Global 10 m Land Use Land Cover Datasets: A Comparison of Dynamic World, World Cover and Esri Land Cover

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(16), P. 4101 - 4101

Published: Aug. 21, 2022

The European Space Agency’s Sentinel satellites have laid the foundation for global land use cover (LULC) mapping with unprecedented detail at 10 m resolution. We present a cross-comparison and accuracy assessment of Google’s Dynamic World (DW), ESA’s Cover (WC) Esri’s Land (Esri) products first time in order to inform adoption application these maps going forward. For year 2020, three LULC show strong spatial correspondence (i.e., near-equal area estimates) water, built area, trees crop classes. However, relative one another, WC is biased towards over-estimating grass cover, Esri shrub scrub DW snow ice. Using ground truth data minimum unit 250 m2, we found that had highest overall (75%) compared (72%) (65%). Across all maps, water was most accurately mapped class (92%), followed by (83%), tree (81%) crops (78%), particularly biomes characterized temperate boreal forests. classes lowest accuracies, tundra biome, included (47%), (34%), bare (57%) flooded vegetation (53%). When using from LUCAS (Land Use/Cover Area Frame Survey) <100 (71%) (66%) (63%), highlighting ability resolve landscape elements more Esri. Although not analyzed our study, discuss advantages due its frequent near real-time delivery both categorical predictions probability scores. recommend should involve critical evaluation their suitability respect purpose, such as aggregate changes ecosystem accounting versus site-specific change detection monitoring, considering trade-offs between thematic resolution, versus. local accuracy, class-specific biases whether analysis necessary. also emphasize importance estimating areas pixel-counting alone but adopting best practices design-based inference estimation quantify uncertainty given study area.

Language: Английский

---

Land Use and Land Cover Mapping Using Sentinel-2, Landsat-8 Satellite Images, and Google Earth Engine: A Comparison of Two Composition Methods

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(9), P. 1977 - 1977

Published: April 20, 2022

Accurate and real-time land use/land cover (LULC) maps are important to provide precise information for dynamic monitoring, planning, management of the Earth. With advent cloud computing platforms, time series feature extraction techniques, machine learning classifiers, new opportunities arising in more accurate large-scale LULC mapping. In this study, we aimed at finding out how two composition methods spectral–temporal metrics extracted from satellite can affect ability a classifier produce maps. We used Google Earth Engine (GEE) platform create cloud-free Sentinel-2 (S-2) Landsat-8 (L-8) over Tehran Province (Iran) as 2020. Two methods, namely, seasonal composites percentiles metrics, were define four datasets based on series, vegetation indices, topographic layers. The random forest was classification identifying most variables. Accuracy assessment results showed that S-2 outperformed L-8 overall class level. Moreover, comparison indicated percentile both series. At level, improved performance related their better about phenological variation different classes. Finally, conclude methodology GEE an fast way be

Language: Английский

---

Assessing and improving the transferability of current global spatial prediction models

Global Ecology and Biogeography, Journal Year: 2023, Volume and Issue: 32(3), P. 356 - 368

Published: Jan. 26, 2023

Abstract Aim Global‐scale maps of the environment are an important source information for researchers and decision makers. Often, these created by training machine learning algorithms on field‐sampled reference data using remote sensing as predictors. Since field samples often sparse clustered in geographic space, model prediction requires a transfer trained to regions where no available. However, recent studies question feasibility predictions far beyond location data. Innovation We propose novel workflow spatial predictive mapping that leverages developments this combines them innovative ways with aim improved transferability performance assessment. demonstrate, evaluate discuss from recently published global environmental maps. Main conclusions Reducing predictors those relevant leads increase map accuracy without decrease quality areas high sampling density. Still, reliable gap‐free were not possible, highlighting their evaluation hampered limited availability

Language: Английский

---

National tree species mapping using Sentinel-1/2 time series and German National Forest Inventory data

Remote Sensing of Environment, Journal Year: 2024, Volume and Issue: 304, P. 114069 - 114069

Published: Feb. 24, 2024

Spatially explicit and detailed information on tree species composition is critical for forest management, nature conservation the assessment of ecosystem services. In many countries, attributes are monitored regularly through sample-based inventories. combination with satellite imagery, data from such inventories have a great potential developing large-area maps. Here, high temporal resolution Sentinel-1 Sentinel-2 has been useful extracting vegetation phenology, that may also be valuable improving mapping. The objective this study was to map main in Germany using combined time series, identify address challenges related use National Forest Inventory (NFI) remote sensing applications. We generated cloud free series 5-day intervals imagery combine those monthly backscatter composites. Further, we incorporate topography, meteorology, climate account environmental gradients. To NFI training machine learning models, following challenges: 1) link pixels variable radius plots, which precise area unknown, 2) efficiently utilize mixed-species plots model validation. past, accuracies pixel-level maps were often estimated solely homogeneous pure-species stands. study, assess how well generalize mixed plot conditions. Our results show mapping large, environmentally diverse landscapes. Classification accuracy pure stands ranged between 72% 97% (F1-score) five dominant species, while less frequent remained challenging. When including assessment, decreased by 4–14 percentage points most groups. highlights importance mixed-forest when validating Based these results, discuss potentials remaining at national level. findings allow further improve national-level medium provide guidance similar approaches other countries where ground-based inventory available.

Language: Английский

---

Comparative validation of recent 10 m-resolution global land cover maps

Remote Sensing of Environment, Journal Year: 2024, Volume and Issue: 311, P. 114316 - 114316

Published: July 18, 2024

Accurate and high-resolution land cover (LC) information is vital for addressing contemporary environmental challenges. With the advancement of satellite data acquisition, cloud-based processing, deep learning technology, Global Land Cover (GLC) map production has become increasingly feasible. a growing number available GLC maps, comprehensive evaluation comparison necessary to assess their accuracy suitability diverse uses. This particularly applies maps lacking statistically robust assessment or sufficient reported detail on validation procedures. study conducts comparative independent recent 10 m namely ESRI Use/Land (LULC), ESA WorldCover, Google World Resources Institute (WRI)'s Dynamic World, examining spatial representation thematic at global, continental, national (for 47 larger countries) levels. Since impacted by reference uncertainty owing geolocation labelling errors, five approaches dealing with were evaluated. Of considered approaches, using sample label supplemented majority within neighborhood found produce more reasonable estimates compared overly optimistic approach any pessimistic direct between labels. Overall global accuracies range 73.4% ± 0.7% (95% confidence interval) 83.8% 0.4% WorldCover having highest followed LULC. The quality varies across different LC classes, continents, countries. maps' was assessed various homogeneity levels 3 × kernel. Although as this reveals that LULC have less than WorldCover. All lower in heterogenous landscapes some countries such Mozambique, Tanzania, Nigeria, Spain. To select most suitable product, users should consider both map's over area interest appropriate application. For future mapping, producers are encouraged adopt standardized class definitions ensure comparability maps. Additionally, heterogeneous key features be improved versions Independent efforts regional levels, well changes, strengthened enhance utility these scales long-term monitoring.

Language: Английский

---

Is green space associated with reduced crime? A national-scale study from the Global South

The Science of The Total Environment, Journal Year: 2022, Volume and Issue: 825, P. 154005 - 154005

Published: Feb. 19, 2022

Assumptions about the link between green space and crime mitigation are informed by literature that overwhelmingly originates in Global North. Little is known association spaces South. We utilized 10 years of precinct-level statistics (n = 1152) over South Africa, a global hotspot, to test hypothesis associated with reduced rates. found that, after controlling for number socio-demographic confounders (unemployment, income, age, education, land use population density), every 1% increase total there 1.2% (0.7 1.7%; 95% confidence interval) decrease violent crime, 1.3% (0.8 1.8%) property no effect on sexual crimes. However, direction changed crimes when exploring characteristics including tree cover park accessibility. Property 0.4% (0.1 0.7%) percentage cover, 0.9% (0.5 1.3%) kilometer proximity public park. Further research, experimental studies, needed better isolate causal mechanisms behind crime-green associations, especially considering may map race income inequality be more reporting affluent areas. Nevertheless, our results provide complementary contribution evidence from North, highlighting need nuanced definitions its links crime. When viewed light broader suite ecosystem services provided space, support urban greening as major strategy towards achieving just sustainable cities towns.

Language: Английский

---

Designing a European-Wide Crop Type Mapping Approach Based on Machine Learning Algorithms Using LUCAS Field Survey and Sentinel-2 Data

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(3), P. 541 - 541

Published: Jan. 23, 2022

One of the most challenging aspects obtaining detailed and accurate land-use land-cover (LULC) maps is availability representative field data for training validation. In this manuscript, we evaluate use Eurostat Land Use Coverage Area frame Survey (LUCAS) 2018 to generate a LULC map with 19 crop type classes two broad categories woodland shrubland, grassland. The were used in combination Copernicus Sentinel-2 (S2) satellite covering Europe. First, spatially temporally consistent S2 image composites (1) spectral reflectances, (2) selection indices, (3) several bio-geophysical indicators created year 2018. From large number features, important selected classification using machine-learning algorithms (support vector machine random forest). Results indicated that could be classified an overall accuracy (OA) 77.6%, independent Our analysis three methods select optimum showed by selecting spectrally different pixels data, best OA achieved, already only 11% total data. Comparing our results similar study Sentinel-1 (S1) can achieve slightly better results, although spatial coverage was reduced due gaps Further ongoing leverage synergies between optical microwave

Language: Английский

---

National wetland mapping using remote-sensing-derived environmental variables, archive field data, and artificial intelligence

Heliyon, Journal Year: 2023, Volume and Issue: 9(2), P. e13482 - e13482

Published: Feb. 1, 2023

While wetland ecosystem services are widely recognized, the lack of fine-scale national inventories prevents successful implementation conservation policies. Wetlands difficult to map due their complex fine-grained spatial pattern and fuzzy boundaries. However, increasing amount open high-spatial-resolution remote sensing data accurately georeferenced field archives, as well progress in artificial intelligence (AI), provide opportunities for mapping. The objective this study was wetlands over mainland France (ca. 550,000 km2) by applying AI environmental variables derived from archive data. A random forest model calibrated using cross-validation according precision-recall area under curve (PR-AUC) index ca. 135,000 soil or flora plots databases, 5 m topographical an airborne DTM a geological map. validated experimentally designed sampling strategy with 3000 collected during ground survey 2021 along non-wetland/wetland transects. Map accuracy then compared those nine existing maps global, European, coverage. model-derived suitability (PR-AUC 0.76) highlights gradual boundaries wetlands. binary is significantly more accurate (F1-score 0.75, overall 0.67) than maps. approach end-results important value planning management since high-resolution enable targeted measures support biodiversity conservation, water resources maintenance, carbon storage.

Language: Английский

---

Uncertainty quantification for probabilistic machine learning in earth observation using conformal prediction

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: July 13, 2024

Abstract Machine learning is increasingly applied to Earth Observation (EO) data obtain datasets that contribute towards international accords. However, these contain inherent uncertainty needs be quantified reliably avoid negative consequences. In response the increased need report uncertainty, we bring attention promise of conformal prediction within domain EO. Unlike previous quantification methods, offers statistically valid regions while concurrently supporting any machine model and distribution. To support for prediction, reviewed EO found only 22.5% incorporated a degree information, with unreliable methods prevalent. Current open implementations require moving large amounts algorithms. We introduced Google Engine native modules compute, facilitating integration into existing traditional deep modelling workflows. demonstrate versatility scalability tools apply them valued applications spanning local global extents, regression, classification tasks. Subsequently, discuss opportunities arising from use in anticipate accessible easy-to-use tools, such as those provided here, will drive wider adoption rigorous EO, thereby enhancing reliability downstream uses operational monitoring decision-making.

Language: Английский

---

The Third Generation of Pan-Canadian Wetland Map at 10 m Resolution Using Multisource Earth Observation Data on Cloud Computing Platform

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2021, Volume and Issue: 14, P. 8789 - 8803

Published: Jan. 1, 2021

Development of the Canadian Wetland Inventory Map (CWIM) has thus far proceeded over two generations, reporting extent and location bog, fen, swamp, marsh, water wetlands across country with increasing accuracy. Each generation this training inventory improved previous results by including additional reference wetland data focusing on processing at scale ecozone, which represent ecologically distinct regions Canada. The first second generations attained relatively highly accurate an average approaching 86% though some overestimated extents, particularly swamp class. current research represents a third refinement map. It was designed to improve overall accuracy (OA) reduce overestimation modifying test train integrating environmental remote sensing datasets, countrywide coverage L-band ALOS PALSAR-2, SRTM, Arctic digital elevation model, nighttime light, temperature, precipitation data. Using random forest classification within Google Earth Engine, OA obtained for CWIM3 is 90.53%, improvement 4.77% results. All ecozones experienced increase 2% or greater individual ecozone range between 94% highest 84% lowest. Visual inspection products demonstrates reduction area compared generations. In study, several scenarios were defined assess effect preprocessing benefits incorporating multisource large-scale mapping. addition, development confidence map helps visualize where are most least reliable given amount recent landscape disturbance (e.g., fire). resulting OAs areal reveal importance adequate scale.

Language: Английский

---