Google Earth Engine for geo-big data applications: A meta-analysis and systematic review

ISPRS Journal of Photogrammetry and Remote Sensing, Journal Year: 2020, Volume and Issue: 164, P. 152 - 170

Published: May 7, 2020

Language: Английский

---

Land-Use Land-Cover Classification by Machine Learning Classifiers for Satellite Observations—A Review

Remote Sensing, Journal Year: 2020, Volume and Issue: 12(7), P. 1135 - 1135

Published: April 2, 2020

Rapid and uncontrolled population growth along with economic industrial development, especially in developing countries during the late twentieth early twenty-first centuries, have increased rate of land-use/land-cover (LULC) change many times. Since quantitative assessment changes LULC is one most efficient means to understand manage land transformation, there a need examine accuracy different algorithms for mapping order identify best classifier further applications earth observations. In this article, six machine-learning algorithms, namely random forest (RF), support vector machine (SVM), artificial neural network (ANN), fuzzy adaptive resonance theory-supervised predictive (Fuzzy ARTMAP), spectral angle mapper (SAM) Mahalanobis distance (MD) were examined. Accuracy was performed by using Kappa coefficient, receiver operational curve (RoC), index-based validation root mean square error (RMSE). Results coefficient show that all classifiers similar level minor variation, but RF algorithm has highest 0.89 MD (parametric classifier) least 0.82. addition, visual cross-validation (correlations between normalised differentiation water index, vegetation index built-up are 0.96, 0.99 1, respectively, at 0.05 significance) comparison other adopted. Findings from literature also proved ANN classifiers, although non-parametric like SAM (Kappa 0.84; area under (AUC) 0.85) better consistent than algorithms. Finally, review concludes classifier, among examined it necessary test morphoclimatic conditions future.

Language: Английский

---

Google Earth Engine Cloud Computing Platform for Remote Sensing Big Data Applications: A Comprehensive Review

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2020, Volume and Issue: 13, P. 5326 - 5350

Published: Jan. 1, 2020

Remote sensing (RS) systems have been collecting massive volumes of datasets for decades, managing and analyzing which are not practical using common software packages desktop computing resources. In this regard, Google has developed a cloud platform, called Earth Engine (GEE), to effectively address the challenges big data analysis. particular, platform facilitates processing geo over large areas monitoring environment long periods time. Although was launched in 2010 proved its high potential different applications, it fully investigated utilized RS applications until recent years. Therefore, study aims comprehensively explore aspects GEE including datasets, functions, advantages/limitations, various applications. For purpose, 450 journal articles published 150 journals between January May 2020 were studied. It observed that Landsat Sentinel extensively by users. Moreover, supervised machine learning algorithms, such as Random Forest, more widely applied image classification tasks. also employed broad range Land Cover/land Use classification, hydrology, urban planning, natural disaster, climate analyses, processing. generally number publications significantly increased during past few years, is expected will be users from fields resolve their challenges.

Language: Английский

---

GLC_FCS30: global land-cover product with fine classification system at 30 m using time-series Landsat imagery

Earth system science data, Journal Year: 2021, Volume and Issue: 13(6), P. 2753 - 2776

Published: June 15, 2021

Abstract. Over past decades, a lot of global land-cover products have been released; however, these still lack map with fine classification system and spatial resolution simultaneously. In this study, novel 30 m for the year 2015 (GLC_FCS30-2015) was produced by combining time series Landsat imagery high-quality training data from GSPECLib (Global Spatial Temporal Spectra Library) on Google Earth Engine computing platform. First, were developed applying rigorous filters to CCI_LC (Climate Change Initiative Global Land Cover) MCD43A4 NBAR (MODIS Nadir Bidirectional Reflectance Distribution Function-Adjusted Reflectance). Secondly, local adaptive random forest model built each 5∘×5∘ geographical tile using multi-temporal spectral texture features corresponding data, GLC_FCS30-2015 product containing types generated tile. Lastly, validated three different validation systems (containing details) 44 043 samples. The results indicated that achieved an overall accuracy 82.5 % kappa coefficient 0.784 level-0 (9 basic types), 71.4 0.686 UN-LCCS (United Nations Cover Classification System) level-1 (16 LCCS 68.7 0.662 level-2 (24 types). comparisons against other (CCI_LC, MCD12Q1, FROM_GLC, GlobeLand30) provides more details than CCI_LC-2015 MCD12Q1-2015 greater diversity FROM_GLC-2015 GlobeLand30-2010. They also showed best 59.1 GlobeLand30-2010 75.9 %. Therefore, it is concluded first dataset 16 as well 14 detailed regional types) high at m. in paper are free access https://doi.org/10.5281/zenodo.3986872 (Liu et al., 2020).

Language: Английский

---

Land Cover Classification using Google Earth Engine and Random Forest Classifier—The Role of Image Composition

Remote Sensing, Journal Year: 2020, Volume and Issue: 12(15), P. 2411 - 2411

Published: July 27, 2020

Land cover information plays a vital role in many aspects of life, from scientific and economic to political. Accurate about land affects the accuracy all subsequent applications, therefore accurate timely is high demand. In classification studies over past decade, higher accuracies were produced when using time series satellite images than single date images. Recently, availability Google Earth Engine (GEE), cloud-based computing platform, has gained attention remote sensing based applications where temporal aggregation methods derived are widely applied (i.e., use metrics such as mean or median), instead GEE, simply select possible fill gaps without concerning how different year/season might affect accuracy. This study aims analyze effect composition methods, well input images, on results. We Landsat 8 surface reflectance (L8sr) data with eight combination strategies produce evaluate maps for area Mongolia. implemented experiment GEE platform algorithm, Random Forest (RF) classifier. Our results show that datasets moderately highly maps, overall 84.31%. Among datasets, two summer scenes (images 1 June 30 September) highest (89.80% 89.70%), followed by median composite same (88.74%). The difference between these three classifications was not significant McNemar test (p > 0.05). However, < 0.05) observed other pairs involving one datasets. indicate (e.g., median) promising method, which only significantly reduces volume (resulting an easier faster analysis) but also produces equally data. spatial consistency among relatively low compared general accuracy, showing selection dataset used any important crucial step, because play essential classification, particularly snowy, cloudy expansive areas like

Language: Английский

---

Mapping cropping intensity in China using time series Landsat and Sentinel-2 images and Google Earth Engine

Remote Sensing of Environment, Journal Year: 2019, Volume and Issue: 239, P. 111624 - 111624

Published: Dec. 30, 2019

Language: Английский

---

The First Wetland Inventory Map of Newfoundland at a Spatial Resolution of 10 m Using Sentinel-1 and Sentinel-2 Data on the Google Earth Engine Cloud Computing Platform

Remote Sensing, Journal Year: 2018, Volume and Issue: 11(1), P. 43 - 43

Published: Dec. 28, 2018

Wetlands are one of the most important ecosystems that provide a desirable habitat for great variety flora and fauna. Wetland mapping modeling using Earth Observation (EO) data essential natural resource management at both regional national levels. However, accurate wetland is challenging, especially on large scale, given their heterogeneous fragmented landscape, as well spectral similarity differing classes. Currently, precise, consistent, comprehensive inventories national- or provincial-scale lacking globally, with studies focused generation local-scale maps from limited remote sensing data. Leveraging Google Engine (GEE) computational power availability high spatial resolution collected by Copernicus Sentinels, this study introduces first detailed, inventory map richest Canadian provinces in terms extent. In particular, multi-year summer Synthetic Aperture Radar (SAR) Sentinel-1 optical Sentinel-2 composites were used to identify distribution five three non-wetland classes Island Newfoundland, covering an approximate area 106,000 km2. The classification results evaluated pixel-based object-based random forest (RF) classifications implemented GEE platform. revealed superiority approach relative mapping. Although was more compared SAR, inclusion types significantly improved accuracies overall accuracy 88.37% Kappa coefficient 0.85 achieved SAR/optical composite RF classification, wherein all correctly identified beyond 70% 90%, respectively. suggest paradigm-shift standard static products approaches toward generating dynamic, on-demand, large-scale coverage through advanced cloud computing resources simplify access processing “Geo Big Data.” addition, resulting ever-demanding Newfoundland interest can be many stakeholders, including federal provincial governments, municipalities, NGOs, environmental consultants name few.

Language: Английский

---

Improved land cover map of Iran using Sentinel imagery within Google Earth Engine and a novel automatic workflow for land cover classification using migrated training samples

ISPRS Journal of Photogrammetry and Remote Sensing, Journal Year: 2020, Volume and Issue: 167, P. 276 - 288

Published: July 29, 2020

Language: Английский

---

The 10-m crop type maps in Northeast China during 2017–2019

Scientific Data, Journal Year: 2021, Volume and Issue: 8(1)

Published: Feb. 2, 2021

Abstract Northeast China is the leading grain production region in where one-fifth of national produced; however, consistent and reliable crop maps are still unavailable, impeding management decisions for regional food security. Here, we produced annual 10-m major crops (maize, soybean, rice) from 2017 to 2019, by using (1) a hierarchical mapping strategy (cropland followed classification), (2) agro-climate zone-specific random forest classifiers, (3) interpolated smoothed 10-day Sentinel-2 time series data, (4) optimized features spectral, temporal, texture characteristics land surface. The resultant have high overall accuracies (OA) spanning 0.81 0.86 based on abundant ground truth data. satellite estimates agreed well with statistical data most municipalities (R 2 ≥ 0.83, p < 0.01). This first effort at resolution, which permits assessing performance soybean rejuvenation plan rotation practice China.

Language: Английский

---

Recent Applications of Landsat 8/OLI and Sentinel-2/MSI for Land Use and Land Cover Mapping: A Systematic Review

Remote Sensing, Journal Year: 2020, Volume and Issue: 12(18), P. 3062 - 3062

Published: Sept. 18, 2020

Recent applications of Landsat 8 Operational Land Imager (L8/OLI) and Sentinel-2 MultiSpectral Instrument (S2/MSI) data for acquiring information about land use cover (LULC) provide a new perspective in remote sensing analysis. Jointly, these sources permit researchers to improve operational classification change detection, guiding better reasoning landscape intrinsic processes, as deforestation agricultural expansion. However, the results their have not yet been synthesized order coherent guidance on effect different well identify promising approaches issues which affect performance. In this systematic review, we present trends, potentialities, challenges, actual gaps, future possibilities L8/OLI S2/MSI LULC mapping detection. particular, highlight possibility using medium-resolution (Landsat-like, 10–30 m) time series multispectral optical provided by harmonization between sensors cube architectures analysis-ready that are permeated publicizations, open policies, science principles. We also reinforce potential exploring more spectral bands combinations, especially three Red-edge two Near Infrared Shortwave S2/MSI, calculate vegetation indices sensitive phenological variations were less frequently applied long time, but turned since mission. Summarizing peer-reviewed papers can guide scientific community data, enable detailed knowledge detection landscapes, natural scenarios.

Language: Английский

---