摘要：Earth observation satellites, such as optical and SAR satellites, require processing such as imaging, radiometric/geometric correction, and continuous accumulation in order to provide high-precision, stable, and time continuous data and features for computer interpretation. Traditional medium and low resolution Earth observation satellites typically perform pixel-by-pixel processing based on the assumption of ideal point targets, which means that the ground object grid has a invariant time-frequency characteristic. However, the working modes of advanced satellite systems, such as high-resolution, wide-swath, large squint angle, and multi-channel, are more complex, and their data processing is very sensitive to the errors generated in the whole chain of the satellite to ground, which puts higher requirements on the accuracy of imaging parameter calibration or estimation. Hence, the method of assuming sensor pixels as ideal point targets for parameter estimation, imaging, and correction processing is no longer able to meet the processing accuracy requirements. Moreover, in recent years, the new development of multi-system satellite network collaboration and fusion applications has made it difficult to characterize and model the features of multi-source and multi-temporal data based on the current ideal point target assumption. To this end, this article proposes a new method for high-precision processing of multi-system remote sensing satellite imaging data. Firstly, the concept and characterization theory of “Hyper-pixel” are proposed, and an accurate imaging model based on hyper-pixels is established. Then, by mining stable features of hyper-pixels, and inspired by generative adversarial learning mechanisms, high-precision estimation and continuous refinement of high coupling imaging parameters are achieved. This effectively improves the accuracy of multi-system remote sensing satellite imaging data products, and provides better data input for computer interpretation.
摘要：Image adaptive filtering is a nonlinear image transformation, which has a wide range of applications. Traditional image adaptive filters are designed by experts, such as bilateral filter and shape adaptive filter. They can determine the shape, size, and weight of the filter based on the local structure and content of the image. They are commonly used to suppress noise while preserving the structural characteristics of the image. Convolutional Neural Networks (CNNs) are an effective tool for feature extraction and nonlinear expression. They can be used to learn and construct image adaptive filters. And this paper explores the application of nonlinear image adaptive filters generated by convolutional neural networks in image interpolation and image fusion.This paper introduces the generation network of an image adaptive filter, including its model structure and objective function. The common network structure usually employs an encoder-decoder architecture, which is mainly composed of three parts: feature extraction, feature recovery, and filter (convolution kernel) estimation. Then, the paper presents two different application scenarios of image adaptive filters: image interpolation and image fusion. The adaptive filter for images enables transformation between different phases during image interpolation and transformation between different bands during image fusion. In these two scenarios, the image adaptive filters are learned by the filter generation network based on the specific application scenario and then applied. In image interpolation applications, the image adaptive filter is used as a nonlinear transformation between two temporal images. The interpolated image is looked at as the mean of adaptive filtering of the previous temporal image and adaptive filtering of the latter temporal image. In image fusion applications, the image adaptive filter is used as a nonlinear fitting method to regress multispectral bands to the panchromatic band. It then extracts spatial details from the difference of the panchromatic band and the simulated panchromatic band, and finally adds spatial details to all the multispectral bands.We conducted experiments in two application scenarios. The first involved nonlinear transformation for image interpolation with different phases simultaneously. The second utilized an image adaptive filter as a nonlinear fitting method for multi-spectral band regression panchromatic band in image fusion. In image interpolation applications, the experimental results show that the interpolated results are consistent with the reference image in spatial and spectral characteristics, and the RMSE of the interpolated image with the reference image is relatively small. The experimental results for image fusion applications indicate that the low-resolution panchromatic band obtained through adaptive filter fitting of the multi-spectral band is more accurate than the traditional component replacement method. The fusion result obtained by nonlinear image adaptive filters has neither obvious spectral distortion nor obvious spatial distortion.From the application of nonlinear image adaptive filters generated by convolutional neural networks in image interpolation and image fusion, we have a glimpse of its application potential of image adaptive filter in constructing image nonlinear transformation. The filter generation network can generate adaptive filters for particular application scenarios, resulting in more accurate and visually pleasing images.
摘要：Land Surface Temperature (LST) is an important parameter for characterizing the surface–air exchange process, which plays an important role in climate change, ecological monitoring, hydrological simulation, and other studies. The traditional LST estimated from Thermal Infrared (TIR) remote sensing is mature in terms of retrieval methods, data production, and quality control. However, the TIR LST has considerable missing data under clouds because of the limitation that the TIR radiation from the ground surface cannot penetrate the clouds. In addition, Passive Microwave (PMW) remote sensing has disadvantages, such as strip gaps and coarse spatial resolution, because of the limitations of the physical mechanisms and imaging methods. Therefore, the all-weather LST unaffected by cloudiness must be obtained to support the subsequent studies. In the present study, we review and organize the basic principles and methods of the acquisition of all-weather LST. The methods are classified into two categories: (i) all-weather LST reconstruction from effective observation and (ii) multisource data integration.The comparative analysis indicates that multisource data integration can combine the advantages of TIR, PMW, and reanalysis data. Thus, it has the highest research value and potential for further research. Multisource data integration can be employed to obtain global long-time all-weather LST products characterized by spatial and temporal continuity. The LST retrieved from PMW remote sensing suffers from coarse spatial resolution and strip gaps. However, it is still an effective method of obtaining land surface information under clouds and an important input parameter for multisource data integration. The reconstructions of all-weather LST based on effective observation only apply to small areas with cloud cover in short periods. They are not practicable for long-term cloudy areas.From the analysis and conclusion, this study also collects and analyzes information about five currently released all-weather surface temperature products. The advantages and disadvantages of the existing products are also summarized. A global all-weather LST product with high quality and spatial resolution is urgently needed by the scientific community. After reviewing the all-weather LST products, we further summarize the applications of all-weather LST. Its applications are still in their infancy. Research on the applications of all-weather LST is relatively small in the current stage. However, all-weather LST has great potential for applications when its products further mature.Finally, further study directions and theoretical development of all-weather LST are discussed and prospected. First, with PMW LST as the basis for all-weather LST, two issues must be addressed: (i) filling the PMW LST strip gap to make the PMW surface temperature a complete spatial coverage; (ii) correcting thermal sampling depth to make that PMW LST obtain the same physical meaning as TIR LST. The reason is that the inconsistent observation caused by the varying thermal sampling depth is the actual reason for the inconsistent physical meaning of PMW and TIR observation information. Second, we should further strengthen the study on estimating all-weather LST from multisource data. The current study of multisource data integration is still in the preliminary stage, and no systematic and effective integration strategy has been developed. Third, the scientific community should enhance the production, publication, and application of all-weather LST products. Few all-weather LST products can be directly applied by users. Generating all-weather LST products with global spatial and temporal continuity and high spatial resolution should be the task of an all-weather LST study. Besides improving the data quality and reliability of all-weather LST, focusing on the operability and cost of the method in practical applications is necessary to make the all-weather LST usable data, thereby truly promoting the progress of the related studies.
关键词：remote sensing;all-weather land surface temperature;reconstruction;interpolation;multi-source data integration
摘要：We investigate the worldwide monitoring of land surface water by satellite remote sensing and the corresponding ability of Chinese satellites for the 14th five-year plan under the general goals of the Ministry of Natural Resources of the People’s Republic of China to plan for the new generation of satellites for water resource monitoring. First, this work reviews the current status of the water resource monitoring by the Chinese and international satellites from several perspectives, including liquid surface water (water extent, water level, water volume, water temperature, and water quality), solid surface water (glacier, snow, and frozen ground), and water vapor in the atmosphere. Then, the capability of land natural resources satellites for land surface water monitoring is inspected. Afterward, the ability of water resources monitoring with various types of remote sensing satellites, including optical, laser, RADAR, and gravity satellites, is summarized and analyzed. Advice and suggestions for Chinese satellite planning of water resources monitoring are proposed by concentrating on the current status and the shortage of water resource monitoring with satellite remote sensing in China. The advice and suggestions include planning the observation, technique, product, and service systems. First, a new generation of cloud water resources monitoring satellites combining infrared and active/passive microwaves is recommended to be developed for the observation system. Moreover, the evolution of radar satellites should be accelerated to make up for the deficiency of optical satellites. Altimetry satellites and gravity satellites must be vigorously cultivated. Furthermore, small satellite constellations for water resources monitoring integrating satellite “communication-navigation-remote sensing” should be promoted. Advanced thermal infrared and hyperspectral satellites with strong temporal and spatial resolution are also recommended to be developed. Second, for the technique system, exploring general remote sensing data processing technologies, including data correction/splicing technology and multisensor data fusion technology, are recommended to improve the quality of domestic satellite data for operational water resources monitoring. Moreover, water resource element extraction/retrieval models must be promoted. The techniques for high-quality long-term water resource products should be also developed. Finally, for the service system, providing a dataset-sharing service of the long-term global water cycle flux and storage elements with high spatiotemporal granularity is recommended. Moreover, the overall development of Chinese satellites for water resources monitoring has started from scratch toward boosting, and these natural resource satellites have basic capabilities for water resources survey. Natural resources satellite datasets are abundant. However, nationwide long-term series of water resources data mainly based on domestically made satellites remain lacking. This gap can be improved from two perspectives.Developing operational natural resources retrieval models from the data perspective following the scientific concept of “remote sensing big data + artificial intelligence” is necessary for timely acquisition, processing, distribution, and providing service while ensuring the stability of satellite remote sensing data. It can realize all-time, all-weather, and all-element satellite remote sensing monitoring for natural resources based on cloud services. From the satellite perspective, the next step for on-orbit satellites is to produce application-oriented operational water resource element products, combining multisource satellite data on the premise of improving satellite data quality. For satellites under planning, relevant departments should work closely to establish goals with different priorities under the guidance of scientific and application issues and full consideration of costs.
关键词：Land surface water resources;water resource monitoring by satellite remote sensing;remote sensing big data;satellite plan suggestions
摘要：Rainfall is a key link in the earth’s water cycle and one of the most important inputs to hydrological and land surface models. Thus far, radar-based rainfall information has been widely used in global surface hydrology process simulation, disastrous weather forecast, flood control, and disaster relief because it provides data with a high spatial and temporal resolution that improves rainfall representation. In recent years, precipitation radar has gradually developed in a multiangle, multifrequency, dual-polarization, multiresolution, and multiantenna direction under different meteorological observation requirements and rainfall application scenarios. The radar observes rainfall in the air. Meteorological and hydrological research have different requirements for rainfall observation on the temporal and spatial scales. Thus, a series of conversion processes from radar rainfall observation in the air to the land surface generates considerable uncertainties. Thus, the accuracy of radar precipitation must be improved by systematic deviation corrections and data processing.In this study, a comprehensive review of the process of radar rainfall inversion aims to provide a general picture of the current state of multimode radar technology. First, the development characteristics of multimode radar remote sensing technology were summarized. Afterward, the basic process of converting aerial rainfall observed by radar to the surface of the land surface was sorted out. Furthermore, we reviewed and summarized the major progress and methods of precipitation radar rainfall inversion. The upscaling and downscaling applications in hydrological and land surface models were compared in the literature review. Then, we analyzed the factors causing the air-land surface rainfall deviation in the radar, such as raindrop evaporation, drift, and fragmentation. The calculation method of raindrop evolution deviation was also summarized. Moreover, we summarized the current methods for correcting radar rainfall based on ground-based reference rainfall, such as the rainfall observed by rain gauges. Finally, on the basis of this review, we discussed the existing major challenges and prospects in multimode radar precipitation, including developing multiscale inversion of surface rainfall, multimodel surface rainfall data fusion, surface rainfall inversion considering microphysical deviation of raindrops, and multimode radar data mining based on machine learning.
摘要：The frequency of extreme rainfall and flooding in North China has increased because of the influence of climate change and human activities. Convective and strong precipitation processes occur in summer. Under the influence of the mixed flow generation mechanism in semihumid and semiarid areas, the flood burst is strong and difficult to forecast. Based on the Weather Research Forecast (WRF) model, namely, coupled WRF-Hydro, this study uses three-dimensional variational data assimilation (3DVAR) in constructing the WRF-3DVAR assimilation system for a rapid hourly update to assimilate high spatial and temporal resolution radar reflectivity data with the traditional meteorological observed data from the Global Telecommunication System (GTS). The study of rainfall-runoff prediction based on the land-atmosphere coupling is conducted by taking the typical rainfall processes of the north and south branches of the Daqinghe River Basin as the research object. Moreover, the performance of the rainfall-runoff prediction method in North China is further verified. The research results have some theoretical and practical values for constructing the data assimilation system of the atmospheric model and flood forecast practice in northern China.We employ three nested domains and adopt the GFS data for driving the WRF model. This study evaluates the improvement effect of WRF on forecasting rainfall and WRF-Hydro forecasting runoff by assimilating radar reflectivity and GTS data. The GTS data are released every 6 h. Thus, in the hourly assimilation scheme, GTS is only assimilated at the 6th, 12th, 18th, and 24th h from the start of the storm. However, radar reflectivity is set to assimilate once every hour. The rainfall evaluation indexes include Root Mean Square Error (RMSE), Mean Bias Error (MBE), and Critical Success Index (CSI). CSI/RMSE is a comprehensive index for evaluating rainfall forecast results. RMSE, MBE, and Nash (Nash-Sutcliffe efficiency coefficient) are used to evaluate runoff.The results show that the precipitation forecasted by the WRF model is always lower than the observed rainfall. However, assimilation systems can increase rainfall. The improved initial conditions in the WRF-3DVAR system via radar data assimilation and GTS data achieve good short-term and convectively strong precipitation. The high assimilation frequency significantly helps trigger and maintain the convective activities in the 3DVAR framework and the storm case applied. The assimilation weather radar combined with the traditional meteorological observed data can effectively improve the rainfall prediction accuracy of the WRF model, particularly for the rainfall with uniform spatial and temporal distribution. The CSI/RMSE index of the forecast rainfall after assimilation is increased by 23.24%—50.00%. Whether data assimilation is carried out or not, the CSI index results show different degrees of rainfall false alarm frequency. In the runoff forecast, the accurate rainfall forecast after data assimilation also improves the runoff forecast results to a certain extent. The peak flow error is reduced by 15.05%, 38.07%, 18.53% and 6.99%, the flood volume error is reduced by 25.99%, 29.32%, 26.02% and 23.95%, and the Nash efficiency coefficient is increased by 0.25, 0.25, 0.29 and 0.48, respectively. However, the forecast results of flood peak discharge and the peak occurrence time for the rainfall with uneven spatial and temporal distribution, large magnitude, and slow water retreat are still not ideal. Moreover, subsequent improvements should be made in terms of accurate calibration of hydrological parameters and real-time correction of forecast errors. The accuracy of the WRF-Hydro runoff forecast in the mixed runoff generation areas of northern China mainly depends on two aspects. One is the accuracy of the rainfall forecast of WRF model, which is related to the driving data and rainfall distribution type. For the rainfall with uneven spatial-temporal distribution, the poor rainfall forecast indirectly affects the runoff forecast effect. On the contrary, it is related to different runoff characteristics, such as the complexity of the runoff process, the magnitude of runoff, the presence or absence of base flow in the early stage, and the soil water content. Data assimilation improves rainfall forecast. Thus, the runoff forecast results of WRF-Hydro are improved to a certain extent by reasonably using a basic flow module, improving land surface initial conditions, such as soil water content, and combining with effective real-time correction technology.
摘要：Ocean rainfall has an important impact on the global atmospheric cycle and local climate. Monitoring rain cells from remote sensing images is vital for ocean weather prediction. The ability of Synthetic Aperture Radar (SAR) to probe with a wide swath and high spatial resolution makes it an effective observation approach for rain cells with a scale of 10—30 km. This study uses the fusion-feature-based Broad Learning System (BLS) to detect the rain cells. The SAR images dataset composed of nine sea surface phenomena obtained by Sentinel-1 wave mode is also used. Results show that the detection accuracy is 98.51%, and the recall rate is 95.24%. These values are equivalent to those of the ResNet50 pretrained model. However, the training time of ResNet50 is 20 times that of BLS under the same calculation conditions. Compared with the structure of a deep learning network, that of BLS is flexible. That is, the model can be optimized and updated by adding nodes or input data. The experiments show that the node incremental learning of BLS can update the model without retraining the whole model. Following the advantages of the incremental learning and retraining schemes, this study proposed a hybrid model-updating scheme for the model-updating task caused by the expansion of the training dataset. This new scheme can ensure the high accuracy of the model and significantly reduce the time cost for model updating.
摘要：The raindrop size distribution (DSD) is used to describe the distribution of raindrop diameters during the rainfall process, which can effectively reflect the microphysical characteristics of raindrops. DSD-derived empirical relationships, including radar reflectivity factor-rainfall intensity (Z-R) and unit rainfall kinetic energy-rainfall intensity (KE-R), are key factors in research fields such as radar quantitative precipitation estimation and soil erosion assessment. At present, the ground disdrometer is generally used to obtain DSD directly at a given site, which is difficult to represent the spatial difference of the large-scale raindrop microphysical process. The dual-frequency precipitation radar (DPR) carried by the global precipitation measurement mission (GPM) core satellite can receive radar echoes of two different bands to obtain more raindrop information, which makes it possible to retrieve spatial three-dimensional DSD parameters. Based on the DSD surface estimations, including the mass weighted mean drop diameter (Dm) and normalised intercept parameter (Nw) of GPM-DRP in its 2ADPR product during the entire 4 years (2017—2020), this study calculated rainfall intensity, unit rainfall kinetic energy, radar reflectivity factor and other parameters of each record, constructed the empirical relationships between Z-R and KE-R in grid scale, and used the observation DSD data of 11 disdrometer stations in the Yangtze River Delta region as a reference to verify and evaluate the reliability of GPM-DPR to estimate DSD parameters and fit microphysical empirical formulas, which is useful for improving the accuracy of large-scale radar rainfall estimation and soil protection decision-making. The results showed that by comparing the DSD estimation of disdrometers and GPM-DPR, it can be found that under the same rainfall intensity class, DPR-derived Dm at most sites is slightly larger than the measured result of the disdrometer at the same location, while DPR-derived Nw is higher than that of the disdrometer. As for rainfall types, the raindrops of disdrometer are mostly stratiform rain, and only a small part is high-intensity convective rain. However, due to radar sensitivity limitation, the DPR-detected raindrops are almost completely distributed in the stratiform. For the empirical formula fitted by rainfall characteristics, the KEs derived from DPR are mainly distributed on both sides of the KE-R empirical formula by corresponding disdromters. In addition, Pearson coefficient of most stations can reach more than 0.60, and at Nantong, Jiaxing and other sites, it even exceed 0.70, which proves that DPR is suitable for inferring the empirical relationship between KE and R. It means that DPR has the ability to infer those rainfall microphysical relationships in place of disdrometers in areas where site data is scarce. What is more, the DPR results perform best when it exceeds 0.5 mm h-1, with small errors and high correlations. Overall, DPR remote sensing has good DSD inversion performance, which is expected to provide new support for large-scale radar quantitative precipitation estimation and soil retention decision-making. However, due to the characteristics of orbital scanning by spaceborne radar, DPR cannot make continuous observations of rainfall events in the same area, which makes it detect a low amount of data in a limited orbital range and limits its application ability to detect rainfall events. On one hand, in order to achieve a more accurate estimation of the rainfall microphysical characteristics, it is necessary to obtain DPR data with a longer duration. On the other hand, the DPR data can be used as a correction tool to be integrated with numerical weath.
关键词：rainfall;Disdrometer;radar remote sensing;GPM;Rainfall Kinetic Energy
摘要：Radiative Transfer Theory (RTT) is one of the essential foundations in astrophysics, engineering thermophysics, computer graphics, biomedical imaging, and remote sensing. RTT is particularly widely used in the field of quantitative remote sensing. However, classical RTT is a phenomenological theory based on heuristic summarizations of experiments instead of being directly derived from the first principles. Given the ignorance of the wave property, RTT cannot explain interference and diffraction phenomena, e.g., the well-known coherent backscattering.The root of RTT dates back to the photometry study by Bouguer, Lambert, and Beer. Von Lommel and Chwolson are believed to propose the integral form of the Radiative Transfer Equation (RTE) for the first time in the 1880s. Afterward, many other scientists, including Schuster, Schwarzschild, Eddington, Milne, Gans, Sobolev, Chandrasekhar, Rozenberg, and Tsang, contributed to the establishment of RTT as a strict theory. However, classical RTT implicitly depends on the assumption of independent scattering, which fails when applied to dense matter. It requires the first-principle approach to bridge the gap between classical RTT and classical electromagnetics and extend the application of RTT. Three ways can be applied: (1) direct derivation, (2) numerical simulations, and (3) controlled experiments.Direct derivation of the RTE from first principles (i.e., Maxwell equations) is the most fundamental approach. Mishchenko and his colleagues’ derivation is currently considered the most rigorous. This derivation is primarily based on previous research on multiple scattering of electromagnetic waves, to which Foldy, Lax, Twersky, and many others have made significant contributions. Mishchenko et al. managed to derive the RTE from Maxwell equations for both coherent and incoherent intensities under the condition of plane wave and discrete random media. The derivation proves that RTT is not a disconnected “island” from the “mainland” of classical electromagnetics.Besides derivations, numerical simulations and controlled experiments help reveal the connection between RTT and numerically exacted computational electromagnetics. In these simulations and experiments, the RTT and electromagnetic computation results are compared under different conditions. Results show that RTT can yield satisfactory results when the volume percentage of scatterers is low. Some corrections, e.g., the Percus-Yevick model, can be introduced to compensate for the errors of RTT when the density of scatterers further increases. Based on these studies, some efforts have been made to extend RTT to the case of dense matter. Notable achievements include the DMRT and the R2T2 theories.Although these studies are still limited to some ideal situations, they have provided some guidance for the mechanistic revision of RTT, thereby expanding the scope of its application. On the contrary, the combination of radiative transfer methods with computational electromagnetics becomes a research direction of interest, along with the development of computer performance and the improvement of relevant algorithms. At present, different methods are used in quantitative remote sensing for different wavebands and different research objects: for example, optical remote sensing and microwave remote sensing for vegetation or vegetation remote sensing and atmospheric remote sensing. Although the names of the methods used are “radiative transfer,” they are based on different assumptions and approximations. The combination of RTT and computational electromagnetics is a promising approach to unifying the remote sensing modeling and inversion studies of different wavelengths and objects.
摘要：Total Phosphorus (TP) and Total Nitrogen (TN) are important indicators of water quality eutrophication and the main parameters for water quality monitoring. Water quality monitoring by spectroscopy has become a hot spot in the current remote sensing water environment research because it is rapid and efficient and has no secondary pollution. The usual TP and TN inversion models are established based on the laboratory configuration standard solution for spectral measurement or the modeling based on the full samples. The model constructed in this way has a good regression effect. However, the actual water body causes the mutual influence of various water quality parameters, the concentration distributions of TP and TN are not uniform, and the predicted value may exceed the training sample range, making the actual prediction effect always unsatisfactory.In this study, the actual water samples in the Baiyangdian area are used as the input values of the inversion model. First, the measured spectral data and the chemical analysis values of TP and TN are used to compare the relationship between the correlation values of different reflectances and water quality parameters. Various inversion models have been constructed for the best relevant bands. The most stable and accurate modeling method has been determined through comparison. Therefore, the modeling samples are divided into uniform, high-value, low-value, median-value, and max-min-value samples according to concentration. Then, the influence of the sample modeling with different concentration ranges on the inversion model is discussed. The model’s predictive ability for samples with concentration values beyond modeling is determined.The extraction results of the characteristic wavebands in the range of 400—100 nm indicate that the reflectance correlation coefficient of TP and TN corresponding to a single wavelength is less than 0.3, which is not high; the maximum correlation coefficient with the first-order value of reflectance is 0.76, which is a moderate correlation; the correlation coefficients with the reflectance ratio are all over 0.8, which is highly correlated. In the inversion effect of the linear regression model, the exponential method and the logarithmic method are inferior to the multiple power method. Moreover, the effect of high power is better than low power. However, the overall effect is not ideal. The model’s R2 is less than 0.6. When the range of the modeling sample concentration is different, the R2 of the model is also different. The result is as follows: max-min method > uniform method > high-value method> middle-value method > low-value method. When the modeling sample concentration covers the predicted sample, the inversion model determination coefficient is R2 > 0.6. The average deviation of the predicted value (ARE) of TP and TN concentrations is less than 20%. When the modeled concentration is higher than the predicted sample, the R2 is approximately 0.6. The predicted value within 12% of the overconcentration range has an ARE of <25%. When the modeling concentration is lower than the predicted value, the R2 is between 0.4 and 0.5. The ARE is ≤30% when the predicted value exceeds the modeling sample concentration. When the sample concentration is on both sides of the predicted value, R2 can reach 0.8, and ARE is <25%. The following is obtained when the modeled sample concentration is between the predicted values: 0.45 < R2 < 0.55; ARE > 35%.When the reflectance method for TP and TN inversion was used, the ratio method can be given priority to the characteristic band when modeling. The regression effect of the partial least square method is significantly better than that of multiple power and exponential models. The model also has a clear physical meaning. Thus, it can be used in the regression study of TP and TN based on reflectance. For the predicted value that is not within the range of the modeled sample concentration, the credibility of the inversion results can be judged based on the relative relationship between its concentration value and the modeled sample concentration value.
关键词：Water quality monitoring;total phosphorus;total nitrogen;concentration inversion;partial least square method
摘要：Tree volume is an important parameter in forest inventory. The reconstruction of the Quantitative Structure Model of Trees (TreeQSM) method based on ground-based LiDAR point clouds can achieve nondestructive acquisition of forest volume. It can also solve the time-consuming and labor-intensive problem of traditional forest in situ investigation. However, the reference volume of the felled timber is difficult to obtain. Thus, the ability of the TreeQSM volume estimation has not been studied at the stem and different branch orders. Moreover, TreeQSM is only applied to the ground-based LiDAR point cloud collected at the tree level but not at the plot level. Therefore, this study proposes to assess the stem and branch volume estimation of TreeQSM from the point cloud collected from the tree and plot levels.In this study, we evaluate the stem and branch volume estimated by TreeQSM using TLS point cloud at the tree and plot levels:(1) Estimating the volume of stem and branch at different orders based on TLS scanning at the tree-level.(2) Estimating and comparing the volume of stem and branch at different orders based on TLS point cloud at the tree and plot levels.(3) Exploring the influence of stand density on the estimation of stem and branch volumes using the TLS point cloud at the plot level.The experimental results showed that the stem and first-order branch volume can be effectively estimated from the point cloud collected from the tree and plot levels. However, the volume estimation of the secondary branch has obvious deviations. At the plot level, the accuracy of the stem and whole tree volume is equivalent to that of the tree level. The deviations are approximately 5% and 10%. However, the first-order branch volume estimation deviation is slightly large, approximately 10% and 15% at the tree and plot levels, respectively. In addition, the stand density is negatively correlated with the accuracy of volume estimation at the plot level. In the low forest density (425, 625, and 925 plants/ha), the stem volume estimation error is within 5%, and the first-order branch volume estimation error is approximately 15%. In addition, the estimation deviations of the total volume in the plot are affected by the partial neutralization effect of the underestimation of the stem and the first-order branch volume and the overestimation of the secondary branch volume. These deviations are all approximately 10%. Thus, it can well estimate the tree stem, first-order branch, and whole tree volumes at the plot level in forests with a low stand density.
关键词：volume;Stem and branch volume;Quantitative Structure Model (QSM);TLS;plot level
摘要：The trade-offs and synergies of ecosystem services are important indicators for understanding regional ecological evolution mechanisms. Many studies have taken administrative regions as a whole to describe the trade-offs and synergies of ecosystem services. However, the researchs on the spatial heterogeneity of regional ecosystem services and the different impact of various ecosystem services remains lacking. In our study, the Fenhe River basin was taken as the study area. Based on the land use classification results obtained by interpreting remote sensing images of 1986, 1995, 2005, and 2015, statistical data, and ecological data, we used the multiscale geographically weighted regression (MGWR) to analyze the trade-off and synergy between ecosystem service’s indicators of the Fenhe River basin. The influence of land transfer on the trade-off synergy and the evolution mechanism of the Fenhe River Basin’s ecosystem services was quantitatively analyzed by combining MGWR with the land use. The results show that: (1) MGWR can detect the spatial heterogeneity in the trade-off synergy of ecosystem service, and thus accurately revealing the internal relationship between multiple ecosystem services. (2) Using MGWR to obtain the best combination of ecosystem services can effectively reduce data redundancy and improve work efficiency. (3) Land use conversion is a major driver of the evolution of ecosystem service’s trade-off synergy. (4) Vegetation degradation in the Fenhe River basin will reduce the trade-off between grain production and water yield and decrease the synergy between other ecological services like the synergy of biological diversity and water yield. (5) From 1986 to 2015, the synergistic relationship between individual ecosystem services in the Fenhe River basin was dominant. The synergistic relationship between water conservation and water yield was the most significant, with an average synergistic rate of 75.6%. The point of this study is the use of MGWR in solving the spatial heterogeneity of regional ecosystem service. We found that MGWR can not only reduce the analysis error caused by spatial heterogeneity but also obtain the optimal ecosystem service combination efficiently, which means it can reduce the data redundancy of ecosystem service analysis. The results can provide a reference for land use optimization and ecosystem service improvement in the Fenhe River basin.
关键词：remote sensing;ecosystem services;Trade-offs and Synergies;MGWR;land use;Fenhe River basin
摘要：Satellite- and ground-based remote sensing methods have unique advantages in monitoring atmospheric pollutants. The comparison and verification of different remote sensing data and collaborative observations use different monitoring platforms, which play a major role in accurately assessing changes in atmospheric pollution. In this study, the tropospheric NO2 vertical column amounts in the winter from November 2018 to February 2019 at the Beijing site were retrieved using the MAX-DOAS spectrometer deployed at the Beijing site. Moreover, the daily and monthly changes in NO2 in Beijing were summarized. The MAX-DOAS spectrometer was also used with TROPOMI’s products to analyze the NO2 pollution during winter in Beijing.The MAX-DOAS measurement spectrum combined with the DOAS inversion algorithm was used to obtain the vertical column amounts of tropospheric NO2 at different times and compare the changes and correlation of the NO2 columns obtained by TROPOMI at the time of satellite overpass. It was also used to analyze the sensitivity of the NO2 columns of ground-based and spaceborne observations at different sampling times and the average sampling distance between the satellite and the ground site at the time of passing territory. Moreover, we counted wind fields in winter weather conditions. The influence of the wind field on the changes in NO2 in Beijing was also analyzed. The two-factor analysis of variance was applied to evaluate the influence of the wind field on the change in the regional NO2 amounts.The results show that the average columns of NO2 in the troposphere in Beijing in November are higher than those in other months in winter. The maximum hourly average columns can reach 4.04 × 1016 molec·cm-2. The average amounts of NO2 in the troposphere in the afternoon of each winter month are significantly higher than those in the morning. The tropospheric NO2 obtained by TROPOMI and MAX-DOAS has a good correlation (r = 0.88). The correlation between satellite-ground-based observations in December 2018 can reach 0.96. However, the NO2 amounts of TROPOMI are overestimated to varying degrees relative to the ground-based MAX-DOAS observation results. Moreover, the sensitivity of satellite-ground comparison shows that within a certain sampling range, the correlation appears to increase significantly with the increase in average time and average distance. The correlation is sensitive to the sampling distance, whereas the relative column deviation is sensitive to the sampling time. This finding provides a reference for the selection of reasonable sampling intervals during data comparison. In addition, wind field analysis indicates that wind speed and the interaction between wind speed and wind direction are the main factors leading to changes in NO2 in Beijing.The results of monitoring NO2 in winter on different platforms indicate that the NO2 in the Beijing area has obvious monthly and diurnal changes. This monitoring is vital for establishing pollution forecasting models and analyzing pollution causes. The comparative observation and sampling sensitivity analysis of the two different observation platforms also provide important reference and data support for the reliability of NO2 inversion on the spaceborne platform.
关键词：NO2;TROPOMI;MAX-DOAS;Two-way ANOVA;changing trend;remote sensing;DOAS;comparison and validation
摘要：The silicification alteration zone is an important prospecting indicator for many kinds of metallic hydrothermal deposits. Identifying the silicification alteration information rapidly and widely by remote sensing technology is vital. However, the width of the silicide alteration zone is only a few meters to hundreds of meters, and it appears as a thin band of a fine ribbon. At present, most of the commonly used thermal infrared data have low spatial resolution and fuzzy spatial details, which limit their applications in identifying microsilicide alteration zones. Based on the study of the land surface temperature anomaly characteristics of the siliceous alteration zone in the Rencha basin, the enhancement recognition model of siliceous alteration with weak information is established to strengthen the recognition and analysis of the siliceous alteration information.In this study, the PAN image high-frequency component map and TIR image are weighted and fused by taking the uranium mining area in the Rencha basin of Guangdong Province as an example. Then, a remote sensing model of silicified alteration weakening information enhancement identification is established. This model is based on the analysis of the topographic characteristics of the silicified alteration zone in the Rencha basin and the discussion of its relationship with LST anomalies.The results show that the weighted fusion of the PAN and TIR image can improve the spatial resolution of the TIR image and preserve the spectral characteristics of the thermal target effectively. In addition, the LST image slightly changes before and after fusion, and the latter improves the ability to identify the thermal anomalies of ground objects. Finally, the silicatization alteration zone, a high-temperature anomaly zone smaller than the fault zone in the approximate LST diagram, is predicted and identified accordingly.In conclusion, the information on the silicified alteration zone in the Rencha basin can be enhanced by the LST retrieval of the weighted fusion image. The silicified bands predicted and identified via this method are consistent with the known ones. This finding illustrates that this method is well applied to identifying the silicified alteration zone in the uranium mining area of the Rencha basin.
关键词：RenCha Basin;Thermal Infrared (TIR);image fusion;Silicified Alteration Zone
摘要：In ASTER images, different lithologic units show obvious multiscale texture features, and wavelet transform has the advantage of extracting multiscale features. Support Vector Machine (SVM) is suitable for solving the classification problem of little training data and nonnormal data distribution. SVM is used to complete lithology classification. The classification results have high classification accuracy and low uncertainty. Using the voting method in selecting lithologic classification results can avoid the uncertainty of lithologic classification results caused by the extraction method of lithologic samples, thereby making the classification results statistically significant. An automatic classification method for ASTER image lithology integrating the wavelet texture, SVM, and voting method is proposed to improve the accuracy of ASTER imagery exploited for mapping assistance. First, the Haar wavelet is utilized for decomposing the ASTER image involving a multiscale wavelet, with the mean value of wavelet coefficients considered texture features. Moreover, the variance, homogeneity, and mean values of the gray-level co-occurrence matrix (GLCM) are extracted concurrently. Then, the feature vectors of the SVM classification are constructed with multiscale texture, GLCM texture, and spectral features. The classification is repeated 10 times. Finally, the lithologic unit is determined by the voting method, and the results are statistically evaluated. The lithologic classification involves 92.1934% accuracy, exceeding the accuracy of spectral classification by 13.3369%, with a kappa coefficient of 0.9202. The multiscale texture extracts detailed lithologic information. The voting method prevents the dynamic lithologic change caused by the spatial variability of samples. The SVM also demonstrates superiority over the maximum likelihood classifier for lithologic classification involving high-dimensional and nonnormal distribution data. The local optimal parameters of SVM are avoided using the artificial bee colony algorithm to search for optimal parameters.
摘要：Hydrocarbon microleakage of oil and gas resources (including coalbed methane) may induce spectral changes in surface soil and vegetation. Detecting surface hydrocarbon microleakage using remote sensing technology, a new method for early exploration of coalbed methane, has a wide range of applications and low cost. At present, the studies of this kind of method mainly focus on bare soil minerals and seldom on widespread vegetated areas. The important reason is that the biophysical process of hydrocarbon microleakage toxicity to vegetation roots is complex, and the spectral characteristics that can be used to extract vegetation anomalies are vague. Moreover, the spectral features selected according to a small number of sampled spectra are accidental, leading to the low accuracy of the extraction results. Therefore, this work first discussed the mechanism of hydrocarbon microleakage poisoning to vegetation roots. Afterward, the vegetation spectral features that effectively reflect the effect of hydrocarbon microleakage were selected based on the PROSAIL model. Moreover, a red-edge position index based on Sentinel-2/MSI band configuration was proposed. Then, we marked the mine sites across our study area, the Qinshui basin, on Google Earth for long-term vegetation spectral characteristics statistics. We compared these mine sites with those of the control area to determine how these spectral features were affected by hydrocarbon microleakage. Finally, the marked samples were divided into training and test sets and then verified. These sets were used to find the optimal spectral feature threshold combination via the threshold space method. The statistical results show that, compared with the control area, the experimental area exhibited an obvious blue shift revealed by the red-edge position index of the mine samples. Moreover, the near-infrared reflectance decreased, and the red valley reflectance increased. These findings were consistent with the mechanism of hydrocarbon microleakage poisoning vegetation and the results of the spectral simulation. In the background mountain forest area, the 80% recall rate of vegetation samples in the mine buffer zone could be balanced with the 5% misclassification rate of vegetation samples, showing the rationality of this method. In this study, we analyzed and optimized the spectral characteristics of hydrocarbon microleakage affecting vegetation. We also used multispectral data to construct a spectral index and extract the hydrocarbon microleakage vegetation anomaly according to the spectral statistics of the mine buffer. This method combines theoretical simulation with large sample statistics, providing a reference for the research of extracting hydrocarbon microleakage vegetation anomaly by remote sensing.
摘要：Studies showed that earthquakes can cause anomalies of methane gas in the atmosphere. The present study selected a certain area in Sichuan-Yunnan Province, and the Luxian earthquake in Sichuan Province in September 2021 was taken as an example. Based on methane gas products obtained by the Atmospheric Infrared Sounder, a hyperspectral sensor was carried by the US Earth observation satellite AQUA/Earth Observation System(AQUA/EOS). The mature Robust Satellite Technique(RST) algorithm was used to extract methane anomaly information before and after the earthquake. The background field characteristics of methane in the study area were analyzed. Moreover, the methane anomaly index time series analysis was performed for earthquakes with a magnitude of 6 or above in the region since 2008. Results showed that the methane background field in the study area is related to topography and geomorphology and has obvious spatial and temporal distribution characteristics. In particular, the difference in biodiversity and temperature is caused by seasonal changes in time, and the methane concentration regularly changes. The spatial variation is mainly related to topography and geomorphology. The seismic-related characteristic is high methane concentration at the fault and plate tectonic junction. The methane anomaly has a certain correspondence with the earthquake. This finding mainly shows that the distribution law of the temporal and spatial characteristics of methane in the regional history is broken. Moreover, the overall changes in the earthquake preparation process show certain characteristics: initial intensification, abnormal intensification, peak attenuation, and calm. The magnitude of the anomaly has no obvious relationship with the earthquake’s magnitude. However, the duration of the anomaly may be related to the magnitude of the earthquake; that is, the methane anomaly caused by the earthquake is not accidental and has a certain duration of the anomaly. In particular, the amplitude of the anomaly is greater than 2, the duration of the anomaly is at least 1 month, the amplitude of the anomaly is greater than 1, and the duration is more than 3 months. The anomaly may correspond to earthquakes in a certain region. Further research should be conducted through the comprehensive analysis of the structural and geological conditions, earthquake magnitude, and comparison of different research area radii in the following years. Conducting seismic anomaly monitoring of methane gas is feasible based on remote sensing in the local Sichuan-Yunnan region. This finding is related to the fact that the region itself is rich in hydrocarbon gas. The occurrence of earthquakes prompts the rapid migration and diffusion of underground massive hydrocarbon gas into the atmosphere along the weak zone, such as rock fissures, fault zones, and unconformity surfaces. This study preliminarily shows that the study of seismic methane anomaly in this local area is suitable for earthquakes with a similar tectonic background within the scope. Given the limitations of the study area, the follow-up study will fully consider the differences in tectonic backgrounds, combined with the distribution of oil and gas reservoirs, to conduct in-depth research outside the local area and further explore the feasibility of seismic methane anomaly monitoring.
关键词：InSAR;point cloud processing;data visualization;WebGL;deformation data