Search

Coastal wetlands are important areas with valuable natural resources and diverse biodiversity. Due to the influence of both natural factors and human activities, the landscape of coastal wetlands undergoes significant changes. It is crucial to systematically monitor and analyze the dynamic changes in coastal wetland cover over a long-term time series. In this paper, a long-term time series coastal wetland remote sensing classification process was proposed, which integrated feature selection and sample migration. Utilizing Google Earth Engine (GEE) and Landsat TM/ETM/OLI remote sensing image data, the selected feature set is combined with the sample migration method to generate the training sample set for each target year. The Simple Non-Iterative Clustering-Random Forest (SNIC-RF) model was ultimately employed to accurately map wetland classes in the Liaohe Estuary from 1985 to 2023 and quantitatively evaluate the spatio-temporal pattern change characteristics of wetlands in the study area. The findings indicate that: (1) After feature selection, the accuracy of the model reached 0.88, and the separation of the selected feature set was good. (2) After sample migration, the overall accuracy of sample classification in the target year ranged from 87 to 94%, along with Kappa coefficients of 0.84 to 0.92, thereby ensuring the validity of classification sample migration. (3) SNIC-RF classification results showed better performance of wetland landscape. Compared with RF classification, the overall classification accuracy was increased by 0.69-5.82%, and the Kappa coefficient was increased by 0.0087-0.0751. (4) From 1985 to 2023, there has been a predominant trend of natural wetlands being converted into artificial wetlands. In recent years, this transition has occurred more gently. Finally, this study offers valuable insights into understanding changes and trends in the surface ecological environment of the Liaohe Estuary. The research method can be extended to other types of wetland classification and the comprehensive application of coastal wetland in hydrology, ecology, meteorology, soil, and environment can be further explored on the basis of this research, laying strong groundwork for shaping policies on ecological protection and restoration. [ABSTRACT FROM AUTHOR]

Reliable and up-to-date training reference samples are imperative for land cover (LC) classification. However, such training datasets are not always available in practice. The sample migration method has shown remarkable success in addressing this challenge in recent years. This work investigated the application of Sentinel-1 (S1) and Sentinel-2 (S2) data in training sample migration. In addition, the impact of various spectral bands and polarizations on the accuracy of the migrated training samples was also assessed. Subsequently, combined S1 and S2 images were classified using the Support Vector Machines (SVM) and Random Forest (RF) classifiers to produce annual LC maps from 2017 to 2021. The results showed a higher accuracy (98.25%) in training sample migrations using both images in comparison to using S1 (87.68%) and S2 (96.82%) data independently. Among the LC classes, the highest accuracy in migrated training samples was found for water, built-up, bare land, grassland, cropland, and wetland. Inquiries on the efficiency of different spectral bands and polarization used in training sample migration showed that bands 4 and 8 and VV polarization in the water class were more important, while for the wetland class, bands 5, 6, 7, 8, and 8A together with VV polarization showed superior performance. The results showed that the RF classifier provided better performance than the SVM (higher overall, producer, and user accuracy). Overall, our findings suggested that shared use of S1 and S2 data can be used as a suitable means for producing up-to-date and high-quality training samples. [ABSTRACT FROM AUTHOR]

The composition and dynamics of wetland plant communities play a critical role in maintaining the functionality of wetland ecosystems and serve as important indicators of wetland degradation and restoration. Accurately identifying wetland plant communities using remote sensing techniques remains challenging due to the complex environment and cloud contamination. Here, we applied a sample migration method based on change vector analysis and a random forest (RF) classifier incorporating SHapley Additive exPlanations (SHAP) to explore the spatiotemporal changes of wetland plant communities in the western Songnen Plain of China between 2016 and 2022, and to better understand the decision logic of the RF model. Our work achieved accurate annual wetland classification at the community scale, with an average overall accuracy of 89.5% and an average kappa coefficient of 0.87. Our analysis revealed different spatiotemporal change characteristics of wetland plant communities in the western Songnen Plain and three national nature reserves. The SHAP model showed that MOS_IRECI is the most important feature determining the prediction results of the RF model, and the importance of the features differs at global and local levels. This study confirms the feasibility of annual dynamic monitoring of wetland plant communities at a regional scale. The results are expected to provide a reference for the fine and sustainable management of wetland resources in the western Songnen Plain, as well as valuable data support for related wetland ecology research.

Myanmar has experienced rapid socio-economic developments in recent decades, which have a greater impact on land cover change. Accurate long time series land cover datasets for Myanmar can be of great help in environmental protection and natural resource management. However, there are relatively few existing studies on long time series land cover datasets in Myanmar, and the acquisition of training samples within different time series is a big challenge. Therefore, this study used Google Earth Engine and Landsat imagery to produce a land cover dataset for every two years from 1990 to 2020 using a training sample migration framework. First, the differences in index change, spectral value change, and spectral shape change were used to determine whether the sample points had changed between the base year and the previous year, and then a small number of samples were manually selected. Second, the spectral features, index information, and texture information of the remote sensing images and the object-oriented segmentation method were used to obtain object-oriented multidimensional features. Finally, the random forest method was employed to train the samples of the previous year to obtain the land cover data of the previous year. The results of the study show that the average overall precision of the land cover classification results for Myanmar for 1990–2020 is 0.83 and Kappa is 0.79. In addition, the land cover classification results for Myanmar of 1990–2020 are significantly better than those of Globeland30-2020, FROM-GLC, and Dynamic World land cover, and comparing with these products showed good agreement.

Wetlands provide vital ecological services for both humans and environment, necessitating continuous, refined and up-to-date mapping of wetlands for conservation and management. In this study, we developed an automated and refined wetland mapping framework integrating training sample migration method, supervised machine learning and knowledge-driven rules using Google Earth Engine (GEE) platform and open-source geospatial tools. We applied the framework to temporally dense Sentinel-1/2 imagery to produce annual refined wetland maps of the Dongting Lake Wetland (DLW) during 2015–2021. First, the continuous change detection (CCD) algorithm was utilized to migrate stable training samples. Then, annual 10 m preliminary land cover maps with 9 classes were produced using random forest algorithm and migrated samples. Ultimately, annual 10 m refined wetland maps were generated based on preliminary land cover maps via knowledge-driven rules from geometric features and available water-related inventories, with Overall Accuracy (OA) ranging from 81.82% (2015) to 93.84% (2020) and Kappa Coefficient (KC) between 0.73 (2015) and 0.91 (2020), demonstrating satisfactory performance and substantial potential for accurate, timely and type-refined wetland mapping. Our methodological framework allows rapid and accurate monitoring of wetland dynamics and could provide valuable information and methodological support for monitoring, conservation and sustainable development of wetland ecosystem.

Long time series land cover classification information is the basis for scientific research on urban sprawls, vegetation change, and the carbon cycle. The rapid development of cloud computing platforms such as the Google Earth Engine (GEE) and access to multi-source satellite imagery from Landsat and Sentinel-2 enables the application of machine learning algorithms for image classification. Here, we used the random forest algorithm to quickly achieve a time series land cover classification at different scales based on the fixed land classification sample points selected from images acquired in 2022, and the year-by-year spectral differences of the sample points. The classification accuracy was enhanced by using multi-source remote sensing data, such as synthetic aperture radar (SAR) and digital elevation model (DEM) data. The results showed that: (i) the maximum difference (threshold) of the sample points without land class change, determined by counting the sample points of each band of the Landsat time series from 1986 to 2022, was 0.25; (ii) the kappa coefficient and observed accuracy of the same sensor from Landsat 8 are higher than the results of the TM and ETM+ sensor data from 2013 to 2022; and (iii) the addition of a mining land cover type increases the kappa coefficient and overall accuracy mean values of the Sentinel 2 image classification for a complex mining and forest area. Among the land classifications via multi-source remote sensing, the combined variables of Spectral band + Index + Terrain + SAR result in the highest accuracy, but the overall improvement is limited. The method proposed is applicable to remotely sensed images at different scales and the use of sensors under complex terrain conditions. The use of the GEE cloud computing platform enabled the rapid analysis of remotely sensed data to produce land cover maps with high accuracy and a long time series. [ABSTRACT FROM AUTHOR]

Reliable and up-to-date training reference samples are imperative for land cover (LC) classification. However, such training datasets are not always available in practice. The sample migration method has shown remarkable success in addressing this challenge in recent years. This work investigated the application of Sentinel-1 (S1) and Sentinel-2 (S2) data in training sample migration. In addition, the impact of various spectral bands and polarizations on the accuracy of the migrated training samples was also assessed. Subsequently, combined S1 and S2 images were classified using the Support Vector Machines (SVM) and Random Forest (RF) classifiers to produce annual LC maps from 2017 to 2021. The results showed a higher accuracy (98.25%) in training sample migrations using both images in comparison to using S1 (87.68%) and S2 (96.82%) data independently. Among the LC classes, the highest accuracy in migrated training samples was found for water, built-up, bare land, grassland, cropland, and wetland. Inquiries on the efficiency of different spectral bands and polarization used in training sample migration showed that bands 4 and 8 and VV polarization in the water class were more important, while for the wetland class, bands 5, 6, 7, 8, and 8A together with VV polarization showed superior performance. The results showed that the RF classifier provided better performance than the SVM (higher overall, producer, and user accuracy). Overall, our findings suggested that shared use of S1 and S2 data can be used as a suitable means for producing up-to-date and high-quality training samples.

Global land cover has undergone extensive and rapid changes as a result of human activities and climate change. These changes have had a significant impact on biodiversity, the surface energy balance, and sustainable development. Global land cover data underpins research on the development of earth system models, resource management, and evaluation of the ecological environment. However, there are limitations in the classification detail, spatial resolution, and rapid change monitoring capability of global land cover change data. Building on the earlier Global Land Cover Mapping (Finer Resolution Observation and Monitoring – Global Land Cover, FROM-GLC), we developed the improved Global Land Cover Change Monitoring Platform (FROM-GLC Plus) using methods such as multi-season sample space-time migration, multi-source data time series reconstruction, and machine learning. The FROM-GLC Plus system provides a capacity for producing global land cover change data set from the 1980s with flexibility in spatio–temporal details. The preliminary results show that FROM-GLC Plus provides a framework for near real-time land cover mapping at multi-temporal (annual to daily) and multi-resolution (30 m to sub-meter) levels.

Long time series land cover classification information is the basis for scientific research on urban sprawls, vegetation change, and the carbon cycle. The rapid development of cloud computing platforms such as the Google Earth Engine (GEE) and access to multi-source satellite imagery from Landsat and Sentinel-2 enables the application of machine learning algorithms for image classification. Here, we used the random forest algorithm to quickly achieve a time series land cover classification at different scales based on the fixed land classification sample points selected from images acquired in 2022, and the year-by-year spectral differences of the sample points. The classification accuracy was enhanced by using multi-source remote sensing data, such as synthetic aperture radar (SAR) and digital elevation model (DEM) data. The results showed that: (i) the maximum difference (threshold) of the sample points without land class change, determined by counting the sample points of each band of the Landsat time series from 1986 to 2022, was 0.25; (ii) the kappa coefficient and observed accuracy of the same sensor from Landsat 8 are higher than the results of the TM and ETM+ sensor data from 2013 to 2022; and (iii) the addition of a mining land cover type increases the kappa coefficient and overall accuracy mean values of the Sentinel 2 image classification for a complex mining and forest area. Among the land classifications via multi-source remote sensing, the combined variables of Spectral band + Index + Terrain + SAR result in the highest accuracy, but the overall improvement is limited. The method proposed is applicable to remotely sensed images at different scales and the use of sensors under complex terrain conditions. The use of the GEE cloud computing platform enabled the rapid analysis of remotely sensed data to produce land cover maps with high accuracy and a long time series.

The spatial and temporal variation analysis of land cover classification is important for studying the distribution and transformation of regional land cover changes. The Qilian Mountain National Park (QMNP), an important ecological barrier in northwestern China, has lacked land cover products for long time series. The Landsat images available on the Google Earth Engine (GEE) make it possible to analyze the land cover changes over the past three decades. The purpose of this study was to generate a long time series of datasets of land cover classification based on the method of sample migration in the QMNP in Northwest China. The Landsat 5, 7, and 8 images and field sample data were combined with multiple image features and the random forest algorithm to complete the land cover classification of the QMNP from 1990 to 2020. The results indicate that (1) the method of Jeffries–Matusita (J-M) distance can reduce image feature redundancy and show that elevation and phenological features have good differentiability among land cover types that were easy to mix with feature classes; (2) the spatial distribution of land cover every 10 years between 1990 and 2020 was consistent in the QMNP, and there were obvious differences in land cover from the east to the west part of the QMNP, with a large area of vegetation distribution in Sunan county in the central part and Tianzhu county in the east part of the QMNP; (3) over the past 30 years, forests and grasslands decreased by 62.2 km2 and 794.7 km2, respectively, while shrubs increased by 442.9 km2 in the QMNP. The conversion of bare land to grassland and the interconversion between different vegetation types were the main patterns of land cover changes, and the land cover changes were mainly concentrated in pastoral areas, meaning that human activity was the main factor of land cover changes; and (4) when the samples of 2020 were migrated to 2010, 2000, and 1990, the overall classification accuracies were 89.7%, 88.0%, 86.0%, and 83.9%, respectively. The results show that the vegetation conservation process in the QMNP was closely related to human activities. [ABSTRACT FROM AUTHOR]

• High-resolution annual cropland maps were made for the Huang-Huai-Hai Plain. • A prior information was used to guide image composition showing superior performance. • A new method was proposed for sample migration cross years. • Mapping with Random Forest and Positive and Unlabeled Learning methods combination. Numerous map products containing cropland data are currently available. However, due to their low temporal and spatial resolutions and limited accuracy, their practical applications are limited. Thus, research on the precise mapping of croplands with high temporal and spatial resolutions is still urgently needed. The Huang-Huai-Hai Plain, a critical grain production area in China, was used as the research area to produce high-resolution annual cropland maps from 2018 to 2022. Sentinel-2 imagery, digital elevation model (DEM) data, and China Agricultural Ecological Zoning (CAEZ) data were collected. Based on these data, the study area was stratified into different zones, and then the distinct characteristics and optimal timing for differentiating cropland from other land cover types within each zone were analyzed and used as prior information for guiding time-series image compositing. Furthermore, the spectral information divergence (SID) index was used to modify the existing sample migration technique to transfer training samples across years. Via the Google Earth Engine (GEE) platform, the random forest (RF) classification method and positive and unlabeled learning (PUL) method were combined to construct cropland maps from composited images and terrain information from DEMs. Compared with the seasonal or spectral-temporal metric (STM) image composition methods, the image composition method based on prior information significantly improved the distinction between cropland and other land cover types. Moreover, the newly proposed sample migration strategy substantially increased the accuracy of training samples transferred across years. Compared with existing cropland maps from products such as the ESA, ESRI, GlobeLand30, GLAD, and CLCD, the cropland maps generated in this study demonstrated superior accuracy, with average overall accuracy (OA) values from 88.81% to 91.66% for different zones, thus outperforming other products with OA values from 77.53% to 88.90%. This study supports the development of related agricultural and ecological research and provides a valuable reference for cropland mapping. [ABSTRACT FROM AUTHOR]

The spatial and temporal variation analysis of land cover classification is important for studying the distribution and transformation of regional land cover changes. The Qilian Mountain National Park (QMNP), an important ecological barrier in northwestern China, has lacked land cover products for long time series. The Landsat images available on the Google Earth Engine (GEE) make it possible to analyze the land cover changes over the past three decades. The purpose of this study was to generate a long time series of datasets of land cover classification based on the method of sample migration in the QMNP in Northwest China. The Landsat 5, 7, and 8 images and field sample data were combined with multiple image features and the random forest algorithm to complete the land cover classification of the QMNP from 1990 to 2020. The results indicate that (1) the method of Jeffries–Matusita (J-M) distance can reduce image feature redundancy and show that elevation and phenological features have good differentiability among land cover types that were easy to mix with feature classes; (2) the spatial distribution of land cover every 10 years between 1990 and 2020 was consistent in the QMNP, and there were obvious differences in land cover from the east to the west part of the QMNP, with a large area of vegetation distribution in Sunan county in the central part and Tianzhu county in the east part of the QMNP; (3) over the past 30 years, forests and grasslands decreased by 62.2 km2 and 794.7 km2, respectively, while shrubs increased by 442.9 km2 in the QMNP. The conversion of bare land to grassland and the interconversion between different vegetation types were the main patterns of land cover changes, and the land cover changes were mainly concentrated in pastoral areas, meaning that human activity was the main factor of land cover changes; and (4) when the samples of 2020 were migrated to 2010, 2000, and 1990, the overall classification accuracies were 89.7%, 88.0%, 86.0%, and 83.9%, respectively. The results show that the vegetation conservation process in the QMNP was closely related to human activities.

Irrigated agricultural expansion is one of the main reasons for water scarcity in the Lake Urmia basin. Although previous studies have analyzed the impact of cropland expansion on the Lake Urmia Shrinkage, there is a lack of comprehensive annual assessment of historical irrigation expansion in the Lake Urmia basin and its impact on water resources of this region. In this study, we developed an automatic and efficient workflow using Landsat and Gravity Recovery and Climate Experiment (GRACE) data, GRACE Follow-On (GRACE-FO) data, and a sample migration technique within the Google Earth Engine cloud computing platform to comprehensively investigate the impact of irrigated agricultural expansion on the shrinkage of Lake Urmia, as one of the most severe environmental crisis in the world. Additionally, using the global surface water data, we proposed a fully automatic procedure to obtain reference samples from water bodies. The Lake Urmia basin was first classified into the water, irrigated, and Non-Water/Irrigated classes using the random forest algorithm. The average overall accuracy of the produced annual land cover maps during 1987–2020 was 92.2%, representing the great potential of the developed method for land cover mapping. We found that the irrigated lands expanded by nearly 890 km2 during the study period. Coincident with this change, although the area of water bodies in Lake Urmia partially recovered after 2015 (reached from 1,050 km2 in 2015 to 3,370 km2 in 2020), it is currently far beyond its original condition (i.e., ∼5,400 km2, average record during 1987–2000). Moreover, the information of the Terrestrial Water Storage (TWS) from the GRACE and GRACE-FO data between 2003 and 2020 showed a dramatic decrease in TWS level (∼−11.5 cm). The findings of this research will assist the local stakeholders and authorities to better understanding the environmental costs of irrigation expansion in the Lake Urmia basin.

We investigate using nutation experiments the spatial distribution of radiofrequency (rf) field, sample, temperature and cross-polarization transfer efficiency in 1.3 mm rotor. First, two-dimensional (2D) 1 H nutation experiments on silicone thin cylinders in the presence of B 0 field gradient generated by shim coils are used to image the spatial distribution of rf field inside the rotor. These experiments show that the rf field is asymmetrical with respect to the center of the rotor. Moreover, they show the large inhomogeneity that still remains across the sample when using spacers, and that even in this case, the rf-field close to the drive cap is decreased to ca. only 20% of its maximum value. Such 2D nutation experiment in the presence of B 0 field gradient are also employed to demonstrate the migration of adamantane sample from the center of the rotor to its ends during Magic-Angle Spinning (MAS). Furthermore, 2D 1 H nutation experiments on nickelocene exhibiting temperature-dependent isotropic chemical shift provides insights into the temperature distribution inside rotor. Finally three-dimensional (3D) 1 H → 13 C Cross-Polarization under MAS (CPMAS) nutation experiment indicates that only nuclei subject to the largest rf field contribute to the CPMAS transfer, when using rf field of constant amplitude on both channels. Such high selectivity allows the determination of accurate dipolar coupling constants in the Cross-Polarization with Variable Contact (CP-VC) experiment under fast MAS, at the expense of low sensitivity. Conversely when using ramped-amplitude on the 1 H channel during the CPMAS transfer, nuclei subject to smaller rf field contributes to the transfer, which increases the sensitivity of CPMAS experiment but does not allow an accurate determination of dipolar coupling constants using CP-VC experiment. [ABSTRACT FROM AUTHOR]