSE) of 1.74 km. Additionally, it improves the problem of the large systematic deviation in the cloud-top height products released by the Japan Meteorological Agency (CTHJMA), especially for ice clouds and multi-layer clouds with ice clouds on the top layer. For water clouds below 2 km and multi-layer clouds with water clouds at the top, the algorithm solves the systematically serious CTHJMA problem. XGBoost can effectively distinguish between different cloud scenarios within the model, which is robust and suitable for CTH retrieval. [ABSTRACT FROM AUTHOR]

Satellite remote sensing of aerosol optical depth (AOD) is essential for detection, characterization, and forecasting of wildfire smoke. In this work, we evaluate the AOD (550 nm) retrievals during the extreme wildfire events over the western U.S. in September 2020. Three products are analyzed, including the Moderate-resolution Imaging Spectroradiometers (MODIS) Multi-Angle Implementation of Atmospheric Correction (MAIAC) product collections C6.0 and C6.1, and the NOAA-20 Visible Infrared Imaging Radiometer (VIIRS) AOD from the NOAA Enterprise Processing System (EPS) algorithm. Compared with the Aerosol Robotic Network (AERONET) data, all three products show strong linear correlations with MAIAC C6.1 and VIIRS presenting overall low bias (<0.06). The accuracy of MAIAC C6.1 is found to be substantially improved with respect to MAIAC C6.0 that drastically underestimated AOD over thick smoke, which validates the effectiveness of updates made in MAIAC C6.1 in terms of an improved representation of smoke aerosol optical properties. VIIRS AOD exhibits comparable uncertainty with MAIAC C6.1 with a slight tendency of increased positive bias over the AERONET AOD range of 0.5–3.0. Averaging coincident retrievals from MAIAC C6.1 and VIIRS provides a lower root mean square error and higher correlation than for the individual products, motivating the benefit of blending these datasets. MAIAC C6.1 and VIIRS are further compared to provide insights on their retrieval strategy. When gridded at 0.1° resolution, MAIAC C6.1 and VIIRS provide similar monthly AOD distribution patterns and the latter exhibits a slightly higher domain average. On daily scale, over thick plumes near fire sources, MAIAC C6.1 reports more valid retrievals where VIIRS tends to have retrievals designated as low or medium quality, which tends to be due to internal quality checks. Over transported smoke near scattered clouds, VIIRS provides better retrieval coverage than MAIAC C6.1 owing to its higher spatial resolution, pixel-level processing, and less strict cloud masking. These results can be used as a guide for applications of satellite AOD retrievals during wildfire events and provide insights on future improvement of retrieval algorithms under heavy smoke conditions. [ABSTRACT FROM AUTHOR]

The atmospheric temperature profiles (ATPs) retrieved through the geostationary Interferometric Infrared Sounder (GIIRS) onboard the FY-4A satellite (GIIRS/FY-4A) can effectively fill the gap of the scarce conventional sounding data in the Taklimakan Desert (TD), the second largest desert in the world, with an area of 330,000 square kilometers. In this study, we take the experimental radiosonde observations (RAOB) from one RAOB station in the hinterland of TD and seven conventional radiosondes in the oasis region around the desert as the true values and analyze the bias distribution characteristics of GIIRS/FY-4A ATPs with quality control (QC) flags 0 or 1 for this region. In addition, a bias comparison is made with GIIRS/FY-4A ATPs, and the fifth generation ECMWF atmospheric reanalysis of the global climate (ERA5) ATPs. The results show that (1) Missing measurements in GIIRS/FY-4A ATPs are the most frequent in the near-surface layer, accounting for more than 80% of all the retrieved grid points. The averaged total proportion of GIIRS/FY-4A ATPs with QC marks 0 or 1 is about 33.06%. (2) The root mean square error (RMSE) of GIIRS/FY-4A ATPs is less than 3 K, smaller than that of ERA5 ATPs. The RMSE of ERA5 ATPs can exceed 10 K in the desert hinterland. The absolute mean biases of GIIRS/FY-4A ATPs and ERA5 ATPs are, respectively, smaller than 3 K and 2 K, the former being slightly larger. The correlation coefficients of GIIRS/FY-4A ATPs with ERA5 ATPs and RAOB ATPs are higher than 0.98 and 0.99, respectively, and the correlation between GIIRS/FY-4A ATPs and RAOB ATPs is inferior to the latter. (3) The overall atmospheric temperature retrieved by GIIRS/FY-4A is 0.08 K higher than the temperature of RAOB, on average, while the overall temperature from ERA5 is 0.13 K lower than that of RAOB, indicating that the temperature profile obtained by integrating GIIRS/FY-4A ATPs and ERA5 ATPs may be much closer to RAOB ATPs. (4) The probability density of the GIIRS/FY-4A ATP biases in the TD region generally follows the Gaussian distribution so that it can be effectively assimilated in the 3-D variational assimilation modules. The probability density distribution characteristics of the GIIRS/FY-4A ATP biases in the desert hinterland and oasis are not much different. However, due to the fusion analysis of the relatively rich multi-source conventional observation data from the oasis stations, the probability density of ERA5 ATPs biases at the oasis stations is nearer to Gaussian distribution than that of the GIIRS/FY-4A ATPs. In the desert hinterland, where conventional observation is not enough, the probability density distributions of the ATPs biases from ERA5 and GIIRS/FY-4A are alike. Therefore, the GIIRS FY4A can contribute to a more accurate estimation of ERA5 ATPs in the TD region.

The atmospheric temperature profile is an important parameter to describe the state of the atmosphere, and it is crucial to climate change research, weather forecasting, and atmospheric parameter retrieval. A machine learning algorithm that incorporates historical observations and ground-based measurements was developed in this study to retrieve the atmospheric temperature profile. Specifically, the deep learning network considered historical observations for the same period and temporally correlated temperature profiles. It combined multi-layer perceptron (MLP) and the convolutional neural network (CNN). MLP derived the features from the ground factors, and CNN captured the essential features associated with the temperature profiles at the current time from latent historical data. Then, the features of the two parts were concatenated to obtain the final network. The construction and parameters of the model were optimized to determine the best model configuration and retrieval performance. The results of the model were evaluated against those of other methods on the same dataset. The model showed a good retrieval precision, which was equivalent to a small retrieval bias, root-mean-square error, and mean absolute error at all altitudes. The analysis of the application of this model to the retrieval of atmospheric temperature profiles indicates that the method is feasible.

The Chinese carbon dioxide measurement satellite (TanSat) has collected a large number of measurements in the solar calibration mode. To improve the accuracy of XCO2 retrieval, the Instrument Line Shape (ILS, also known as the slit function) must be accurately determined. In this study, we characterized the on-orbit ILS of TanSat by fitting measured solar irradiance from 2017 to 2018 with a well-calibrated high-spectral-resolution solar reference spectrum. We used various advanced analytical functions and the stretch/sharpen of the tabulated preflight ILS to represent the ILS for each wavelength window, footprint, and band. Using super Gaussian+P7 and the stretch/sharpen functions substantially reduced the fitting residual in O 2 A-band and weak CO 2 band compared with using the preflight ILS. We found that the difference between the derived ILS width and on-ground preflight ILS was up to −3.5% in the weak CO 2 band, depending on footprint and wavelength. The large amplitude of the ILS wings, depending on the wavelength, footprint, and bands, indicated possible uncorrected stray light. Broadening ILS wings will cause additive offset (filling-in) on the deep absorption lines of the spectra, which we confirmed using offline bias correction of the solar-induced fluorescence retrieval. We estimated errors due to the imperfect ILS using simulated TanSat spectra. The results of the simulations showed that XCO2 retrieval is sensitive to errors in the ILS, and 4% uncertainty in the full width of half maximum (FWHM) or 20% uncertainty in the ILS wings can induce an error of up to 1 ppm in the XCO2 retrieval. [ABSTRACT FROM AUTHOR]

The space-borne microwave radiometers and scatterometers can effectively measure global sea surface winds under non-precipitation. However, the measurements in rainy conditions significantly degrade, which are usually flagged as poor quality or invalidated for some scientific purposes. This paper develops a combined active–passive wind vector retrieval model for rainy conditions based on the HY-2B radiometer and scatterometer measurements. In our model, the polarization ratio of brightness temperatures at 6.925 GHz (PR06) is used as an indicator to implicitly represent the rain effect. For wind speed retrieval, a statistical regression model is trained as a function of PR06 and brightness temperatures of the radiometer. Moreover, two new geophysical model functions, including rain effect, are developed for wind direction inversion. Comparisons between HY-2B retrieval results and ERA5 wind products indicate that the retrieval model performs well under all rainy conditions. The overall root mean squared errors (RMSEs) of wind speed and direction retrievals are 1.60 m/s and 20.60°, respectively. With an increase in the rain rate, the wind retrieval performance degrades slightly and still provides a reliable retrieval result. [ABSTRACT FROM AUTHOR]

This paper describes the initial results of precipitation estimates from the Time-Resolved Observations of Precipitation structure and storm Intensity with a Constellation of Smallsats (TROPICS) Millimeter-wave Sounder (TMS) using the Precipitation Retrieval and Profiling Scheme (PRPS). The TROPICS mission consists of a Pathfinder CubeSat and a constellation of six CubeSats, providing a low-cost solution to the frequent sampling of precipitation systems across the Tropics. The TMS instrument is a 12-channel cross-track scanning radiometer operating at frequencies of 91.655 to 204.8 GHz, providing similar resolutions to current passive microwave sounding instruments. These retrievals showcase the potential of the TMS instrument for precipitation retrievals. The PRPS has been modified for use with the TMS using a database based upon observations from current sounding sensors. The results shown here represent the initial postlaunch version of the retrieval scheme, as analyzed for the Pathfinder CubeSat launched on 30 June 2021. In terms of monthly precipitation estimates, the results fall within the mission specifications and are similar in performance to retrievals from other sounding instruments. At the instantaneous scale, the results are very promising. [ABSTRACT FROM AUTHOR]

Mapping soil heavy metal concentration using machine learning models based on readily available satellite remote sensing images is highly desirable. Accurate mapping relies on appropriate data, feature extraction, and model selection. To this end, a data processing pipeline for soil copper (Cu) concentration estimation has been designed. First, instead of using single Landsat scenes, the utilization of multiple Landsat scenes of the same location over time is considered. Second, to generate a preferred feature set as input to a regression model, a number of feature extraction methods are motivated and compared. Third, to find a preferred regression model, a variety of approaches are implemented and compared for accuracy. In this research, 11 Landsat-8 images from 2013 to 2017 of Gulin County, Sichuan China, and 138 soil samples with lab-measured Cu concentrations collected from the area in 2015 are used. A variety a metrics under cross-validation are used for comparison. The results indicate that multi-temporal images increase accuracy compared to single Landsat images. The preferred feature extraction varies based on the regression model used; however, the best results are obtained using support vector regression and the original data. The final soil Cu map generated using the recommended data processing pipeline shows a consistent spatial pattern with a ground-truth land cover classification map. These results indicate that machine learning has the ability to perform large-scale soil heavy metal mapping from widely available satellite remote sensing images. [ABSTRACT FROM AUTHOR]

Chlorophyll-a is an important parameter that characterizes the eutrophication of water bodies. The advantage of ZY1-02D hyperspectral satellite subdivision in the visible light and near-infrared bands is that it highlights the unique characteristics of water bodies in the spectral dimension, and it helps to assess the Class II water bodies of inland lakes and reservoirs, making it an important tool for refined remote sensing detection of the environment. In this study, the Baiyangdian Nature Reserve in northern China, which contains a typical inland lake and wetland, was chosen as the study area. Using ZY1-02D hyperspectral synchronization transit images and in situ measured chlorophyll-a concentration as the data source, remote sensing of the chlorophyll-a concentration of inland lakes was conducted. By analyzing the correlation between the spectral reflectance of the ZY1-02D hyperspectral image and the chlorophyll-a concentration and using algorithms such as the single band, band ratio, and three bands to compare and filter characteristic wavelengths, a quantitative hyperspectral model of the chlorophyll-a concentration was established to determine the chlorophyll-a concentration of Baiyangdian Lake. The dynamic monitoring of the water body and the assessment of the nutritional status of the water body were determined. The results revealed that the estimation of the chlorophyll-a concentration of Baiyangdian Lake based on the hyperspectral Fluorescence Line Height (FLH) model was ideal, with an R2 value of 0.78. The FLH model not only comprehensively considers the effects of suspended solids, yellow substances, and backscattering of the water body on the estimation of the chlorophyll-a concentration, but also considers the influence of the elastic scattering efficiency of the chlorophyll. Based on the ZY1-02D hyperspectral data, a spatial distribution map of the chlorophyll-a concentration of Baiyangdian Lake was created to provide new ideas and technical support for monitoring inland water environments. [ABSTRACT FROM AUTHOR]

Several global leaf area index (LAI) products were generated using neural networks, but the training dataset for the neural networks was sensor specific, and the construction of the training dataset was time consuming. In this paper, an unsupervised domain adaptation-based method was proposed to estimate LAI from the Visible Infrared Imaging Radiometer Suite (VIIRS) surface reflectance dataset based on a training dataset constructed from the Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance dataset. A transfer component analysis (TCA) algorithm was first utilized to map the MODIS and VIIRS surface reflectance into the same subspace to reduce the distribution discrepancies between the MODIS and VIIRS surface reflectance. Then, the embedded data obtained from MODIS surface reflectance dataset, along with the LAI values produced by fusing the MODIS and the Carbon cYcle and Change in Land Observational Products from an Ensemble of Satellites (CYCLOPES) products, were employed to train general regression neural networks (GRNNs). Finally, for retrieving the LAI values, the embedded data acquired from the VIIRS surface reflectance dataset was input into the trained GRNNs. For multiple field sites with different biome types, we used this developed method to retrieve LAI values based on the VIIRS surface reflectance dataset. The results indicate that, based on the training dataset built from MODIS surface reflectance dataset, the domain adaptation-based retrieval method can effectively estimate LAI values from VIIRS surface reflectance dataset. By comparison with the VIIRS and MODIS LAI products, the retrieved LAI values with TCA are more consistent with the reference LAI values acquired from high-resolution remote sensing images. The coefficient of determination (R2) and root mean square error (RMSE) of the retrieved LAI values with TCA at all selected sites are 0.88 and 0.68, respectively. Furthermore, the accuracy of the retrieved LAI values with TCA is higher than the retrieved LAI values without TCA with the R2 0.81 and the RMSE 0.79. [ABSTRACT FROM AUTHOR]

Nitrous oxide (N2O) is the third most abundant anthropogenous greenhouse gas (after carbon dioxide and methane), with a long atmospheric lifetime and a continuously increasing concentration due to human activities, making it an important gas to monitor. In this work, we present a new method to retrieve N2O concentration profiles (with up to two degrees of freedom) from each cloud-free satellite observation by the Infrared Atmospheric Sounding Interferometer (IASI), using spectral micro-windows in the N2O ν 3 band, the Radiative Transfer for TOVS (RTTOV) tools and the Tikhonov regularization scheme. A time series of ten years (2011–2020) of IASI N2O profiles and integrated partial columns has been produced and validated with collocated ground-based Network for the Detection of Atmospheric Composition Change (NDACC) and Total Carbon Column Observing Network (TCCON) data. The importance of consistency in the ancillary data used for the retrieval for generating consistent time series has been demonstrated. The Nitrous Oxide Profiling from Infrared Radiances (NOPIR) N2O partial columns are of very good quality, with a positive bias of 1.8 to 4% with respect to the ground-based data, which is less than the sum of uncertainties of the compared values. At high latitudes, the comparisons are a bit worse, due to either a known bias in the ground-based data, or to a higher uncertainty in both ground-based and satellite retrievals. [ABSTRACT FROM AUTHOR]

Spaceborne snow water retrievals over oceans are assessed using a multiyear coincident dataset of CloudSat Cloud Profiling Radar (CPR) and Global Precipitation Mission (GPM) Dual-frequency Precipitation Radar (DPR). Various factors contributing to differences in snow water retrievals between CPR and DPR are carefully considered. A set of relationships between radar reflectivity (Ze) and snow water content (SWC) at Ku- and W-bands is developed using the same microphysical assumptions. It is found that surface snow water contents from CPR are much larger than those from DPR at latitudes above 60°, while surface snow water contents from DPR slightly exceed those from CPR at latitudes below 50°. Coincident snow water content profiles between CPR and DPR are further divided into two conditions. One is that only CPR detects the falling snow. Another is that both CPR and DPR detect the falling snow. The results indicate that about 88% of all snow water content profiles are under the first condition and usually associated with light snowfall events. The remaining snow water content profiles are generally associated with moderate and heavy snowfall events. Moreover, CPR surface snow water contents are larger than DPR ones at high latitudes because most light snowfall events are misdetected by DPR due to its low sensitivity. DPR surface snow water contents exceed CPR ones at low latitudes because CPR may experience a significant reduction in backscattering efficiency of large particles and attenuation in heavy snowfall events. The low sensitivity of DPR also causes a noticeable decrease in detected snow layer depth. The results presented here can help in developing global snowfall retrieval algorithms using multi-radars.

Satellite remote sensing of aerosol optical depth (AOD) is essential for detection, characterization, and forecasting of wildfire smoke. In this work, we evaluate the AOD (550 nm) retrievals during the extreme wildfire events over the western U.S. in September 2020. Three products are analyzed, including the Moderate-resolution Imaging Spectroradiometers (MODIS) Multi-Angle Implementation of Atmospheric Correction (MAIAC) product collections C6.0 and C6.1, and the NOAA-20 Visible Infrared Imaging Radiometer (VIIRS) AOD from the NOAA Enterprise Processing System (EPS) algorithm. Compared with the Aerosol Robotic Network (AERONET) data, all three products show strong linear correlations with MAIAC C6.1 and VIIRS presenting overall low bias (

Different cloud-top heights (CTHs) have different degrees of atmospheric heating, which is an important factor for weather forecasting and aviation safety. AHIs (Advanced Himawari Imagers) on the Himawari-8 satellite are a new generation of visible and infrared imaging spectrometers characterized by a wide observation range and a high temporal resolution. In this paper, a cloud-top height retrieval algorithm based on XGBoost is proposed. The algorithm comprehensively utilizes AHI L1 multi-channel radiance data and calculates the input parameters of the generated model according to the characteristics of the cloud phase, texture, and the local brightness temperature change of the cloud. In addition, the latitude, longitude, solar zenith angle and satellite zenith angle are input into the model to further constrain the influence of the geographical and spatial factors such as the sea and land location, on CTH. Compared with the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) cloud-top height data (CTHCAL), the results show that: the algorithm retrieved the cloud-top height (CTHXGB) with a mean error (ME) of 0.3 km, a standard deviation (Std) of 1.72 km, and a root mean square error (RMSE) of 1.74 km. Additionally, it improves the problem of the large systematic deviation in the cloud-top height products released by the Japan Meteorological Agency (CTHJMA), especially for ice clouds and multi-layer clouds with ice clouds on the top layer. For water clouds below 2 km and multi-layer clouds with water clouds at the top, the algorithm solves the systematically serious CTHJMA problem. XGBoost can effectively distinguish between different cloud scenarios within the model, which is robust and suitable for CTH retrieval.

The Chinese carbon dioxide measurement satellite (TanSat) has collected a large number of measurements in the solar calibration mode. To improve the accuracy of XCO2 retrieval, the Instrument Line Shape (ILS, also known as the slit function) must be accurately determined. In this study, we characterized the on-orbit ILS of TanSat by fitting measured solar irradiance from 2017 to 2018 with a well-calibrated high-spectral-resolution solar reference spectrum. We used various advanced analytical functions and the stretch/sharpen of the tabulated preflight ILS to represent the ILS for each wavelength window, footprint, and band. Using super Gaussian+P7 and the stretch/sharpen functions substantially reduced the fitting residual in O2 A-band and weak CO2 band compared with using the preflight ILS. We found that the difference between the derived ILS width and on-ground preflight ILS was up to −3.5% in the weak CO2 band, depending on footprint and wavelength. The large amplitude of the ILS wings, depending on the wavelength, footprint, and bands, indicated possible uncorrected stray light. Broadening ILS wings will cause additive offset (filling-in) on the deep absorption lines of the spectra, which we confirmed using offline bias correction of the solar-induced fluorescence retrieval. We estimated errors due to the imperfect ILS using simulated TanSat spectra. The results of the simulations showed that XCO2 retrieval is sensitive to errors in the ILS, and 4% uncertainty in the full width of half maximum (FWHM) or 20% uncertainty in the ILS wings can induce an error of up to 1 ppm in the XCO2 retrieval.

This paper describes the initial results of precipitation estimates from the Time-Resolved Observations of Precipitation structure and storm Intensity with a Constellation of Smallsats (TROPICS) Millimeter-wave Sounder (TMS) using the Precipitation Retrieval and Profiling Scheme (PRPS). The TROPICS mission consists of a Pathfinder CubeSat and a constellation of six CubeSats, providing a low-cost solution to the frequent sampling of precipitation systems across the Tropics. The TMS instrument is a 12-channel cross-track scanning radiometer operating at frequencies of 91.655 to 204.8 GHz, providing similar resolutions to current passive microwave sounding instruments. These retrievals showcase the potential of the TMS instrument for precipitation retrievals. The PRPS has been modified for use with the TMS using a database based upon observations from current sounding sensors. The results shown here represent the initial postlaunch version of the retrieval scheme, as analyzed for the Pathfinder CubeSat launched on 30 June 2021. In terms of monthly precipitation estimates, the results fall within the mission specifications and are similar in performance to retrievals from other sounding instruments. At the instantaneous scale, the results are very promising.

The space-borne microwave radiometers and scatterometers can effectively measure global sea surface winds under non-precipitation. However, the measurements in rainy conditions significantly degrade, which are usually flagged as poor quality or invalidated for some scientific purposes. This paper develops a combined active–passive wind vector retrieval model for rainy conditions based on the HY-2B radiometer and scatterometer measurements. In our model, the polarization ratio of brightness temperatures at 6.925 GHz (PR06) is used as an indicator to implicitly represent the rain effect. For wind speed retrieval, a statistical regression model is trained as a function of PR06 and brightness temperatures of the radiometer. Moreover, two new geophysical model functions, including rain effect, are developed for wind direction inversion. Comparisons between HY-2B retrieval results and ERA5 wind products indicate that the retrieval model performs well under all rainy conditions. The overall root mean squared errors (RMSEs) of wind speed and direction retrievals are 1.60 m/s and 20.60°, respectively. With an increase in the rain rate, the wind retrieval performance degrades slightly and still provides a reliable retrieval result.

Mapping soil heavy metal concentration using machine learning models based on readily available satellite remote sensing images is highly desirable. Accurate mapping relies on appropriate data, feature extraction, and model selection. To this end, a data processing pipeline for soil copper (Cu) concentration estimation has been designed. First, instead of using single Landsat scenes, the utilization of multiple Landsat scenes of the same location over time is considered. Second, to generate a preferred feature set as input to a regression model, a number of feature extraction methods are motivated and compared. Third, to find a preferred regression model, a variety of approaches are implemented and compared for accuracy. In this research, 11 Landsat-8 images from 2013 to 2017 of Gulin County, Sichuan China, and 138 soil samples with lab-measured Cu concentrations collected from the area in 2015 are used. A variety a metrics under cross-validation are used for comparison. The results indicate that multi-temporal images increase accuracy compared to single Landsat images. The preferred feature extraction varies based on the regression model used; however, the best results are obtained using support vector regression and the original data. The final soil Cu map generated using the recommended data processing pipeline shows a consistent spatial pattern with a ground-truth land cover classification map. These results indicate that machine learning has the ability to perform large-scale soil heavy metal mapping from widely available satellite remote sensing images.

Chlorophyll-a is an important parameter that characterizes the eutrophication of water bodies. The advantage of ZY1-02D hyperspectral satellite subdivision in the visible light and near-infrared bands is that it highlights the unique characteristics of water bodies in the spectral dimension, and it helps to assess the Class II water bodies of inland lakes and reservoirs, making it an important tool for refined remote sensing detection of the environment. In this study, the Baiyangdian Nature Reserve in northern China, which contains a typical inland lake and wetland, was chosen as the study area. Using ZY1-02D hyperspectral synchronization transit images and in situ measured chlorophyll-a concentration as the data source, remote sensing of the chlorophyll-a concentration of inland lakes was conducted. By analyzing the correlation between the spectral reflectance of the ZY1-02D hyperspectral image and the chlorophyll-a concentration and using algorithms such as the single band, band ratio, and three bands to compare and filter characteristic wavelengths, a quantitative hyperspectral model of the chlorophyll-a concentration was established to determine the chlorophyll-a concentration of Baiyangdian Lake. The dynamic monitoring of the water body and the assessment of the nutritional status of the water body were determined. The results revealed that the estimation of the chlorophyll-a concentration of Baiyangdian Lake based on the hyperspectral Fluorescence Line Height (FLH) model was ideal, with an R2 value of 0.78. The FLH model not only comprehensively considers the effects of suspended solids, yellow substances, and backscattering of the water body on the estimation of the chlorophyll-a concentration, but also considers the influence of the elastic scattering efficiency of the chlorophyll. Based on the ZY1-02D hyperspectral data, a spatial distribution map of the chlorophyll-a concentration of Baiyangdian Lake was created to provide new ideas and technical support for monitoring inland water environments.

Nitrous oxide (N2O) is the third most abundant anthropogenous greenhouse gas (after carbon dioxide and methane), with a long atmospheric lifetime and a continuously increasing concentration due to human activities, making it an important gas to monitor. In this work, we present a new method to retrieve N2O concentration profiles (with up to two degrees of freedom) from each cloud-free satellite observation by the Infrared Atmospheric Sounding Interferometer (IASI), using spectral micro-windows in the N2O ν3 band, the Radiative Transfer for TOVS (RTTOV) tools and the Tikhonov regularization scheme. A time series of ten years (2011–2020) of IASI N2O profiles and integrated partial columns has been produced and validated with collocated ground-based Network for the Detection of Atmospheric Composition Change (NDACC) and Total Carbon Column Observing Network (TCCON) data. The importance of consistency in the ancillary data used for the retrieval for generating consistent time series has been demonstrated. The Nitrous Oxide Profiling from Infrared Radiances (NOPIR) N2O partial columns are of very good quality, with a positive bias of 1.8 to 4% with respect to the ground-based data, which is less than the sum of uncertainties of the compared values. At high latitudes, the comparisons are a bit worse, due to either a known bias in the ground-based data, or to a higher uncertainty in both ground-based and satellite retrievals.

Several global leaf area index (LAI) products were generated using neural networks, but the training dataset for the neural networks was sensor specific, and the construction of the training dataset was time consuming. In this paper, an unsupervised domain adaptation-based method was proposed to estimate LAI from the Visible Infrared Imaging Radiometer Suite (VIIRS) surface reflectance dataset based on a training dataset constructed from the Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance dataset. A transfer component analysis (TCA) algorithm was first utilized to map the MODIS and VIIRS surface reflectance into the same subspace to reduce the distribution discrepancies between the MODIS and VIIRS surface reflectance. Then, the embedded data obtained from MODIS surface reflectance dataset, along with the LAI values produced by fusing the MODIS and the Carbon cYcle and Change in Land Observational Products from an Ensemble of Satellites (CYCLOPES) products, were employed to train general regression neural networks (GRNNs). Finally, for retrieving the LAI values, the embedded data acquired from the VIIRS surface reflectance dataset was input into the trained GRNNs. For multiple field sites with different biome types, we used this developed method to retrieve LAI values based on the VIIRS surface reflectance dataset. The results indicate that, based on the training dataset built from MODIS surface reflectance dataset, the domain adaptation-based retrieval method can effectively estimate LAI values from VIIRS surface reflectance dataset. By comparison with the VIIRS and MODIS LAI products, the retrieved LAI values with TCA are more consistent with the reference LAI values acquired from high-resolution remote sensing images. The coefficient of determination (R2) and root mean square error (RMSE) of the retrieved LAI values with TCA at all selected sites are 0.88 and 0.68, respectively. Furthermore, the accuracy of the retrieved LAI values with TCA is higher than the retrieved LAI values without TCA with the R2 0.81 and the RMSE 0.79.

The spatial calculation of vector data is crucial for geochemical analysis in geological big data. However, large volumes of geochemical data make for inefficient management. Therefore, this study proposed a shapefile storage method based on MongoDB in GeoJSON form (SSMG) and a shapefile storage method based on PostgreSQL with open location code (OLC) geocoding (SSPOG) to solve the problem of low efficiency of electronic form management. The SSMG method consists of a JSONification tier and a cloud storage tier, while the SSPOG method consists of a geocoding tier, an extension tier, and a storage tier. Using MongoDB and PostgreSQL as databases, this study achieved two different types of high-throughput and high-efficiency methods for geochemical data storage and retrieval. Xinjiang, the largest province in China, was selected as the study area in which to test the proposed methods. Using geochemical data from shapefile as a data source, several experiments were performed to improve geochemical data storage efficiency and achieve efficient retrieval. The SSMG and SSPOG methods can be applied to improve geochemical data storage using different architectures, so as to achieve management of geochemical data organization in an efficient way, through time consumed and data compression ratio (DCR), in order to better support geological big data. The purpose of this study was to find ways to build a storage method that can improve the speed of geochemical data insertion and retrieval by using excellent big data technology to help us efficiently solve problem of geochemical data preprocessing and provide support for geochemical analysis.

The interferometric imaging radar altimeter (InIRA) aboard the Chinese Tiangong-2 space laboratory is the first spaceborne imaging radar working at low incidence angles. This study focuses on the retrieval of significant wave heights (SWHs) from InIRA data. The retrieved SWHs can be used for correcting the sea state bias of InIRA-derived sea surface heights and can supplement SWH products from other spaceborne sensors. First, we analyzed tilt, range bunching and velocity bunching wave modulations at low incidence angles, and we found clear dependencies between the SWH and two defined factors, range and azimuth integration, for ocean waves in the range and azimuth directions, respectively. These dependencies were further confirmed using InIRA measurements and collocated WaveWatch III (WW3) data. Then, an empirical orthogonal SWH model using the range and azimuth integration factors as model inputs was proposed. The model was segmented by the incidence angle, and the model coefficients were estimated by fitting the collocation at each incidence angle bin. Finally, the SWHs were retrieved from InIRA data using the proposed model. The retrievals were validated using both WW3 and altimeter (JASON2, JASON3, SARAL, and HY2A) SWHs. The validation with WW3 data shows a root mean square error (RMSE) of 0.43 m, while the average RMSE with all traditional altimeter data is 0.48 m. This indicates that the InIRA can be used to measure SWHs.

Moisture content in tidal flats changes frequently and spatially on account of tidal fluctuations, which greatly influence the reflectance of the tidal flat surface. Precise prediction of the spatial-temporal variation of tidal flats’ moisture content is an important foundation of surface bio-geophysical information research by remote sensing. In this paper, we first measured the multi-angle reflectance of soil samples obtained from tidal flats in the northeastern Dongtai, Jiangsu Province, China, in the laboratory. Then, based on the particle swarm optimization (PSO) algorithm, we retrieved the photometric characteristics of the soil surface by employing the SOILSPECT bidirectional reflectance model. Finally, the soil moisture content was retrieved by introducing the equivalent water thickness of the soil. The results showed that: (i) A significant correlation existed between the retrieved equivalent water thickness and the measured soil moisture content. The SOILSPECT model is capable of estimating soil moisture with high precision by using multi-angle reflectance. (ii) Retrieved values of single scattering albedo (ω) were consistent with the variation of soil moisture content. The roughness parameter (h) and the asymmetry factor (Θ) were consistent with the structure and particle composition of the soil surface in dry soil samples. (iii) When the soil samples were soaked with water, the roughness parameter (h) and the type of scattering on the soil surface both showed irregular changes. These results support the importance of using the measured soil particle size as one of the parameters for the retrieval of soil moisture content, which is a method that should be used cautiously, especially in tidal flats.

The retrieval of aerosol and cloud properties such as their optical thickness and/or layer/top height requires the selection of a model that describes their microphysical properties. We demonstrate that, if there is not enough information for an appropriate microphysical model selection, the solution’s accuracy can be improved if the model uncertainty is taken into account and appropriately quantified. For this purpose, we design a retrieval algorithm accounting for the uncertainty in model selection. The algorithm is based on (i) the computation of each model solution using the iteratively regularized Gauss–Newton method, (ii) the linearization of the forward model around the solution, and (iii) the maximum marginal likelihood estimation and the generalized cross-validation to estimate the optimal model. The algorithm is applied to the retrieval of aerosol optical thickness and aerosol layer height from synthetic measurements corresponding to the Earth Polychromatic Imaging Camera (EPIC) instrument onboard the Deep Space Climate Observatory (DSCOVR) satellite. Our numerical simulations show that the heuristic approach based on the thesolution minimizing the residual, which is frequently used in literature, is completely unrealistic when both the aerosol model and surface albedo are unknown.

Land surface temperature (LST) is an important variable in the physics of land–surface processes controlling the heat and water fluxes over the interface between the Earth’s surface and the atmosphere. Space-borne remote sensing provides the only feasible way for acquiring high-precision LST at temporal and spatial domain over the entire globe. Passive microwave (PMW) satellite observations have the capability to penetrate through clouds and can provide data under both clear and cloud conditions. Nonetheless, compared with thermal infrared data, PMW data suffer from lower spatial resolution and LST retrieval accuracy. Various methods for estimating LST from PMW satellite observations were proposed in the past few decades. This paper provides an extensive overview of these methods. We first present the theoretical basis for retrieving LST from PMW observations and then review the existing LST retrieval methods. These methods are mainly categorized into four types, i.e., empirical methods, semi-empirical methods, physically-based methods, and neural network methods. Advantages, limitations, and assumptions associated with each method are discussed. Prospects for future development to improve the performance of LST retrieval methods from PMW satellite observations are also recommended.

A large amount of fresh water resources are stored in the snowpack, which is the primary source of water for streamflow in many places at middle-to-high latitude areas. Therefore, snow water equivalent (SWE) is a key parameter in the water cycle. Active and passive microwave remote sensing methods have been used to retrieve SWE due to relatively poor resolution of current in situ interpolated maps with good accuracy. However, estimation of SWE has proved challenging, despite several decades of efforts to develop retrieval approaches. Active sensors provide higher-resolution observations. Two recent promising retrieval algorithms using active data are dual frequency dual polarization backscattered power and differential interferometry. These retrieval algorithms have some restrictions on snow characteristics, the environment, and instrument properties. The restrictions limit the snow that is suitable for the specific retrieval algorithm. In order to better understand how much of the snowpack satisfies the precondition of these retrieval approaches, we use a 4 km gridded snowpack product over the contiguous US for years 1997 and 2015. We use a simple scattering model to simulate the scattering characteristics of snow. The snow property maps, simulated scattering characteristics of snow, and environmental conditions are used to filter the suitable snow for each retrieval algorithm. We show that snow wetness and vegetation coverage are the two main limiting conditions for these retrieval algorithms. We show that 39% and 44% of the grid-points with snow satisfy the preconditions of dual polarization dual frequency retrieval algorithms at 13.5 GHz (one of the recommended frequencies for this algorithm in the literature) in 1997 and 2015, respectively. The most important limiting factors for dual polarization dual frequency retrieval method are dryness of snow, penetration depth, and vegetation-free constraints. The backscattered power in dual polarization dual frequency method is more sensitive to snow density and grain radius rather than to snow depth. We also show that 55% and 53% of the grid-points with snow satisfy the precondition of differential interferometry retrieval algorithms at 1 GHz (one of the recommended frequencies for this algorithm in the literature) in 1997 and 2015, respectively. The most important precondition-limiting factors for differential interferometry are dryness of snow and vegetation-free constraints. The differential interferometry phase retrieval algorithm is equally sensitive to snow height and snow density variations and is independent of snow grain radius.

Remote sensing of global ocean color is a valuable tool for understanding the ecology and biogeochemistry of the worlds oceans, and provides critical input to our knowledge of the global carbon cycle and the impacts of climate change. Ocean polarized reflectance contains information about the constituents of the upper ocean euphotic zone, such as colored dissolved organic matter (CDOM), sediments, phytoplankton, and pollutants. In order to retrieve the information on these constituents, remote sensing algorithms typically rely on radiative transfer models to interpret water color or remote-sensing reflectance; however, this can be resource-prohibitive for operational use due to the extensive CPU time involved in radiative transfer solutions. In this work, we report a fast model based on machine learning techniques, called Neural Network Reflectance Prediction Model (NNRPM), which can be used to predict ocean bidirectional polarized reflectance given inherent optical properties of ocean waters. This supervised model is trained using a large volume of data derived from radiative transfer simulations for coupled atmosphere and ocean systems using the successive order of scattering technique (SOS-CAOS). The performance of the model is validated against another large independent test dataset generated from SOS-CAOS. The model is able to predict both polarized and unpolarized reflectances with an absolute error (AE) less than 0.004 for 99% of test cases. We have also shown that the degree of linear polarization (DoLP) for unpolarized incident light can be predicted with an AE less than 0.002 for 99% of test cases. In general, the simulation time of SOS-CAOS depends on optical depth, and required accuracy. When comparing the average speeds of the NNRPM against the SOS-CAOS model for the same parameters, we see that the NNRPM is able to predict the Ocean BRDF 6000 times faster than SOS-CAOS. Both ultraviolet and visible wavelengths are included in the model to help differentiate between dissolved organic material and chlorophyll in the study of the open ocean and the coastal zone. The incorporation of this model into the retrieval algorithm will make the retrieval process more efficient, and thus applicable for operational use with global satellite observations.

Between 2006 and 2011, the Advanced Land Observing Satellite (ALOS) Phased Array L-type Synthetic Aperture Radar (PALSAR) instrument acquired multi-temporal datasets under several environmental conditions and multiple configurations of look angle and polarization. The extensive archive of SAR backscatter observations over the forest test sites of Krycklan (boreal) and Remningstorp (hemi-boreal), Sweden, was used to assess the retrieval of stem volume at stand level. The retrieval was based on the inversion of a simple Water Cloud Model with gaps; single estimates of stem volume are then combined to obtain the final multi-temporal estimate. The model matched the relationship between the SAR backscatter and the stem volume under all configurations. The retrieval relative Root Mean Square Error (RMSE) differed depending upon environmental conditions, polarization and look angle. Stem volume was best retrieved in Krycklan using only HV-polarized data acquired under unfrozen conditions with a look angle of 34.3° (relative RMSE: 44.0%). In Remningstorp, the smallest error was obtained using only HH-polarized data acquired under predominantly frozen conditions with a look angle of 34.3° (relative RMSE: 35.1%). The relative RMSE was below 30% for stands >20 ha, suggesting high accuracy of ALOS PALSAR estimates of stem volumes aggregated at moderate resolution.

The Geophysical Model Function (GMF) XMOD1 provides a linear algorithm for sea surface wind field retrievals for the Spaceborne Imaging Radar-C/X-band Synthetic Aperture Radar (SIR-C/X-SAR). However, the relationship between the normalized radar cross section (NRCS) and the sea surface wind speed, wind direction and incidence angles is non-linear. Therefore, in this paper, XMOD1 is revisited using the full dataset of X-SAR acquired over the ocean. We analyze the detailed relationship between the X-SAR NRCS, incidence angle and sea surface wind speed. Based on the C-band GMF CMOD_IFR2, an updated empirical retrieval model of the sea surface wind field called SIRX-MOD is derived. In situ buoy measurements and the scatterometer data of ERS-1/SCAT are used to validate the retrieved sea surface wind speeds from the X-SAR data with SIRX-MOD, which respectively yield biases of 0.13 m/s and 0.16 m/s and root mean square (RMS) errors of 1.83 m/s and 1.63 m/s.