Abstract:The RADARSAT Constellation Mission (RCM) Compact Polarimetry (CP) data has become a key asset in crop mapping and monitoring for diverse agricultural landscapes. This study utilizes the unique capabilities of the RCM CP data for crop mapping. It performs a detailed comparison between single-date and multi-date classification to underscore the prowess of multi-temporal CP data in crop mapping. The novelty of our approach is in the thorough investigation of real CP data, a significant advancement from previous studies that mainly relied on simulated CP data. The CP data used in this study were acquired on July 1, July 30, and August 27, 2021, over southern Quebec, Canada, including soy, corn, hay, and cereal classes. Various features were extracted from the CP data, and the Random Forest classifier was utilized for crop mapping. The experimental results demonstrated the superiority of multi-temporal CP data for crop classification. The Overall Accuracy (OA) for single-date classifications on July 1, July 30, and August 27 were 61.10%, 75.00%, and 86.45%, respectively. In contrast, the multi-date analysis showed a marked increase in OA (91.20%). This substantial improvement underscores the significant benefit of incorporating multi-date CP data, which delivers a robust and precise framework for crop mapping. Les données radar à synthèse d'ouverture en mode polarimétrie compacte (PC) de la Mission Constellation RADARSAT (MCR) sont devenues un atout majeur pour la cartographie et la. surveillance des cultures dans divers paysages agricoles. Cette étude utilise les capacités uniques des données P C de la mission RCM pour la cartographie des cultures. Elle effectue une comparaison détaillée entre la classification à une date et la classification à plusieurs dates pour souligner la petinance des données PC multitemporelles dans la cartographie des cultures. La nouveauté de notre approche réside dans l'évaluation approfondi des données CP réelles, ce qui constitue une avancée significative par rapport aux études précédentes qui s'appuyaient principalement sur des données PC simulées. Les données de PC utilisées dans cette étude ont été acquises le 1er juillet, le 30 juillet et le 27 août 2021, dans une region au sud du Québec, Canada, et comprennent des classes de soja, de maïs, de foin et de céréales. Diverses caractéristiques ont été extraites des données P C, et le classificateur Forêt des arbes aléatoires (Random Forest) a été utilisé pour la cartographie des cultures. Les résultats expérimentaux ont démontré la supériorité des données PC multitemporelles pour la classification des cultures. La précision globale (Overaal Accuracy: OA) pour les classifications à une seule date, le 1er juillet, le 30 juillet et le 27 août, était respectivement de 61,10 %, 75,00 % et 86,45 %. En revanche, l'analyse multi-dates a montré une nette augmentation de l'OA (91,20 %). Cette amélioration substantielle souligne l'avantage significatif de l'incorporation de données P C multi-dates, qui fournit un cadre robuste et précis pour la cartographie des cultures.

Simultaneous multi-crop land suitability prediction from remote sensing data using semi-supervised learning

Interaction of climate, topography and soil properties with cropland and cropping pattern using remote sensing data and machine learning methods

Multimodal Crop Type Classification Fusing Multi-Spectral Satellite Time Series with Farmers Crop Rotations and Local Crop Distribution

Incorporating Multi-Temporal Remote Sensing and a Pixel-Based Deep Learning Classification Algorithm to Map Multiple-Crop Cultivated Areas

Pre-Constrained Machine Learning Method for Multi-Year Mapping of Three Major Crops in a Large Irrigation District

Agricultural land suitability classification and crop suggestion using machine learning and spatial multicriteria decision analysis in semi-arid ecosystem

Crop Yield Prediction through Proximal Sensing and Machine Learning Algorithms

Mapping Seasonal Agricultural Land Use Types Using Deep Learning on Sentinel-2 Image Time Series

Crop Monitoring using remote sensing land use and land change data: Comparative analysis of deep learning methods using pre-trained CNN models

Field-Level Crop Classification Using an Optimal Dataset of Multi-temporal Sentinel-1 and Polarimetric RADARSAT-2 SAR Data with Machine Learning Algorithms

Remote Sensing Crop Recognition by Coupling Phenological Features and Off-Center Bayesian Deep Learning

County-level corn yield prediction using supervised machine learning

Improved method for cropland extraction of seasonal crops from multi-sensor satellite data

The Canadian Cropland Dataset: A New Land Cover Dataset for Multitemporal Deep Learning Classification in Agriculture

From data to harvest: Leveraging ensemble machine learning for enhanced crop yield predictions across Canada amidst climate change

Advancing crop classification in smallholder agriculture: A multifaceted approach combining frequency-domain image co-registration, transformer-based parcel segmentation, and Bi-LSTM for crop classification

Crop Classification Using Multi-Temporal RADARSAT Constellation Mission Compact Polarimetry SAR Data

Optimal County-Level Crop Yield Prediction Using MODIS-based Variables and Weather Data: A Comparative Study on Machine Learning Models

Enhanced crop classification through integrated optical and SAR data: a deep learning approach for multi-source image fusion

Evaluating the Effectiveness of Machine Learning and Deep Learning Models Combined Time-Series Satellite Data for Multiple Crop Types Classification over a Large-Scale Region

Multimodal crop cover identification using deep learning and remote sensing