The Classification Method Study of Crops Remote Sensing with Deep Learning, Machine Learning, and Google Earth Engine

Remote Sensing, Год журнала: 2022, Номер 14(12), С. 2758 - 2758

Опубликована: Июнь 8, 2022

The extraction and classification of crops is the core issue agricultural remote sensing. precise crop types great significance to monitoring evaluation planting area, growth, yield. Based on Google Earth Engine Colab cloud platform, this study takes typical oasis area Xiangride Town, Qinghai Province, as an example. It compares traditional machine learning (random forest, RF), object-oriented (object-oriented, OO), deep neural networks (DNN), which proposes a random forest combined with network (RF+DNN) framework. In study, spatial characteristics band information, vegetation index, polarization main in were constructed using Sentinel-1 Sentinel-2 data. temporal phenology growth state analyzed curve curvature method, data screened time space. By comparing analyzing accuracy four methods, advantages RF+DNN model its application value illustrated. results showed that for during period good development, better result could be obtained whose accuracy, training, predict spent than DNN alone. overall Kappa coefficient 0.98 0.97, respectively. also higher (OA = 0.87, 0.82), object oriented 0.78, 0.70) 0.93, 0.90). scalable simple method proposed paper gives full play platform operation, can effectively improve accuracy. Timely accurate at different scales pattern change, yield estimation, safety warning.

Язык: Английский

---

Mapping Vegetation Changes in Mongolian Grasslands (1990–2024) Using Landsat Data and Advanced Machine Learning Algorithm

Remote Sensing, Год журнала: 2025, Номер 17(3), С. 400 - 400

Опубликована: Янв. 24, 2025

Grassland ecosystems provide a range of services in semi-arid and arid regions. However, they have significantly declined due to overgrazing desertification. In the current study, we employed Landsat time series data (TM, OLI, OLI-2) spanning from 1990 2024, combined with vegetation indices such as NDVI SAVI, along NDWI digital elevation models (DEMs), analyze land cover dynamics Ugii Lake watershed area, Mongolia. By integrating multisource remote sensing into advanced XGBoost (extreme gradient boosting) machine learning algorithm, achieved high classification accuracy, overall accuracies exceeding 94% Kappa coefficients greater than 0.92. The results revealed decline montane grasslands (−6.2%) an increase other grassland types, suggesting ecosystem redistribution influenced by climatic anthropogenic factors. Cropland exhibited resilience, recovering significant 1990s moderate growth 2024. Our findings highlight stability barren underscore pressures ecological degradation human activities. This study provides up-to-date statistical support decision-making conservation sustainable management Mongolia under changing conditions.

Язык: Английский

---

Land Use/Land Cover Mapping Based on GEE for the Monitoring of Changes in Ecosystem Types in the Upper Yellow River Basin over the Tibetan Plateau

Remote Sensing, Год журнала: 2022, Номер 14(21), С. 5361 - 5361

Опубликована: Окт. 26, 2022

The upper Yellow River basin over the Tibetan Plateau (TP) is an important ecological barrier in northwestern China. Effective LULC products that enable monitoring of changes regional ecosystem types are great importance for their environmental protection and macro-control. Here, we combined 18-class classification scheme based on with Sentinel-2 imagery, Google Earth Engine (GEE) platform, random forest method to present new a spatial resolution 10 m 2018 2020 Basin TP conducted types. results indicated that: (1) In 2020, overall accuracy (OA) maps ranged between 87.45% 93.02%. (2) Grassland was main first-degree class research area, followed by wetland water bodies barren land. For second-degree class, grassland, broadleaf shrub marsh. (3) types, largest area progressive succession (positive) grassland–shrubland (451.13 km2), whereas retrogressive (negative) grassland–barren (395.91 km2). areas were grassland–broadleaf (344.68 km2) desert land–grassland (302.02 shrubland–grassland (309.08 grassland–bare rock (193.89 northern southwestern parts study showed trend towards positive succession, south-central Huangnan, northeastern Gannan, central Aba Prefectures signs purpose this provide basis data basin-scale analysis more detailed categories reliable accuracy.

Язык: Английский

---

Early Identification of Crop Type for Smallholder Farming Systems Using Deep Learning on Time-Series Sentinel-2 Imagery

Sensors, Год журнала: 2023, Номер 23(4), С. 1779 - 1779

Опубликована: Фев. 5, 2023

Climate change and the COVID-19 pandemic have disrupted food supply chain across globe adversely affected security. Early estimation of staple crops can assist relevant government agencies to take timely actions for ensuring Reliable crop type maps play an essential role in monitoring crops, estimating yields, maintaining smooth supplies. However, these are not available developing countries until matured about be harvested. The use remote sensing accurate crop-type mapping first few weeks sowing remains challenging. Smallholder farming systems diverse types further complicate challenge. For this study, a ground-based survey is carried out map fields by recording coordinates planted respective fields. time-series images mapped acquired from Sentinel-2 satellite. A deep learning-based long short-term memory network used at early growth stage. Results show that including rice, wheat, sugarcane, classified with 93.77% accuracy as four sowing. proposed method applied on large scale effectively smallholder farms stage, allowing authorities plan seamless availability food.

Язык: Английский

---

CARM30: China annual rapeseed maps at 30 m spatial resolution from 2000 to 2022 using multi-source data

Scientific Data, Год журнала: 2024, Номер 11(1)

Опубликована: Апрель 8, 2024

Rapeseed is a critical cash crop globally, and understanding its distribution can assist in refined agricultural management, ensuring sustainable vegetable oil supply, informing government decisions. China the leading consumer third-largest producer of rapeseed. However, there lack widely available, long-term, large-scale remotely sensed maps on rapeseed cultivation China. Here this study utilizes multi-source data such as satellite images, GLDAS environmental variables, land cover maps, terrain to create annual at 30 m spatial resolution from 2000 2022 (CARM30). Our product was validated using independent samples showed average F1 scores 0.869 0.971 for winter spring The CARM30 has high consistency with existing 10 20 maps. Additionally, CARM30-derived planted area significantly correlated statistics (R2 = 0.65-0.86; p < 0.001). obtained information serve reference stakeholders farmers, scientific communities, decision-makers.

Язык: Английский

---

AI4Boundaries: an open AI-ready dataset to map field boundaries with Sentinel-2 and aerial photography

Earth system science data, Год журнала: 2023, Номер 15(1), С. 317 - 329

Опубликована: Янв. 18, 2023

Abstract. Field boundaries are at the core of many agricultural applications and a key enabler for operational monitoring production to support food security. Recent scientific progress in deep learning methods has highlighted capacity extract field from satellite aerial images with clear improvement object-based image analysis (e.g. multiresolution segmentation) or conventional filters Sobel filters). However, these need labels be trained on. So far, no standard data set exists easily robustly benchmark models state art. The absence such further impedes proper comparison against existing methods. Besides, there is consensus on which evaluation metrics should reported (both pixel levels). As result, it currently impossible compare new To fill gaps, we introduce AI4Boundaries, readily usable train boundary detection. AI4Boundaries includes two specific sets: (i) 10 m Sentinel-2 monthly composites large-scale analyses retrospect (ii) 1 orthophoto regional-scale analyses, as automatic extraction Geospatial Aid Application (GSAA). All have been sourced GSAA that made openly available (Austria, Catalonia, France, Luxembourg, Netherlands, Slovenia, Sweden) 2019, representing 14.8 M parcels covering 376 K km2. Data were selected following stratified random sampling drawn based landscape fragmentation metrics, perimeter/area ratio area covered by parcels, thus considering diversity landscapes. resulting “AI4Boundaries” dataset consists 7831 samples 256 pixels 512 orthophoto. Both datasets provided corresponding vector ground-truth parcel delineation (2.5 47 105 km2), raster version already pre-processed ready use. Besides providing this open foster computer vision developments methods, discuss perspectives limitations various types agriculture domain consider possible improvements. JRC Open Catalogue: http://data.europa.eu/89h/0e79ce5d-e4c8-4721-8773-59a4acf2c9c9 (European Commission, Joint Research Centre, 2022).

Язык: Английский

---

Investigating the Potential of Crop Discrimination in Early Growing Stage of Change Analysis in Remote Sensing Crop Profiles

Remote Sensing, Год журнала: 2023, Номер 15(3), С. 853 - 853

Опубликована: Фев. 3, 2023

Currently, remote sensing crop identification is mostly based on all available images acquired throughout growth. However, the image and data resources in early growth stage are limited, which makes challenging. Different types have different phenological characteristics seasonal rhythm characteristics, their rates at times. Therefore, making full use of to augment difference information times key identification. In this study, we first calculated differential features between periods as new during stage. Secondly, multi-temporal each period were constructed by combination, then a feature optimization method was used obtain optimal set possible combinations crops, well change explored. Finally, performance classification regression tree (Cart), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Support Vector Machine (SVM) classifiers recognizing crops analyzed. The results show that: (1) There differences with rice changing significantly F, corn E, M, L, H, soybean N, H. (2) For rice, land surface water index (LSWI), simple ratio (SR), B11, normalized tillage (NDTI) contributed most, while red-edge3 (NDRE3), LSWI, green vegetation (VIgreen), red-edge spectral (RESI), red-edge2 (NDRE2) greatly (3) Rice could be identified 13 May, PA UA high 95%. Corn soybeans 7 July, 97% 94%, respectively. (4) With addition more temporal features, recognition accuracy increased. GBDT RF performed best identifying three This study demonstrates feasibility using for recognition, can provide idea recognition.

Язык: Английский

---

Innovative Decision Fusion for Accurate Crop/Vegetation Classification with Multiple Classifiers and Multisource Remote Sensing Data

Remote Sensing, Год журнала: 2024, Номер 16(9), С. 1579 - 1579

Опубликована: Апрель 29, 2024

Obtaining accurate and real-time spatial distribution information regarding crops is critical for enabling effective smart agricultural management. In this study, innovative decision fusion strategies, including Enhanced Overall Accuracy Index (E-OAI) voting the Index-based Majority Voting (OAI-MV), were introduced to optimize use of diverse remote sensing data various classifiers, thereby improving accuracy crop/vegetation identification. These strategies utilized integrate classification outcomes from distinct feature sets (including Gaofen-6 reflectance, Sentinel-2 time series vegetation indices, biophysical variables, Sentinel-1 backscatter coefficients, their combinations) using classifiers (Random Forests (RFs), Support Vector Machines (SVMs), Maximum Likelihood (ML), U-Net), taking two grain-producing areas (Site #1 Site #2) in Haixi Prefecture, Qinghai Province, China, as research area. The results indicate that employing U-Net on feature-combined yielded highest overall (OA) 81.23% 91.49% #2, respectively, single classifier experiments. E-OAI strategy, compared original OAI boosted OA by 0.17% 6.28%. Furthermore, OAI-MV strategy achieved 86.02% 95.67% respective study sites. This highlights strengths features discerning different crop types. Additionally, proposed effectively harness benefits multisource features, significantly enhancing classification.

Язык: Английский

---

Seasonal Land Use and Land Cover Mapping in South American Agricultural Watersheds Using Multisource Remote Sensing: The Case of Cuenca Laguna Merín, Uruguay

Sensors, Год журнала: 2025, Номер 25(1), С. 228 - 228

Опубликована: Янв. 3, 2025

Recent advancements in Earth Observation sensors, improved accessibility to imagery and the development of corresponding processing tools have significantly empowered researchers extract insights from Multisource Remote Sensing. This study aims use these technologies for mapping summer winter Land Use/Land Cover features Cuenca de la Laguna Merín, Uruguay, while comparing performance Random Forests, Support Vector Machines, Gradient-Boosting Tree classifiers. The materials include Sentinel-2, Sentinel-1 Shuttle Radar Topography Mission imagery, Google Engine, training validation datasets quoted methods involve creating a multisource database, conducting feature importance analysis, developing models, supervised classification performing accuracy assessments. Results indicate low significance microwave inputs relative optical features. Short-wave infrared bands transformations such as Normalised Vegetation Index, Surface Water Index Enhanced demonstrate highest importance. Accuracy assessments that various classes is optimal, particularly rice paddies, which play vital role country’s economy highlight significant environmental concerns. However, challenges persist reducing confusion between classes, regarding natural vegetation versus seasonally flooded vegetation, well post-agricultural fields/bare land herbaceous areas. Forests Trees exhibited superior compared Machines. Future research should explore approaches Deep Learning pixel-based object-based integration address identified challenges. These initiatives consider data combinations, including additional indices texture metrics derived Grey-Level Co-Occurrence Matrix.

Язык: Английский

---

Advancements in crop mapping through remote sensing: A comprehensive review of concept, data sources, and procedures over four decades

Remote Sensing Applications Society and Environment, Год журнала: 2025, Номер unknown, С. 101527 - 101527

Опубликована: Март 1, 2025

Язык: Английский

---