EARice10: a 10 m resolution annual rice distribution map of East Asia for 2023

Earth system science data, Journal Year: 2025, Volume and Issue: 17(2), P. 661 - 683

Published: Feb. 11, 2025

Abstract. Timely and accurate high-resolution annual mapping of rice distribution is essential for food security, greenhouse gas emissions assessment, support sustainable development goals. East Asia (EA), a major global rice-producing region, accounts approximately 29.3 % the world's production. Therefore, to acquire latest EA, this study proposed novel method based on Google Earth Engine (GEE) platform, producing 10 m resolution map (EARice10) EA 2023. A new synthetic aperture radar (SAR)-based index (SRMI) was firstly combined with optical indices generate representative samples. In addition, stacking-based optical–SAR adaptive fusion model designed fully integrate features Sentinel-1 Sentinel-2 data high-precision in EA. The accuracy EARice10 evaluated using more than 90 000 validation samples achieved an overall 90.48 %, both user producer exceeding %. reliability product verified by R2 values ranging between 0.94 0.98 respect official statistics 0.79 previous products. accessible at https://doi.org/10.5281/zenodo.13118409 (Song et al., 2024).

Language: Английский

---

The 20 m Africa rice distribution map of 2023

Earth system science data, Journal Year: 2025, Volume and Issue: 17(5), P. 1781 - 1805

Published: May 6, 2025

Abstract. In recent years, the demand for rice in Africa has been growing rapidly, and, order to meet this demand, cultivation area is also expanding rapidly; thus, it of great significance monitor Africa. The spatial and temporal distribution complex, making difficult use phenology-based identification methods, existing products are all made up grid-based statistical data with a low resolution, unable obtain accurate field location available labels. To address these two difficulties, based on time series optical dual-polarization synthetic aperture radar (SAR) data, study proposes sample set construction method by means fast-coarse-positioning-assisted visual interpretation feature-importance-guided supervised classification combining multiple SAR features reduce impact diversity Firstly, we vertical transmit, horizontal receive (VH) fast coarse positioning screening possible areas combine auxiliary construct set; secondly, complementary information 20 m map 2023 was completed object-oriented segmentation results images pixel-based after feature selection. average accuracy proposed validation more than 85 %, R2 linear fit various 0.9, which proves that can achieve mapping under complex climatic conditions large region, providing crucial support monitoring agricultural policy development. dataset at https://doi.org/10.5281/zenodo.13729353 (Jiang et al., 2024).

Language: Английский

---

Classification and spatio-temporal evolution analysis of coastal wetlands in the Liaohe Estuary from 1985 to 2023: based on feature selection and sample migration methods

Frontiers in Forests and Global Change, Journal Year: 2024, Volume and Issue: 7

Published: Aug. 16, 2024

Coastal wetlands are important areas with valuable natural resources and diverse biodiversity. Due to the influence of both factors human activities, landscape coastal undergoes significant changes. It is crucial systematically monitor analyze dynamic changes in wetland cover over a long-term time series. In this paper, series remote sensing classification process was proposed, which integrated feature selection sample migration. Utilizing Google Earth Engine (GEE) Landsat TM/ETM/OLI image data, selected set combined migration method generate training for each target year. The Simple Non-Iterative Clustering-Random Forest (SNIC-RF) model ultimately employed accurately map classes Liaohe Estuary from 1985 2023 quantitatively evaluate spatio-temporal pattern change characteristics study area. findings indicate that: (1) After selection, accuracy reached 0.88, separation good. (2) migration, overall year ranged 87 94%, along Kappa coefficients 0.84 0.92, thereby ensuring validity (3) SNIC-RF results showed better performance landscape. Compared RF classification, increased by 0.69–5.82%, coefficient 0.0087–0.0751. (4) From 2023, there has been predominant trend being converted into artificial wetlands. recent years, transition occurred more gently. Finally, offers insights understanding trends surface ecological environment Estuary. research can be extended other types comprehensive application hydrology, ecology, meteorology, soil, further explored on basis research, laying strong groundwork shaping policies protection restoration.

Language: Английский

---

EARice10: A 10 m Resolution Annual Rice Distribution Map of East Asia for 2023

Published: Aug. 28, 2024

Abstract. Timely and accurate high-resolution annual mapping of rice distribution is essential for food security, greenhouse gas emissions assessment supporting sustainable development goals. East Asia (EA), a major global rice-producing region, accounts approximately 29.3 % the world's production. Therefore, to acquire latest EA, this study proposed novel method based on Google Earth Engine (GEE) platform, producing 10-meter-resolution map (EARice10) EA 2023. A new Synthetic Aperture Radar (SAR)-based Rice Mapping Index (SRMI) was firstly combined with optical indices generate representative samples. In addition, stacking-based optical-SAR adaptive fusion model designed fully integrate features Sentinel-1 Sentinel-2 data high-precision in EA. The accuracy EARice10 evaluated using more than 90,000 validation samples achieved an overall 90.48 %, both user’s producer’s accuracies exceeding 90 %. reliability product verified by R2 values ranging between 0.94 0.98 respect official statistics, 0.79 previous products. accessible at https://doi.org/10.5281/zenodo.13118409 (Song et al., 2024).

Language: Английский

---

Machine learning-based early prediction of rice-growing fields using multi-temporal Sentinel-1 synthetic aperture radar and Sentinel-2 multispectral data

Journal of Applied Remote Sensing, Journal Year: 2024, Volume and Issue: 18(03)

Published: Aug. 14, 2024

Rice is the most important food crop in Taiwan. Early information on rice-growing conditions thus vital for estimating rice production to guarantee national security and grain exports. The rice-harvested area conventionally inspected twice a year by costly interpretation of aerial photographs intensive labor-field surveys. However, such methods monitoring are inadequate providing government with timely rice-cultivated conditions. This study aims use time series Sentinel-1 synthetic aperture radar Sentinel-2 multispectral data develop machine-learning approach early prediction fields An object-based random forest (OBRF) was developed process remotely sensed rice-cropping seasons from 2018 2021. results compared reference showed that could be accurately predicted before harvest, about three months first two second crop. F-score Kappa coefficient values achieved were 0.87 0.85, those 0.72 0.71, respectively. These findings reaffirmed close agreement between official statistics estimated satellite data, correlation determination (R2) value greater than 0.96. A large portion crop's areas abandoned or converted upland cultivation crop, which confirmed visual Landsat images statistics. Ultimately, this proved efficacy using Sentinel-1/2 OBRF Quantitative geographical produced essential estimation nationally address concerns.

Language: Английский

---

Automated rice mapping using multitemporal Sentinel-1 SAR imagery using dynamic threshold and slope-based index methods

Remote Sensing Applications Society and Environment, Journal Year: 2024, Volume and Issue: unknown, P. 101410 - 101410

Published: Nov. 1, 2024

Language: Английский

---

Object Based Classification in Google Earth Engine Combining SNIC and Machine Learning Methods (Case Study: Lake Köyceğiz)

Turkish Journal of Remote Sensing and GIS, Journal Year: 2024, Volume and Issue: unknown, P. 125 - 137

Published: March 17, 2024

Köyceğiz Lake is one of our country’s most critical coastal barrier lakes, rich in sulfur, located at the western end Mediterranean Region. Lake, connected to via Dalyan Strait, 7 lakes world with this feature. In study, water change analysis was carried out by integrating Object-Based Image Classification method CART (Classification and Regression Tree), RF (Random Forest), SVM (Support Vector Machine) algorithms, which are machine learning algorithms. SNIC (Simple Non-iterative Clustering) segmentation used, allows a detailed object level dividing image into super pixels. Sentinel 2 Harmonized images study area were obtained from Google Earth Engine (GEE) platform for 2019, 2020, 2021, 2022,and all calculations made GEE. When classification accuracies four years examined, it seen that accuracies(OA, UA, PA, Kappa) lake above 92%, F-score 0.98 methods using object-based combination algorithm CART, RF, It has been determined higher evaluation metrics determining than methods.

Language: Английский

---

Research on insurance decision-making based on the TOPSIS evaluation model and K-means clustering algorithm

Highlights in Science Engineering and Technology, Journal Year: 2024, Volume and Issue: 101, P. 867 - 873

Published: May 20, 2024

Increasing losses caused by extreme weather events led to property insurance costs soaring and becoming more challenging obtain, leaving companies owners in a severe crisis. This study aims address the challenges posed climate change industry explore impact of on insurance. We introduce TOPSIS evaluation model K-means clustering algorithm assess risk level different regions provide optimal basis for decisions. By analyzing global disaster data emission indicators, this paper found that some face rejecting underwriting, used algorithm. The results show increase aggravates industry, but through comprehensive assessment cluster analysis, we can accurate decision support companies. assessing effective classification regions, it provides insurers with support. research are great significance dealing problems help better adapt change.

Language: Английский

---

Optimizing Feature Selection for Solar Park Classification: Approaches with OBIA and Machine Learning

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 286 - 301

Published: Jan. 1, 2024

Language: Английский

---

Dinámica de inundaciones ambientales en humedales de la Cuenca baja del Rio Grijalva: enfoque espaciotemporal a través de imágenes Landsat

Revista de Teledetección, Journal Year: 2024, Volume and Issue: 64, P. 75 - 87

Published: July 29, 2024

La diversidad de metodologías existentes para definir y analizar la dinámica las superficies agua muestra dificultad que genera investigar su comportamiento, aunado a existen variables dificultan delimitación tales como precipitación o evapotranspiración. Este trabajo tuvo objetivo espaciotemporal humedales alto impacto socioambiental en Cuenca Baja del Rio Grijalva el periodo 1986 2018. Para análisis se integró una base datos satelital con 169 imágenes Landsat 5 8. Se calcularon índices espectrales (MNDWI MBWI) identificaron los umbrales caracterizan área estudio. Los resultados mostraron MBWI fue superior estimación agua. Finalmente, generaron mapas probabilidades mayor importancia ecológica económica CBRG. Estos revelaron periodos retorno procesos expansión retroceso longitudinal Niña formación temporales podría estar asociado saturación manto freático no aportes superficiales.

---