Search

Remote sensing and land resource surveys have been used in recent decades for land use/land cover (LULC) mapping; however, keeping the developed LULC up-to-date and consistent with land survey statistics remains challenging. This study developed a practical and effective framework to automatically update existing LULC products and bridge the gap between remote sensing classification results and land survey data. This study employed Landsat imagery time series, change detection algorithms, sample migration, and random forests to develop a framework for updating existing LULC products in China from 1980–2015 to 1980–2022. The updated LULC maps reflect the post-2015 LULC changes well and maintain continuity with the pre-2015 products. Additionally, a statistical space allocation method based on the minimum cross-entropy strategy was proposed to optimize the LULC maps, increasing the correlation coefficient (r) with China's second and third national land survey statistics from 0.41–0.89 to 0.86–0.99. Thus, the framework and products developed in this study provide valuable tools for sustainable land use and policy planning. [ABSTRACT FROM AUTHOR]

Accurate and timely mapping of essential urban land use categories (EULUC) is vital to understanding urban land use distribution, pattern, and composition. Recent advances in leveraging big open data and machine learning algorithms have demonstrated the possibility of large-scale mapping of EULUC in a new cost-effective way. However, they are still limited by the transferability of samples, models, and classification results across space, particularly across different cities. Given the heterogeneities of environmental and socioeconomic conditions among cities, in-depth studies of data and model adaptation towards city-specific EULUC mappings are highly required to support policy making, and urban renewal planning and management practices. In addition, the trending need for timely and detailed small land unit data processing with finer data granularity becomes increasingly important. We proposed a City Meta Unit (CMU) data model and classification framework driven by multisource data and artificial intelligence (AI) algorithms to address these challenges. The CMU Framework was innovatively applied to systematically set up a grid-based data model and classify urban land use with an improved AI algorithm by applying Moore neighborhood correlations. Specifically, we selected Xiamen, Fujian, in China, a coastal city, as the typical testbed to implement this proposed framework and apply an AI transfer learning technique for grid and parcel land-use study. Experimental results with our proposed CMU framework showed that the grid-based land use classification performance achieves overall accuracies of 81.17% and 76.55% for level I (major classes) and level II (minor classes), which is much higher than the parcel-based land use classification (overall accuracies of 72.37% for level I, and 68.99% for level II). We further investigated the relationship between training sample size and classification performance and quantified the contribution of different data sources to urban land use classifications. The CMU framework makes data collections and processing intelligent and efficient, with finer granularity, saving time and cost by using existing open social data. Incorporating the CMU framework with the proposed grid-based model is an effective and new approach for urban land use classification, which can be flexibly extended and applied to various cities. [ABSTRACT FROM AUTHOR]

Cropland greatly impacts food security, energy supply, biodiversity, biogeochemical cycling, and climate change. Accurately and systematically understanding the effects of agricultural activities requires cropland spatial information with high resolution and a long time span. In this study, the first 1 km resolution global cropland proportion dataset for 10000 BCE-2100 CE was produced. With the cropland map initialized in 2010 CE, we first harmonized the cropland demands extracted from the History Database of the Global Environment 3.2 (HYDE 3.2) and the Land-Use Harmonization 2 (LUH2) datasets, and then spatially allocated the demands based on the combination of cropland suitability, kernel density, and other constraints. According to our maps, cropland originated from several independent centers and gradually spread to other regions, influenced by some important historical events. The spatial patterns of future cropland change differ in various scenarios due to the different socioeconomic pathways and mitigation levels. The global cropland area generally shows an increasing trend over the past years, from 0 million km2 in 10000 BCE grows to 2.8 million km2 in 1500 CE, 6.2 million km2 in 1850 CE, and 16.4 million km2 in 2010 CE. It then follows diverse trajectories under future scenarios, with the growth rate ranging from 18.6 % to 82.4 % between 2010 CE and 2100 CE. There are large area disparities among different geographical regions. The mapping result coincides well with widely-used datasets at present in both distribution pattern and total amount. With improved spatial resolution, our maps can better capture the cropland distribution details and spatial heterogeneity. The spatiotemporally continuous and conceptually consistent global cropland dataset serves as a more comprehensive alternative for long-term earth system simulations and other precise analyses. The flexible and efficient harmonization and downscaling framework can be applied to specific regions or extended to other land use/cover types through the adjustable parameters and open model structure. The 1 km global cropland maps are available at https://doi.org/10.5281/zenodo.5105689 (Cao et al., 2021a). [ABSTRACT FROM AUTHOR]

Annual land use land cover (LULC) change information at medium spatial resolution (i.e. at 30 m) is required in numerous subjects, such as biophysical modelling, land management and global change studies. Annual LULC information, however, is usually not available at continental or national scale due to reasons such as insufficient remote sensing data coverage or lack of computational capabilities. Here we integrate high temporal resolution and coarse spatial resolution satellite images (i.e., Moderate Resolution Imaging Spectroradiometer (MODIS) and Global Inventory Modelling and Mapping Studies (GIMMS) normalized difference vegetation index (NDVI)) with high spatial resolution datasets (China's Land-Use/cover Datasets (CLUDs) derived from 30-meter Landsat TM/ETM+/OLI) to generate reliable annual nominal 30 m LULC maps for the whole of China between 1980 and 2015. We also test the performance of a statistical based change detection algorithm (Breaks for Additive Seasonal and Trend), originally designed for tracking forest change, in classifying all-type LULC change. As a result, a nominal 30 m annual land use/land cover datasets (CLUD-A) from 1980 to 2015 was developed for the whole China. The mapping results were assessed with a change sample dataset, a regional annual validation sample set and a three-year China sample set. Of the detected change years, 75.61% matched the exact time of conversion within ±1 year. Annual mapping results provided a detail process of urbanization, deforestation, afforestation, water and cropland dynamics over the past 36 years. The consistent characterization of land change dynamics for China can be further used in scientific research and to support land management for policy-makers. [ABSTRACT FROM AUTHOR]

Accurate and up-to-date land use land cover (LULC) mapping has long been a challenge in Africa. Recently, three LULC maps with moderate spatial resolution (20 m to 100 m) have been developed using multiple Earth observation datasets for 2015–2016 for the whole continent, which provide unprecedented spatial detail of the land surface for Africa. This study aimed to compare these three recent African LULC maps (i.e. the Copernicus Global Land Service Land Cover (CGLS-LC100, 100 m), European Space Agency Sentinel-2A Land Cover (ESA-S2-LC20, 20 m) and Finer Resolution Observation and Monitoring of Global Land Cover for Africa version 2 (FROM-GLC-Africa30, 30 m)) using a validation sample set and statistics from the FAO. The results indicated that the accuracy of the three datasets was unevenly distributed in spatial extent and area estimation. All the three datasets achieve an accuracy of above 60% and the fraction layer of CGLS-LC100 showed the best consistency with FAO statistics in the area. However, great disagreement in spatial details was found among three products, with 43.12% of the total area in Africa was in low agreement. The LULC mapping regions with the highest uncertainty were southeast Africa, the Sahel region and the Eastern Africa Plateau. Uncertainty was most closely related to elevation and precipitation changes along latitude/longitude. [ABSTRACT FROM AUTHOR]

Accurate land-use/land-cover mapping based on remote-sensing images depends on clear and frequent observations. This study aimed to explore how many Landsat images were needed within a year and when they should be acquired, for cropland mapping in Africa. Three Landsat footprints in Egypt (Path/Row: 177/039, 127 images), Ethiopia (Path/Row: 168/054, 98 images), and South Africa (Path/Row: 170/078, 207 images) from 1984 to 2016 were used together with spectral indices and a 30-m digital elevation model in a random forest-based supervised classification. Detailed exploration was conducted into the number and temporal distribution of Landsat images required. Our results indicated that average cropland mapping accuracies for these three sites ranged from 81.17% to 87.59% (Egypt), 54.43% to 79.72% (Ethiopia), and 28.11% to 59.35% (South Africa) using different numbers of images within a year. The overall cropland accuracies were improved with an increase in available Landsat images within a year and reached a relatively stable stage when more than five images were acquired in all three sites. Growing season images played a key role in identifying cropland, accounting for a 13.22% average accuracy improvement compared with non-growing season images. Therefore, at least five images are recommended from a computational efficiency perspective, although fewer images, as low as two growing season images, can also achieve good results in specific regions. [ABSTRACT FROM AUTHOR]

A technically transparent and freely available reference sample set for validation of global land cover mapping was recently established to assess the accuracies of land cover maps with multiple resolutions. This sample set can be used to estimate areas because of its equal-area hexagon-based sampling design. The capabilities of these sample set-based area estimates for cropland were investigated in this paper. A 30-m cropland map for China was consolidated using three thematic maps (cropland, forest and wetland maps) to reduce confusion between cropland and forest/wetland. We compared three area estimation methods using the sample set and the 30 m cropland map. The methods investigated were: (1) pixel counting from a complete coverage map, (2) direct estimation from reference samples, and (3) model-assisted estimation combining the map with samples. Our results indicated that all three methods produced generally consistent estimates which agreed with cropland area measured from an independent national land use dataset. Areas estimated from the reference sample set were less biased by comparing with a National Land Use Dataset of China (NLUD-C). This study indicates that the reference sample set can be used as an alternative source to estimate areas over large regions. [ABSTRACT FROM AUTHOR]

Satellite based human settlement extraction at medium resolution (30 m) with supervised classification has been widely carried out. However, adequate training sample collection and mapping accuracy are two hindering factors over large regions. Here we propose a new framework for efficient human settlement extraction from Landsat images over large areas. First, an inventory-based training set is adopted to obtain some statistical parameters required to build a non-settlement mask. The mask can not only reduce unnecessary computation but also reduce the impact of background noise. Thereafter, for the un-masked areas we calculate the similarity of each image pixel to pre-collected sample points, and only those within certain threshold are treated as the settlement class. This approach is very fast and has been applied to three rapidly developing regions in China. Accuracy assessment indicates that the mean overall accuracies are 87%, 89% and 89% for Jing-Jin-Ji region, Yangtze River Delta and Pearl River Delta, respectively. This work may be applied to human settlement extraction at even broader spatial scales. [ABSTRACT FROM AUTHOR]

High-quality training and validation samples are critical components of land-cover and land-use mapping tasks in remote sensing. For large area mapping it is much more difficult to build such sample sets due to the huge amount of work involved in sample collection and image processing. As more and more satellite data become available, a new trend emerges in land-cover mapping that takes advantage of images acquired beyond the greenest season. This has created the need for constructing sample sets that can be used in classifying images of multiple seasons. On the other hand, seasonal land-cover information is also becoming a new demand in land and climate change studies. Here we produce the first training and validation data sets with seasonal labels in order to support the production of seasonal land-cover data for entire Africa. Nonetheless, for the first time, two classification systems were created for the same set of samples. We adapted the finer resolution observation and monitoring of global land cover (FROM-GLC) and the Food and Agriculture Organization (FAO) Land Cover Classification System legends. Locations of training-sample units of FROM-GLC were repurposed here. Then we designed a process to enlarge the training-sample units to increase the density of samples in the feature space of spectral characteristics of Moderate Resolution Imaging Spectroradiometer (MODIS) time-series and Landsat imagery. Finally, we obtained 15,799 training-sample units and 7430 validation-sample units. The land-cover type at each point was recorded at the time of maximum greenness in addition to four seasons in a year. Nearly half of the sample units were also suitable for 500 m resolution MODIS data. We analysed the representativeness of the training and validation sets and then provided some suggestions about their use in improving classification accuracies of Africa. [ABSTRACT FROM AUTHOR]

China has long suffered from failures in urban governance, arable land conservation, and environmental conservation because of the lack of accurate and reliable land use data. To fill these gaps, China has been conducting a once-a-decade census of national land use status. On August 26, 2021, China released the third national land survey data, which revealed several challenges that require further attention. In this paper, we first review the land resource surveys that have been and are being conducted in the world's major economically developed countries, and then we compare China's three national land surveys in terms of the data used, core survey technologies, the land use classification system, and main outcomes. Second, according to the major data results of the second national land survey and the third national land survey and other auxiliary data, using such methods as the data envelopment analysis and the land cover conversion matrix to highlight the existing land use issues such as shrinking arable land, inefficient construction land use, and low targeting efficiency of ecological restoration programs, as well as the drivers of these issues. Finally, we conclude the paper by discussing the next steps necessary to achieve the goal of sustainable land use in China, and the potential of satellite remote sensing technology and its derived land cover products to better support future national land surveys. • We review the land resource surveys that have been conducted in the world's major economically developed countries. • We compare China's three national land surveys since the 1980 s. • The land use issues were quantitatively highlighted using released land survey data, combined with other ancillary data. • The next steps necessary to achieve the goal of sustainable land use in China was discussed. • The potential of satellite remote sensing technology to better support future national land surveys was investigated. [ABSTRACT FROM AUTHOR]

A new 1 km global IIASA- IFPRI cropland percentage map for the baseline year 2005 has been developed which integrates a number of individual cropland maps at global to regional to national scales. The individual map products include existing global land cover maps such as GlobCover 2005 and MODIS v.5, regional maps such as AFRICOVER and national maps from mapping agencies and other organizations. The different products are ranked at the national level using crowdsourced data from Geo-Wiki to create a map that reflects the likelihood of cropland. Calibration with national and subnational crop statistics was then undertaken to distribute the cropland within each country and subnational unit. The new IIASA- IFPRI cropland product has been validated using very high-resolution satellite imagery via Geo-Wiki and has an overall accuracy of 82.4%. It has also been compared with the EarthStat cropland product and shows a lower root mean square error on an independent data set collected from Geo-Wiki. The first ever global field size map was produced at the same resolution as the IIASA- IFPRI cropland map based on interpolation of field size data collected via a Geo-Wiki crowdsourcing campaign. A validation exercise of the global field size map revealed satisfactory agreement with control data, particularly given the relatively modest size of the field size data set used to create the map. Both are critical inputs to global agricultural monitoring in the frame of GEOGLAM and will serve the global land modelling and integrated assessment community, in particular for improving land use models that require baseline cropland information. These products are freely available for downloading from the website. [ABSTRACT FROM AUTHOR]

Global land cover data could provide continuously updated cropland acreage and distribution information, which is essential to a wide range of applications over large geographical regions. Cropland area estimates were evaluated in the conterminous USA from four recent global land cover products: MODIS land cover (MODISLC) at 500-m resolution in 2010, GlobCover at 300-m resolution in 2009, FROM-GLC and FROM-GLC-agg at 30-m resolution based on Landsat imagery circa 2010 against the US Department of Agriculture survey data. Ratio estimators derived from the 30-m resolution Cropland Data Layer were applied to MODIS and GlobCover land cover products, which greatly improved the estimation accuracy of MODISLC by enhancing the correlation and decreasing mean deviation (MDev) and RMSE, but were less effective on GlobCover product. We found that, in the USA, the CDL adjusted MODISLC was more suitable for applications that concern about the aggregated county cropland acreage, while FROM-GLC-agg gave the least deviation from the survey at the state level. Correlation between land cover map estimates and survey estimates is significant, but stronger at the state level than at the county level. In regions where most mismatches happen at the county level, MODIS tends to underestimate, whereas MERIS and Landsat images incline to overestimate. Those uncertainties should be taken into consideration in relevant applications. Excluding interannual and seasonal effects,R2of the FROM-GLC regression model increased from 0.1 to 0.4, and the slope is much closer to one. Our analysis shows that images acquired in growing season are most suitable for Landsat-based cropland mapping in the conterminous USA. [ABSTRACT FROM AUTHOR]

Abstract: Data from three thermal sensors with different spatial resolution were assessed for urban surface temperature retrieval over the Yokohama City, Japan. The sensors are Thermal Airborne Broadband Imager (TABI), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and MODerate resolution Imaging Spectroradiometer (MODIS). Two algorithms were developed for land surface temperature (LST) retrieval from TABI image and ASTER thermal infrared (TIR) channels 13 and 14. In addition, ASTER LST and MODIS LST products were also collected. All the LST images were assessed by analyzing the relationship between LST and normalized difference vegetation index (NDVI) and by spatial distributions of LST profiles, derived from typical transects over the LST images. In this study, a strong negative relationship between LST and NDVI has been demonstrated although the degree of correlation between NDVI and LST varies slightly among the different LST images. Cross-validation among the LST images retrieved from the three thermal sensors of different spatial resolutions indicates that the LST images retrieved from the 2 channel ASTER data and a single band TABI thermal image using our developed algorithms are reliable. The LST images retrieved from the three sensors should have different potential to urban environmental studies. The MODIS thermal sensor can be used for the synoptic overview of an urban area and for studying urban thermal environment. The ASTER, with its TIR subsystem of 90-m resolution, allows for a more accurate determination of thermal patterns and properties of urban land use/land cover types, and hence, a more accurate determination of the LST. In consideration of the high heterogeneity of urban environment, the TABI thermal image, with a high spatial resolution of 2m, can be used for rendering and assessing complex urban thermal patterns and detailed distribution of LST at the individual house level more accurately. [Copyright &y& Elsevier]

The potential adverse impact of international trade on the environment has received growing attention in recent years. Growing environmental pressure poses significant challenges to sustainable development, further highlighting the need for a comprehensive response to tackling the unsustainable food use driven by the soybean trade. Although a significant amount of literature on trade-related land footprints already exists, globally, there has been only a limited amount of research seeking to identify the main issues related to agriculturally driven land-use change and food trade flows that have high land-use impacts. In addition, existing research does not fully reveal the ecological significance of land footprints but instead focuses mainly on physical quantities. There have been few studies that shed any light on the underlying correlations between environmental footprints and the food trade. To address these challenges, in this study, a multi-region input-output (MRIO) model was used to study agricultural land use and changes in carbon losses related to the soybean trade along global supply chains in 2013; the bilateral flows of land and economic value between countries were also modeled. The empirical results of this research indicate that the total land footprint embodied in the global soybean trade in 2013 amounted to 16.51 Mha. Globally, China used the most land resources for the soybean trade and accounted for 9.69 Mha of this footprint. The countries where the soybean trade had the greatest impact on the land and the economy were the USA and Brazil, where 6.74 Mha and 5.76 Mha of land were used for soybeans, respectively. Most of the soybeans from these countries were exported to China. The situation on the supply side was similar as China was also ranked in the top ten countries on the supply side; however, its supply-side land footprint was sufficient to meet internal demand. To further assess the environmental impact of the international soybean trade, carbon loss values (represented by the loss of carbon sequestration capacity) were estimated for the soybean trade at a country level. The total global carbon loss and the social cost of carbon due to the soybean trade in 2013 were estimated at $93.27 billion and $15.48 billion, respectively, with Brazil, the USA, and other countries in South America having the largest figures. It was found that, following a peak in the expansion of the amount of cropland used for planting soybeans, these ecological costs had declined since 2005. Based on these results, we suggest that soybean exporting countries should focus more on improving land-use efficiency and ecological protection in order to minimize the net land footprint of soybeans. • We applied MRIO model to trace agricultural land use and environmental footprints related to global soybean trade in 2013. • The total land footprint embodied in the global soybean trade amounted to 16.51 Mha. • The carbon loss values were estimated for the soybean trade process at a country level. • Global soybean trade's carbon loss and social cost of carbon in 2013 were $93.27 billion and $15.48 billion respectively. • We suggest export countries lay more focus on ecological protection to minimize the net land footprint of soybeans. [ABSTRACT FROM AUTHOR]

Artificial impervious areas are predominant indicators of human settlements. Timely, accurate, and frequent information on artificial impervious areas is critical to understanding the process of urbanization and land use/cover change, as well as of their impacts on the environment and biodiversity. Despite their importance, there still lack annual maps of high-resolution Global Artificial Impervious Areas (GAIA) with longer than 30-year records, due to the high demand of high performance computation and the lack of effective mapping algorithms. In this paper, we mapped annual GAIA from 1985 to 2018 using the full archive of 30-m resolution Landsat images on the Google Earth Engine platform. With ancillary datasets, including the nighttime light data and the Sentinel-1 Synthetic Aperture Radar data, we improved the performance of our previously developed algorithm in arid areas. We evaluated the GAIA data for 1985, 1990, 1995, 2000, 2005, 2010, and 2015, and the mean overall accuracy is higher than 90%. A cross-product comparison indicates the GAIA data are the only dataset spanning over 30 years. The temporal trend in GAIA agrees well with other datasets at the local, regional, and global scales. Our results indicate that the GAIA reached 797,076 km2 in 2018, which is 1.5 times more than that in 1990. China and the United States (US) rank among the top two in artificial impervious area, accounting for approximately 50% of the world's total in 2018. The artificial impervious area of China surpassed that of the US in 2015. By 2018, the remaining eight among the top ten countries are India, Russia, Brazil, France, Italy, Germany, Japan, and Canada. The GAIA dataset can be freely downloaded from http://data.ess.tsinghua.edu.cn. • We improved the performance of "Exclusion/Inclusion" approach in arid regions. • We mapped global artificial impervious areas (GAIA) with Google Earth Engine. • The mean overall accuracy over multiple years is higher than 90%. • GAIA reached 797,076 km2 by 2018, more than 2.5 times that of 1990. • The top five countries are China, US, India, Russia, and Brazil. [ABSTRACT FROM AUTHOR]

Mediterranean coastal lagoons and their peripheral areas often provide a collection of habitats for many species, and they often face significant threats from anthropogenic activities. Diverse human activities in such areas directly affect the spatio-temporal dynamic of surface water and its ecological characteristics. Monitoring the surface water dynamic, and understanding the impact of human activities are of great significance for coastal lagoon conservation. The Regional Natural Park of Narbonne includes a typical Mediterranean lagoon complex where surface water dynamic and its potential link with local diverse human activities has not yet been studied. In this context, based on all the available Landsat images covering the study area during 2002–2016, this study identified the water and non-water classes for each satellite observation by comparing three widely used spectral indices (i.e., NDVI, NDWI and MNDWI) and using the Otsu method. The yearly water frequency index was then computed to present the spatio-temporal dynamic of surface water for each year, and three water dynamic scenarios were also identified for each year: permanent water (PW), non-permanent water (NPW) and non-water (NW). The spatial and inter-annual variation in the patterns of the three water scenarios were characterized by computing the landscape metrics at scenario-level quantifying area/edge, shape, aggregation and fragmentation. Finally, the quantitative link between different land use and land cover (LULC) types derived from the LULC maps of 2003, 2012 and 2015 and the surface water dynamic scenarios was established in each of the 300 m × 300 m grid cells covering the study area to determine the potential impact of human activities on the surface water dynamic. In terms of the inter-annual variation during 2002–2016, PW presented an overall stability, and NPW occupied only a small part of the water surface in each year and presented an inter-annual fluctuation. NPW had a smaller patch size, with lower connectivity degree and higher fragmentation degree. In terms of spatial variation during 2002–2016, NPW often occurred around PW, and its configurational features varied from place to place. Moreover, PW mostly corresponded to the natural lagoon, and salt marsh (as a part of lagoons), and NPW had a strong link with arable land (agricultural irrigation) and salt marsh (salt production), sand beach/dune, coastal wetlands and lagoon for the LULC maps of 2003, 2012 and 2015. However, more in-depth analysis is required for understanding the impact of sand beach/dune, coastal wetlands and lagoon on surface water dynamics. This study covers the long-term variations of surface water patterns in a Mediterranean lagoon complex having intense and diverse human activities, and the potential link between LULC types and the water dynamic scenarios was investigated on different dates. The results of the study should be useful for environmental management and protection of coastal lagoons. [ABSTRACT FROM AUTHOR]

As the most biodiversity-rich part of the protected areas system, habitats within the pilot national parks have long been threatened by drastic human-induced land use and land cover changes. The growing concern about habitat loss has spurred China's national park project to shift from pilot to construction phase with the official establishment of China's first group of national parks (CFGNPs) in October 2021. But far too little attention has been paid to the synergistic work concerning the habitat quality (HQ) dynamics of all five national parks. Here, the InVEST model, combined with a satellite-derived land use and land cover product and a hot spot analysis (HSA) method, was used to investigate the HQ dynamics at the park- and pixel-scale within the CFGNPs. Our results demonstrate that the past ecological conservation practices within national parks have been unpromising, especially in Giant Panda National Park, Northeast China Tiger and Leopard National Park (NCTL), and Wuyi Mountain National Park (WYM), where HQ as a whole showed a significant decline. Furthermore, more than half of Hainan Tropical Rainforest National Park (87.2%), WYM (77.4%), and NCTL (52.9%) showed significant HQ degradation from 1980 to 2019. Besides, increasing trends in the area shares of HQ degraded pixels were observed in all five national parks from 1980–1999 to 2000–2019. The HSA implied that the hot spots of high HQ degradation rates tend to occur in areas closer to urban settlements or on the edge of national parks, where human activities are intensive. Despite these disappointing findings, we highlighted from the observed local successes and the HQ plateau that the construction of CFGNPs is expected to reverse the deteriorating HQ trends. Thus, we concluded our paper by proposing an HSA-based regulatory zoning scheme that includes five subzones to guide the future construction of China's national park system. [Display omitted] • The first known investigation to reveal the habitat quality trends for China's first group of national parks was presented. • The past ecological conservation practices within national parks have been unpromising over the past four decades. • The area shares of habitat quality degraded pixels show an increasing trend in all five national parks from 1980 to 2019. • We developed a regulatory zoning scheme for habitat protection or ecosystem restoration within these national parks. [ABSTRACT FROM AUTHOR]

Land use and land cover maps provide fundamental information that has been used in different kinds of studies, ranging from climate change to city planning. However, despite substantial efforts in recent decades, large-scale 30-m land cover maps still suffer from relatively low accuracy in terms of land cover type discrimination (especially for the vegetation and impervious types), due to limits in relation to the data, method, and design of the workflow. In this work, we improved the land cover classification accuracy by integrating free and public high-resolution Google Earth images (HR-GEI) with Landsat Operational Land Imager (OLI) and Enhanced Thematic Mapper Plus (ETM+) imagery. Our major innovation is a hybrid approach that includes three major components: (1) a deep convolutional neural network (CNN)-based classifier that extracts high-resolution features from Google Earth imagery; (2) traditional machine learning classifiers (i.e., Random Forest (RF) and Support Vector Machine (SVM)) that are based on spectral features extracted from 30-m Landsat data; and (3) an ensemble decision maker that takes all different features into account. Experimental results show that our proposed method achieves a classification accuracy of 84.40% on the entire validation dataset in China, improving the previous state-of-the-art accuracies obtained by RF and SVM by 4.50% and 4.20%, respectively. Moreover, our proposed method reduces misclassifications between certain vegetation types, and improves identification of the impervious type. Evaluation applied over an area of around 14,000 km2 confirms little improvement for land cover types (e.g., forest) of which the classification accuracies are already over 80% when using traditional machine learning approaches, yet improvements in accuracy of 7% for cropland and shrubland, 9% for grassland, 23% for impervious and 25% for wetlands were achieved when compared with traditional machine learning approaches. The results demonstrate the great potential of integrating features of datasets at different resolutions and the possibility to produce more reliable land cover maps. • Combining Google Earth images and Landsat data for land cover mapping • Fusing high-resolution spatial features and medium-resolution spectral features • Improving the previous highest OA from 80% to 84% on all samples in China • Reducing confusions among different vegetation and impervious types • Validating the method with 5 selected regions with a total area of about 14,000 km2 [ABSTRACT FROM AUTHOR]