Figure 1 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Source publication
To study global and regional environment protection and sustainable development and also to optimize mapping methods, it is of great significance to compare three existing 10 m resolution global land cover products in terms of accuracy: FROM-GLC10, the ESRI 2020 land cover product (ESRI2020), and the European Space Agency world cover 2020 product (...
Contexts in source publication
Context 1
... Southeast Asia, land cover products are of great significance for ecological protection, efficient use of land, and the rational planning of resources [10,37]. Thus, we took Southeast Asia as the research area (Figure 1). Through the comparative validation of three different land cover products, we wish to provide a reference of how to use the land cover products for studying the local human-land relationship, climate change, and ecosystem and environmental protection [38]. ...
Context 2
... Southeast Asia, land cover products are of great significance for ecological prote tion, efficient use of land, and the rational planning of resources [10,37]. Thus, we too Southeast Asia as the research area (Figure 1). Through the comparative validation three different land cover products, we wish to provide a reference of how to use the lan cover products for studying the local human-land relationship, climate change, and ec system and environmental protection [38]. ...
Context 3
... Spatial consistency analysis results Figure A1 presents the spatial analysis results of each class. The spatial consistency analysis results of each class show that the consistency results in the Indo-China Peninsula region, including Myanmar, Thailand, and Cambodia, were worse than those in the Malay Archipelago region. ...
Context 4
... made an assessment of the random errors when sampling using boxplots. Figure 10a presents a boxplot of the OA values of 100 samples for the three land cover products. According to the boxplot of the overall precision of the sampled data, these 100-sample precision estimators are almost unbiased and outliers exist, but the number does not exceed five. ...
Context 5
... made an assessment of the random errors when sampling using boxplots. Fig 10a presents a boxplot of the OA values of 100 samples for the three land cover produ According to the boxplot of the overall precision of the sampled data, these 100-sam precision estimators are almost unbiased and outliers exist, but the number does no ceed five. Therefore, on the basis of 100 sampling points, it is scientifically reasonable we chose the mean as the final accuracy validation result. ...
Context 6
... take OA as an example. The results are shown in Figure 11. The results show that the verification result of FROM_GLC10 under the proportion of 100% field collection points and 0% manual densification points is 59.1%. ...
Context 7
... ESRI2020 verification results changed from 75.2% to 80.8% in the process of changing the proportion of the field collection points from 100% to 0%. The ESA2020 verification results changed from 79.8% to 81.2% in the process of changing the proportion of the field collection points from 100% to 0%. Figure 11. The mean OA value of 100 samples for the different mixing ratios of the field collection points and the manual densification points. ...
Context 8
... ESA2020 has a separate class for moss and lichen forest, which most likely contains some grassland, so the PA and the UA of the grassland in ESA2020 are not high. We did not have a separate class of moss and lichen forest in the Figure 11. The mean OA value of 100 samples for the different mixing ratios of the field collection points and the manual densification points. ...
Context 9
... to the boxplot of OA (Figure 10a), compared with ESA2020 and ESRI2020, FROM_GLC10 is more affected by changes in the sample points. According to the boxplot of the sampling data for the individual classes, the errors of the different classes are quite different. ...
Context 10
... to the boxplot of the sampling data for the individual classes, the errors of the different classes are quite different. For the PA, it can be seen from Figure 10b, the grassland box is drawn longer than the boxes for the other classes in ESA2020; the built-up area box and the shrubland box are drawn longer than the boxes for other classes in ESRI2020; and the built-up area, grassland, and shrubland boxes are drawn longer than those for other classes in FROM_GLC10. This shows that these classes are greatly affected by the sampling. ...
Context 11
... the fuzzy definitions of the grassland and shrubland classes themselves cause classification errors. For the UA, it can be seen from Figure 10c that the sampling errors of bare land and wetland in ESRI2020 and ESA2020 are large. There were a few outliers for wetland in FROM_GLC10. ...
Context 12
... it is crucial to calculate the number of points required for each class according to the area ratio. Figure 11 reveals that the mixing ratio of the field collection points and the manual densification points can influence the validation results by as much as 19.5%. This effect is related to whether the sample points are evenly distributed, and it may also be related to the intensity of human interference in the distribution area of our field collection points. ...
Context 13
... validation criterion was that both the UA and the PA of a single class should be 50%. The results are presented in Figure 12. ...
Similar publications
High-accuracy land use and land cover maps (LULC) are increasingly in demand for environmental management and decision-making. Despite the limitation, Machine learning classifiers (MLC) fill the gap in any complex issue related to LULC data accuracy. Visualizing land-cover information is critical in mitigating Côte d’Ivoire’s deforestation and land...
Citations
... In general, across different datasets, tree cover, cropland, built-up, and water had relatively higher PA, UA, and F1 score values, while shrubland, grass land, bare/sparse vegetation, and herbaceous wetland need further improvement. The serious misclassification and omission of these four land cover classes is consistent with the conclusions reached by Ding, et al. [41] and Kang, et al. [42], who conducted the accuracy assessment of ESA 2020 in Southeast Asia and Northwestern China. The high label errors of ESA 2020 used for model training and testing can lead to an error accumulation especially for uneasily classified types. ...
Generating high-resolution land cover maps using relatively lower-resolution remote sensing images is of great importance for subtle analysis. However, the domain gap between real lower-resolution and synthetic images has not been permanently resolved. Furthermore, super-resolution information is not fully exploited in semantic segmentation models. By solving the aforementioned issues, a deeply fused super resolution guided semantic segmentation network using 30 m Landsat images is proposed. A large-scale dataset comprising 10 m Sentinel-2, 30 m Landsat-8 images, and 10 m European Space Agency (ESA) Land Cover Product is introduced, facilitating model training and evaluation across diverse real-world scenarios. The proposed Deeply Fused Super Resolution Guided Semantic Segmentation Network (DFSRSSN) combines a Super Resolution Module (SRResNet) and a Semantic Segmentation Module (CRFFNet). SRResNet enhances spatial resolution, while CRFFNet leverages super-resolution information for finer-grained land cover classification. Experimental results demonstrate the superior performance of the proposed method in five different testing datasets, achieving 68.17–83.29% and 39.55–75.92% for overall accuracy and kappa, respectively. When compared to ResUnet with up-sampling block, increases of 2.16–34.27% and 8.32–43.97% were observed for overall accuracy and kappa, respectively. Moreover, we proposed a relative drop rate of accuracy metrics to evaluate the transferability. The model exhibits improved spatial transferability, demonstrating its effectiveness in generating accurate land cover maps for different cities. Multi-temporal analysis reveals the potential of the proposed method for studying land cover and land use changes over time. In addition, a comparison of the state-of-the-art full semantic segmentation models indicates that spatial details are fully exploited and presented in semantic segmentation results by the proposed method.
... Empirical evidence has demonstrated that remote sensing is a more efficient and effective approach for obtaining accurate land cover information, compared to the time consuming and substantial investment required for field surveys [12,13]. Over the past several decades, satellite sensing technology and algorithms have rapidly developed, resulting in an increasing number of LC datasets with higher spatial resolution [14,15]. Originally, several LC maps were produced in a 1 km resolution, such as the IGBP DISCover (1992)(1993) [16], the UMD (1992-1993) [17], and the GLC(2000) [18]. ...
... LUCAS [15,[39][40][41].In particular, based on the CHINA2000 reference dataset, Song et al. identified obvious confusion of grass in China within GLOBCOVER, MODIS2005, MODIS2000, and GLC2000 products . Gao et al. evaluated the capability of three LC datasets by comparing them to the LUCAS reference dataset across the European Union, and showed that disagreement primarily occurred in regions characterized by heterogeneity [27]. ...
The Tibetan Plateau (TP) region contains maximal alpine grassland ecology at the mid-latitudes. This region is also recognized as an ecologically fragile and sensitive area under the effects of global warming. Regional climate modeling and ecosystem research depend on accurate land cover (LC) information. In order to obtain accurate LC information over the TP, the reliability and precision of five moderate/high-resolution LC products (MCD12Q1, C3S-LC, GlobeLand30, GLC_FCS30, and ESA2020 in 2020) were analyzed and evaluated in this study. The different LC products were compared with each other in terms of areal/spatial consistency and assessed with four reference sample datasets (Geo-Wiki, GLCVSS, GOFC-GOLD, and USGS) using the confusion matrix method for accuracy evaluation over the TP. Based on the paired comparison of these five LC datasets, all five LC products show that grass is the major land cover type on the TP, but the range of grass coverage identified by the different products varies noticeably, from 43.35% to 65.49%. The fully consistent spatial regions account for 43.72% of the entire region of the TP, while, in the transition area between grass and bare soil, there is still a large area of medium-to-low consistency. In addition, a comparison of LC datasets using integrated reference datasets shows that the overall accuracies of MCD12Q1, C3S-LC, GlobeLand30, GLC_FCS30, and ESA2020 are 54.29%, 49.32%, 53.03%, 53.73%, and 60.11%, respectively. The producer accuracy of the five products is highest for grass, while glaciers have the most reliable and accurate characteristics among all LC products for users. These findings provide valuable insights for the selection of rational and appropriate LC datasets for studying land-atmosphere interactions and promoting ecological preservation in the TP.
... For remote sensing picture classification, the most effective and practical validation tool is the confusion matrix method. 52 The vast majority of operations are consolidated into the error matrix, which use producer accuracy (PA), user accuracy (UA), and overall accuracy (OA) as indicators; this method is therefore successful. 53 A test of the classification results' correctness was performed to gauge the degree of precision of the usage map produced by the digital classification technique. ...
Mangroves maintain coastal balance and have the greatest potential for carbon sequestration. Most mapping studies on mangroves have focused on their extent and distribution and rarely featured mangrove species. Therefore, the objective of our study is to investigate mangrove species mapping from integrated Sentinel-2 imagery and field spectral data using a random forest (RF) algorithm. Study areas are located in East and South Lampung, Indonesia. The field samples used represented 144 points of mangrove species. The classification method used an RF algorithm and four models with varying parameters: model 1 with Sentinel-2; model 2 with both Sentinel-2 and field spectral data; model 3 with Sentinel-2, field spectral data, and spectrally transformed data; and model 4 only with spectrally transformed data. The results showed that Rhizophora mucronata, Sonneratia alba, Avicennia lanata, and Avicennia marina were the most common mangrove species in these areas, with reflectance values in the range of 0.002 to 0.493, 0.006 to 0.833, 0.014 to 0.768, and 0.002 to 0.758. Permutation importance (PI) that affects the classification model is the red band, near-infrared, and green normalized difference vegetation index, where the most PI in model 3 is 0.283. The highest level of agreement for mangrove species is found in model 3. Model 3 is the best parameter for RF classification that showed the best mapping accuracy, with the overall accuracy, user accuracy, producer accuracy, and kappa value being 81.25%, 81.68%, 81.25%, and 0.80, respectively.
... To facilitate the accuracy assessment, we obtained and processed the sample points from reference [23]. These sample points encompass a wide range of types, considering the quantity, richness, and spatial distribution of the samples, including field validation points and manual densification points from Google high-resolution images. ...
... This outcome can be attributed to the utilization of deep learning, scalable cloud-based computing, and a large-scale labeled sample dataset [14]. These findings align with those presented in the study [23]. Consequently, we recommend considering the inclusion of sample points during dataset production to enhance classification accuracy. ...
Currently, many globally accessible forest mapping products can be utilized to monitor and assess the status of and changes in forests. However, substantial disparities exist among these products due to variations in forest definitions, classification methods, and remote sensing data sources. This becomes particularly conspicuous in regions characterized by significant deforestation, like Southeast Asia, where forest mapping uncertainty is more pronounced, presenting users with challenges in selecting appropriate datasets across diverse regions. Moreover, this situation impedes the further enhancement of accuracy for forest mapping products. The aim of this research is to assess the consistency and accuracy of six recently produced forest mapping products in Southeast Asia. These products include three 10 m land cover products (Finer Resolution Observation and Monitoring Global LC (FROM-GLC10), ESA WorldCover 10 m 2020 (ESA2020), and ESRI 2020 Land Cover (ESRI2020)) and three forest thematic mapping products (Global PALSAR-2 Forest/Non-Forest map (JAXA FNF2020), global 30 m spatial distribution of forest cover in 2020 (GFC30_2020), and Generated_Hansen2020, which was synthesized based on Hansen TreeCover2010 (Hansen2010) and Hansen Global Forest Change (Hansen GFC) for the year 2020). Firstly, the research compared the area and spatial consistency. Next, accuracy was assessed using field validation points and manual densification points. Finally, the research analyzed the geographical environmental and biophysical factors influencing consistency. The results show that ESRI2020 had the highest overall accuracy for forest, followed by ESA2020, FROM-GLC10, and Generated_Hansen2020. Regions with elevations ranging from 200 to 3000 m and slopes below 15° or above 25° showed high spatial consistency, whereas other regions showed low consistency. Inconsistent regions showed complex landscapes heavily influenced by human activities; these regions are prone to being confused with shrubs and cropland and are also impacted by rubber and oil palm plantations, significantly affecting the accuracy of forest mapping. Based on the research findings, ESRI2020 is recommended for mountainous areas and abundant forest regions. However, in areas significantly affected by human activities, such as forest and non-forest edges and mixed areas of plantations and natural forests, caution should be taken with product selection. The research has identified areas of forest inconsistency that require attention in future forest mapping. To enhance our understanding of forest mapping and generate high-precision forest cover maps, it is recommended to incorporate multi-source data, subdivide forest types, and increase the number of sample points.
... Using procedures described in the relevant literature [33][34][35][36], several data inputs were selected as primary sources, which defined the area under analysis. An overview of this approach including a summary of the core files used in this study, the stated data types, the temporal coverage, the data source, and the data format is shown in Table 2. ...
Tracking changes in the structure of landscape dynamics as a result of flood activity is a complex process. This study presents a model for determining changes to landscapes caused by flood events by evaluating a specific territory in Eastern Slovakia, which has been affected by repeated large-scale flood events in the past. The area has not been subject to a comprehensive monitoring of changes in the landscape structure. Based on the observation of several sets of data, a combination of statistical methods and GIS spatial analysis tools (visualizing tools for compare categories, mapping, and modelling techniques, spatial analysis models for land use change and flood modelling) were used to identify changes in the landscape structure in the period from 1998 to 2021. The results point to the significance of the year 2010, with the precipitation totals for this year showing a level significantly higher than the rolling average and confirming the occurrence of an extreme flood event. The dynamics of landscape structure changes were evaluated based on changes in the representation of selected types of land cover classes. The results of a spatial evaluation of the Corine Land Cover demonstrate that the most-significant area changes were recorded in 2012 in the pasture class, with a decrease of 31% or approximately 96.5 ha. The identified difference in the frequency of representation of individual values of the normalized differential vegetation index confirms the loss of landscape diversity and the emergence of a more homogeneous type of landscape. An assessment of the state of pastures in the study area shows that this class has completely disappeared from the site near the watercourse.
... For example, ESRI has released the 2020 Land Cover product. It has good classification results for agricultural land, forests, water bodies, and building sites, but poor classification results for shrubs, wetlands, or fallow lands [27]. Additionally, Myanmar is relatively backward. ...
Land use/cover change (LUCC) research occupies an important place in the study of global change. It is important for the ecological protection and long-term development of a place. Current research is lacking in the study of dynamic changes at the national level in Myanmar over long time periods and sequences. Quantitative research on the driving factors of LUCC is also lacking. This paper uses the GLC_FCS30 (Global Land-Cover product with Fine Classification System) dataset and socio-economic statistical data in Myanmar to conduct the study. The dynamic change process of LUC (land use/cover) was investigated using the land use dynamic degree, land use transfer matrix, and Sankey diagram. Principal component analysis was used to derive the main drivers of LUCC. The drivers were quantified using multiple linear stepwise regression analysis and specific factors were analyzed. The spatial scope of the study is Myanmar, and the temporal scope is 2000–2020. Results: (1) In 2020, the spatial distribution of LUC in Myanmar shows predominantly forests and croplands. Forests account for 56.64% of the country’s total area. Agricultural land accounts for 25.59% of the country’s total area. (2) Over the time scale of the study, the trend of LUCC in Myanmar showed significant shrinkage of evergreen broad-leaved forest and deciduous broad-leaved forest (a total shrinkage of −3.34 × 104 km2) and expansion of the other land types. (3) Over the time scale of the study, the dynamic changes in LUCC in Myanmar most occurred as an interconversion between two land types, such as between cropland and deciduous broad-leaved forest, evergreen broad-leaved forest and shrubland, deciduous broad-leaved forest and shrubland, evergreen broad-leaved forest and evergreen needle-leaved forest, and evergreen broad-leaved forest and deciduous broad-leaved forest. (4) The dynamics of LUC in Myanmar is mainly influenced by the socio-economic level of the country. Among them, the impact of agricultural level is the most obvious. Specifically, Myanmar’s LUCC is mainly driven by urban population, urbanization rate, industrial value added, food production, and total population. Our research will enable the Myanmar government to make more scientific and rational land management and planning and to make more informed decisions. After understanding the basic situation of LUCC in Myanmar, the hydrological effects, biodiversity changes, and ecological service function changes due to land change in the region can be explored. This is the direction of future research.
... used a landcover map which is a composite of land use/ land cover (LULC) predictions for 10 classes throughout the year to generate a representative snapshot of 2020. It is derived by Esri (Environmental Systems Research Institute) from ESA Sentinel-2 imagery at 10m resolution (Karra et al. 2021) with a recognised global high accuracy (Ding et al. 2022;Mirmazloumi et al. 2022). The assessment of the pixel-based accumulation of NDVI (Normalized Difference Vegetation Index) anomalies over the last nine years (2013-2021) used the 'MYD13A1.006 ...
Water conflicts arise from geostrategic factors that hide behind visible ones such as the
construction of dams. In the case of the Grand Ethiopian Renaissance Dam (GERD), it
will have a significant impact on development in Ethiopia, but its filling has worried
downstream countries about altering the flow of the Nile and associated ecosystems. In
this study, several high spatiotemporal resolution remote sensing products were used
on the basis of artificial intelligence in Google Earth Engine. The results show that the
two first filling phases had no effect on the reservoirs of the dams in Sudan and Egypt,
nor on the vegetation cover. On the contrary, significant reserves of water have been
stored in response to unusual floods in the White Nile, and the recent trend of
increasing vegetation cover has not been affected likely due to groundwater
contributions and judicious anticipation, and the provision of resources for food
security. In general, the parties must take long-term collaborative measures to ensure
effective management and reduce waste, especially in the upcoming filling.
Land-cover mapping plays a crucial role in resource detection, ecological environmental protection, and sustainable development planning. The existing large-scale land-cover products with coarse spatial resolution have a wide range of categories, but they suffer from low mapping accuracy. Conversely, land-cover products with fine spatial resolution tend to lack diversity in the types of land cover they encompass. Currently, there is a lack of large-scale land-cover products simultaneously possessing fine-grained classifications and high accuracy. Therefore, we propose a mapping framework for fine-grained land-cover classification. Firstly, we propose an iterative method for developing fine-grained classification systems, establishing a classification system suitable for Sentinel-2 data based on the target area. This system comprises 23 fine-grained land-cover types and achieves the most stable mapping results. Secondly, to address the challenges in large-scale scenes, such as varying scales of target features, imbalanced sample quantities, and the weak connectivity of slender features, we propose an improved network based on Swin-UNet. This network incorporates a pyramid pooling module and a weighted combination loss function based on class balance. Additionally, we independently trained models for roads and water. Guided by the natural spatial relationships, we used a voting algorithm to integrate predictions from these independent models with the full classification model. Based on this framework, we created the 2017 Beijing–Tianjin–Hebei regional fine-grained land-cover product JJJLC-10. Through validation using 4254 sample datasets, the results indicate that JJJLC-10 achieves an overall accuracy of 80.3% in the I-level validation system (covering seven land-cover types) and 72.2% in the II-level validation system (covering 23 land-cover types), with kappa coefficients of 0.7602 and 0.706, respectively. In comparison with widely used land-cover products, JJJLC-10 excels in accurately depicting the spatial distribution of various land-cover types and exhibits significant advantages in terms of classification quantity and accuracy.
The present review examines existing technologies for mapping wetland ecosystems (WE) based on remote sensing data. WEs are valuable ecosystems with significant conservation importance. There are numerous classifications of WEs, encompassing dozens of types. However, existing global or national-level WE maps typically do not consider their landscape specificity and are limited to only a few classes. Peatlands, swamp forests, high-productivity meadows, and riparian vegetation formations store substantial carbon reserves, which are released due to fires. The intensity of these fires has been increasing in recent years as a result of climate change. These issues necessitate the development of new large-scale monitoring methods for assessing the state of WEs, including mapping their types, determining biomass and carbon stocks, assessing the consequences of landscape fires, and evaluating greenhouse gas emissions and other combustion byproducts. First and foremost, it is necessary to develop a classification system for Russia’s WEs that considers their landscape diversity while being sufficiently generalized for satellite monitoring and annual updating of WE maps. Various remote sensing data types, including aerial imagery, lidar, and radar data, are used for WE mapping. The most promising direction for the development of WE monitoring technologies at the national level involves the use of long-term homogeneous time series data from MODIS (Moderate Resolution Imaging Spectroradiometer) and VIIRS (Visible Infrared Imaging Radiometer Suite), in combination with ground-based calibration measurements and high-resolution satellite optical and radar data.
