Automatic extraction of land cover statistics from satellite imagery by deep learning1

2021 ◽  
pp. 1-17
Author(s):  
Eleonora Bernasconi ◽  
Fabrizio De Fausti ◽  
Francesco Pugliese ◽  
Monica Scannapieco ◽  
Diego Zardetto
Keyword(s):  
Land Cover ◽  
Satellite Imagery ◽  
Ground Truth ◽  
Careful Analysis ◽  
Production Costs ◽  
Moderate Resolution ◽  
Estimation System ◽  
Land Cover Maps ◽  
Sentinel 2 ◽  
Good Agreement

In this paper, we address the challenge of producing fully automated land cover estimates from satellite imagery through Deep Learning algorithms. We developed our system according to a tile-based, classify-and-count design. To implement the classification engine of the system, we adopted a cutting-edge Convolutional Neural Network model named Inception-V3, which we customized and trained for scene classification on the EuroSAT dataset. We tested and validated our system on two Sentinel-2 images representing quite different Italian territories (with an area of 751 km2 and 443 km2, respectively). Because no genuine ground-truth is available for the land cover of these sub-regional territories, we built a pseudo ground-truth by integrating land cover information from flagship European projects CORINE and LUCAS. A critical and careful analysis shows that our automatic land cover estimates are in good agreement with the pseudo ground-truth and offers extensive evidence of the remarkable segmentation ability of our system. The limits of our approach are also critically discussed in the paper and possible countermeasures are illustrated. When compared with traditional projects like CORINE and LUCAS, our automatic land cover estimation system exhibits three fundamental advantages: it can dramatically reduce production costs; it can allow delivering very timely and frequent land cover statistics; it can enable land cover estimation for very small territorial areas, well beyond the NUTS-2 level. As an additional outcome of land cover estimation, our system also automatically generates moderate resolution land cover maps that might be used in cartography projects as an agile first-level tool for map update or change detection purposes.

scholarly journals Continental-Scale Land Cover Mapping at 10 m Resolution Over Europe (ELC10)

2021 ◽  
Vol 13 (12) ◽  
pp. 2301
Author(s):  
Zander Venter ◽  
Markus Sydenham
Keyword(s):  
Land Cover ◽  
Atmospheric Correction ◽  
Google Earth ◽  
Tree Planting ◽  
Training Data ◽  
Classification Model ◽  
Night Time ◽  
Continental Scale ◽  
Land Cover Maps ◽  
Sentinel 2

Land cover maps are important tools for quantifying the human footprint on the environment and facilitate reporting and accounting to international agreements addressing the Sustainable Development Goals. Widely used European land cover maps such as CORINE (Coordination of Information on the Environment) are produced at medium spatial resolutions (100 m) and rely on diverse data with complex workflows requiring significant institutional capacity. We present a 10 m resolution land cover map (ELC10) of Europe based on a satellite-driven machine learning workflow that is annually updatable. A random forest classification model was trained on 70K ground-truth points from the LUCAS (Land Use/Cover Area Frame Survey) dataset. Within the Google Earth Engine cloud computing environment, the ELC10 map can be generated from approx. 700 TB of Sentinel imagery within approx. 4 days from a single research user account. The map achieved an overall accuracy of 90% across eight land cover classes and could account for statistical unit land cover proportions within 3.9% (R2 = 0.83) of the actual value. These accuracies are higher than that of CORINE (100 m) and other 10 m land cover maps including S2GLC and FROM-GLC10. Spectro-temporal metrics that capture the phenology of land cover classes were most important in producing high mapping accuracies. We found that the atmospheric correction of Sentinel-2 and the speckle filtering of Sentinel-1 imagery had a minimal effect on enhancing the classification accuracy (< 1%). However, combining optical and radar imagery increased accuracy by 3% compared to Sentinel-2 alone and by 10% compared to Sentinel-1 alone. The addition of auxiliary data (terrain, climate and night-time lights) increased accuracy by an additional 2%. By using the centroid pixels from the LUCAS Copernicus module polygons we increased accuracy by <1%, revealing that random forests are robust against contaminated training data. Furthermore, the model requires very little training data to achieve moderate accuracies—the difference between 5K and 50K LUCAS points is only 3% (86 vs. 89%). This implies that significantly less resources are necessary for making in situ survey data (such as LUCAS) suitable for satellite-based land cover classification. At 10 m resolution, the ELC10 map can distinguish detailed landscape features like hedgerows and gardens, and therefore holds potential for aerial statistics at the city borough level and monitoring property-level environmental interventions (e.g., tree planting). Due to the reliance on purely satellite-based input data, the ELC10 map can be continuously updated independent of any country-specific geographic datasets.

scholarly journals An Assessment of Changes the Land Use Land Cover and Impact of the It Park at Talwade Pune On the Nearby Villages

2020 ◽  
pp. 172-179
Author(s):  
Ujjwala Khare ◽  
Prajakta Thakur
Keyword(s):  
Land Use ◽  
Land Cover ◽  
Rapid Growth ◽  
Urban Areas ◽  
Satellite Images ◽  
Ground Truth ◽  
Land Use Land Cover ◽  
Landsat Images ◽  
Tertiary Sector ◽  
Land Cover Maps

<p>The expansion of urban areas is common in metropolitan cities in India. Pune also has experienced rapid growth in the fringe areas of the city. This is mainly on account of the development of the Information Technology (IT) Parks. These IT Parks have been established in different parts of Pune city. They include Hinjewadi, Kharadi, Talwade and others like the IT parks in Magarpatta area. The IT part at Talwade is located to close to Pune Nashik Highway has had an impact on the villages located around it. The surrounding area includes the villages of Talwade, Chikhli, Nighoje, Mahalunge, Khalumbre and Sudumbre.</p> <p>The changes in the land use that have occurred in areas surrounding Talwade IT parks during the last three decades have been studied by analyzing the LANDSAT images of different time periods. The satellite images of the 1992, 2001 and 2011 were analyzed to detect the temporal changes in the land use and land cover.</p> <p>This paper attempts to study the changes in land use / land cover which has taken place in these villages in the last two decades. Such a study can be done effectively with the help of remote sensing and GIS techniques. The tertiary sector has experienced a rapid growth especially during the last decade near the IT Park. The occupation structure of these villages is also related to the changes due to the development of the IT Park.</p> <p>The land use of study area has been analysed using the ground truth applied to the satellite images at decadal interval. Using the digital image processing techniques, the satellite images were then classified and land use / land cover maps were derived. The results show that the area under built-up land has increased by around 14 per cent in the last 20 years. On the contrary, the land under agriculture, barren, pasture has decreased significantly.</p>

scholarly journals Mapping avalanches with satellites – evaluation of performance and completeness

2020 ◽  
Author(s):  
Elisabeth D. Hafner ◽  
Frank Techel ◽  
Silvan Leinss ◽  
Yves Bühler
Keyword(s):  
Satellite Imagery ◽  
Ground Truth ◽  
Probability Of Detection ◽  
Hazard Zonation ◽  
Alpine Regions ◽  
Evaluation Of Performance ◽  
Measure Design ◽  
Time Periods ◽  
Avalanche Warning ◽  
Sentinel 2

Abstract. The spatial distribution and size of avalanches are essential parameters for avalanche warning, avalanche documentation, mitigation measure design and hazard zonation. Despite its importance, this information is incomplete today and only available for limited areas and limited time periods. Manual avalanche mapping from satellite imagery has recently been applied to reduce this gap achieving promising results. However, their reliability and completeness were not yet verified satisfactorily. In our study we attempt a full validation of the completeness of visually detected and mapped avalanches from optical SPOT-6, Sentinel-2 and radar Sentinel-1 imagery. We examine manually mapped avalanches from two avalanche periods in 2018 and 2019 for an area of approximately 180 km2 around Davos, Switzerland relying on ground- and helicopter-based photographs as ground truth. For the quality assessment, we investigate the Probability of Detection (POD) and the Positive Predictive Value (PPV). Additionally, we relate our results to conditions which potentially influence avalanche detection in the satellite imagery. We statistically confirm the high potential of SPOT for comprehensive avalanche mapping for selected periods (POD = 0.74, PPV = 0.88) as well as the reliability of Sentinel-1 for the mapping of larger avalanches (POD = 0.27, PPV = 0.87). Furthermore, we proof that Sentinel-2 is unsuitable for the mapping of most avalanches due to its spatial resolution (POD = 0.06, PPV = 0.81). Because we could apply the same reference avalanche events for all three satellite mappings, our validation results are robust and comparable. We demonstrate that satellite-based avalanche mapping has the potential to fill the existing avalanche documentation gap over large areas, making alpine regions safer.

scholarly journals ASSESSMENTS OF SENTINEL-2 VEGETATION RED-EDGE SPECTRAL BANDS FOR IMPROVING LAND COVER CLASSIFICATION

2017 ◽  
Vol XLII-2/W7 ◽  
pp. 871-874 ◽  
Author(s):  
S. Qiu ◽  
B. He ◽  
C. Yin ◽  
Z. Liao
Keyword(s):  
Land Cover ◽  
Land Cover Classification ◽  
Relative Importance ◽  
East Texas ◽  
Spectral Bands ◽  
Red Edge ◽  
Sentinel 2A ◽  
Land Cover Maps ◽  
Confusion Matrices ◽  
Sentinel 2

The Multi Spectral Instrument (MSI) onboard Sentinel-2 can record the information in Vegetation Red-Edge (VRE) spectral domains. In this study, the performance of the VRE bands on improving land cover classification was evaluated based on a Sentinel-2A MSI image in East Texas, USA. Two classification scenarios were designed by excluding and including the VRE bands. A Random Forest (RF) classifier was used to generate land cover maps and evaluate the contributions of different spectral bands. The combination of VRE bands increased the overall classification accuracy by 1.40&amp;thinsp;%, which was statistically significant. Both confusion matrices and land cover maps indicated that the most beneficial increase was from vegetation-related land cover types, especially agriculture. Comparison of the relative importance of each band showed that the most beneficial VRE bands were Band 5 and Band 6. These results demonstrated the value of VRE bands for land cover classification.

scholarly journals Towards High-Resolution Land-Cover Classification of Greenland: A Case Study Covering Kobbefjord, Disko and Zackenberg

2021 ◽  
Vol 13 (18) ◽  
pp. 3559
Author(s):  
Daniel Alexander Rudd ◽  
Mojtaba Karami ◽  
Rasmus Fensholt
Keyword(s):  
Land Cover ◽  
Arctic Region ◽  
Ecological Monitoring ◽  
Land Cover Classification ◽  
Temporal Analysis ◽  
Google Earth ◽  
The Arctic ◽  
Arctic Ecosystems ◽  
Land Cover Maps ◽  
Sentinel 2

Mapping of the Arctic region is increasingly important in light of global warming as land cover maps can provide the foundation for upscaling of ecosystem properties and processes. To this end, satellite images provide an invaluable source of Earth observations to monitor land cover in areas that are otherwise difficult to access. With the continuous development of new satellites, it is important to optimize the existing maps for further monitoring of Arctic ecosystems. This study presents a scalable classification framework, producing novel 10 m resolution land cover maps for Kobbefjord, Disko, and Zackenberg in Greenland. Based on Sentinel-2, a digital elevation model, and Google Earth Engine (GEE), this framework classifies the areas into nine classes. A vegetation land cover classification for 2019 is achieved through a multi-temporal analysis based on 41 layers comprising phenology, spectral indices, and topographical features. Reference data (1164 field observations) were used to train a random forest classifier, achieving a cross-validation accuracy of 91.8%. The red-edge bands of Sentinel-2 data proved to be particularly well suited for mapping the fen vegetation class. The study presents land cover mapping in the three study areas with an unprecedented spatial resolution and can be extended via GEE for further ecological monitoring in Greenland.

scholarly journals A development of the Enhanced Built-up and Bareness Index (EBBI) based on combination of multi-resolution Landsat 8 and Sentinel 2 MSI images

2021 ◽  
Vol 62 (1) ◽  
pp. 1-9
Author(s):  
Hung Le Trinh ◽  
Ha Thu Thi Le ◽  
Loc Duc Le ◽  
Long Thanh Nguyen ◽  
Keyword(s):  
Land Cover ◽  
Satellite Imagery ◽  
Near Infrared ◽  
Urban Land ◽  
Urban Land Use ◽  
Landsat 8 ◽  
Infrared Band ◽  
Bare Land ◽  
Sentinel 2

Classification of built-up land and bare land on remote sensing images is a very difficult problem due to the complexity of the urban land cover. Several urban indices have been proposed to improve the accuracy in classifying urban land use/land cover from optical satellite imagery. This paper presents an development of the EBBI (Enhanced Built-up and Bareness Index) index based on the combination of Landsat 8 and Sentinel 2 multi-resolution satellite imagery. Near infrared band (band 8a), short wave infrared band (band 11) of Sentinel 2 MSI image and thermal infrared band (band 10) Landsat 8 image were used to calculate EBBI index. The results obtained show that the combination of Landsat 8 and Sentinel 2 satellite images improves the spatial resolution of EBBI index image, thereby improving the accuracy of classification of bare land and built-up land by about 5% compared with the case using only Landsat 8 images.

scholarly journals Evaluation of Satellite Imagery for Monitoring Pacific Walruses at a Large Coastal Haulout

2021 ◽  
Vol 13 (21) ◽  
pp. 4266
Author(s):  
Anthony S. Fischbach ◽  
David C. Douglas
Keyword(s):  
Satellite Imagery ◽  
Chukchi Sea ◽  
Late Summer ◽  
Visual Interpretation ◽  
Odobenus Rosmarus Divergens ◽  
Radar Imagery ◽  
Moderate Resolution ◽  
Satellite Sensors ◽  
Sar Imagery ◽  
Sentinel 2

Pacific walruses (Odobenus rosmarus divergens) are using coastal haulouts in the Chukchi Sea more often and in larger numbers to rest between foraging bouts in late summer and autumn in recent years, because climate warming has reduced availability of sea ice that historically had provided resting platforms near their preferred benthic feeding grounds. With greater numbers of walruses hauling out in large aggregations, new opportunities are presented for monitoring the population. Here we evaluate different types of satellite imagery for detecting and delineating the peripheries of walrus aggregations at a commonly used haulout near Point Lay, Alaska, in 2018–2020. We evaluated optical and radar imagery ranging in pixel resolutions from 40 m to ~1 m: specifically, optical imagery from Landsat, Sentinel-2, Planet Labs, and DigitalGlobe, and synthetic aperture radar (SAR) imagery from Sentinel-1 and TerraSAR-X. Three observers independently examined satellite images to detect walrus aggregations and digitized their peripheries using visual interpretation. We compared interpretations between observers and to high-resolution (~2 cm) ortho-corrected imagery collected by a small unoccupied aerial system (UAS). Roughly two-thirds of the time, clouds precluded clear optical views of the study area from satellite. SAR was unaffected by clouds (and darkness) and provided unambiguous signatures of walrus aggregations at the Point Lay haulout. Among imagery types with 4–10 m resolution, observers unanimously agreed on all detections of walruses, and attained an average 65% overlap (sd 12.0, n 100) in their delineations of aggregation boundaries. For imagery with ~1 m resolution, overlap agreement was higher (mean 85%, sd 3.0, n 11). We found that optical satellite sensors with moderate resolution and high revisitation rates, such as PlanetScope and Sentinel-2, demonstrated robust and repeatable qualities for monitoring walrus haulouts, but temporal gaps between observations due to clouds were common. SAR imagery also demonstrated robust capabilities for monitoring the Point Lay haulout, but more research is needed to evaluate SAR at haulouts with more complex local terrain and beach substrates.

scholarly journals Key Landscapes for Conservation Land Cover and Change Monitoring Thematic and Validation Datasets for Sub-Saharan Africa

2020 ◽  
Author(s):  
Zoltan Szantoi ◽  
Andreas Brink ◽  
Andrea Lupi ◽  
Claudio Mannone ◽  
Gabriel Jaffrain
Keyword(s):  
Land Cover ◽  
Satellite Imagery ◽  
Time Series Data ◽  
Sub Saharan Africa ◽  
Series Data ◽  
Ecological Zones ◽  
Complete Dataset ◽  
Risk Areas ◽  
Sub Saharan ◽  
Land Cover Maps

Abstract. Threats to biodiversity pose an enormous challenge for Africa. Mounting social and economic demands on natural resources increasingly threaten key areas for conservation. Effective protection of sites of strategic conservation importance requires timely and highly detailed geospatial monitoring. Larger ecological zones and wildlife corridors warrant monitoring as well, as these areas have an even higher degree of pressure and habitat loss. To address this, a satellite imagery based monitoring workflow to cover at-risk areas at various details was developed. During the program's first phase, a total of 560,442 km2 area in Sub-Saharan Africa was covered, from which 153,665 km2 were mapped with 8 land cover classes while 406,776 km2 were mapped with up to 32 classes. Satellite imagery was used to generate dense time series data from which thematic land cover maps were derived. Each map and change map were fully verified and validated by an independent team to achieve our strict data quality requirements. The independent validation datasets for each KLCs are also described and presented here (The complete dataset available at Szantoi et al., 2020B https://doi.org/10.1594/PANGAEA.914261, and a demonstration dataset at Szantoi et al., 2020C https://doi.pangaea.de/10.1594/PANGAEA.915849).

scholarly journals The Spatial and Temporal Land Cover Patterns of the Qazaly Irrigation Zone in 2003–2018: The Case of Syrdarya River’s Lower Reaches, Kazakhstan

2019 ◽  
Vol 11 (15) ◽  
pp. 4035 ◽  
Author(s):  
Kanat Samarkhanov ◽  
Jilili Abuduwaili ◽  
Alim Samat ◽  
Gulnura Issanova
Keyword(s):  
Land Cover ◽  
Vegetation Index ◽  
Normalized Difference Vegetation Index ◽  
Learning Algorithm ◽  
Aral Sea ◽  
Landsat 8 ◽  
Spatial And Temporal Patterns ◽  
Moderate Resolution ◽  
Moderate Resolution Imaging Spectroradiometer ◽  
Land Cover Maps

In this study, the spatial and temporal patterns of the land cover were monitored within the Qazaly irrigation zone located in the deltaic zone of the Syrdarya river in the surroundings of the former Aral Sea. A 16-day MODIS (Moderate Resolution Imaging Spectroradiometer) Aqua NDVI (Normalized Difference Vegetation Index) data product with a spatial resolution of 250 meters was used for this purpose, covering the period between 2003 and 2018. Field survey results obtained in 2018 were used to build a sample dataset. The random forests supervised classification machine learning algorithm was used to map land cover, which produced good results with an overall accuracy of about 0.8. Statistics on land cover change were calculated and analyzed. The correctness of obtained classes was checked with Landsat 8 (OLI, The Operational Land Imager) images. Detailed land cover maps, including rice cropland, were derived. During the observation period, the rice croplands increased, while the generally irrigated area decreased.

Sign in / Sign up

Export Citation Format

Share Document