Crop Mapping without Labels: Investigating Temporal and Spatial Transferability of Crop Classification Models Using a 5-Year Sentinel-2 Series and Machine Learning

被引:8
作者
Rusnak, Tomas [1 ]
Kasanicky, Tomas [2 ]
Malik, Peter [2 ]
Mojzis, Jan [2 ]
Zelenka, Jan [2 ]
Svicek, Michal [3 ]
Abraham, Dominik [3 ]
Halabuk, Andrej [1 ]
机构
[1] Slovak Acad Sci, Inst Landscape Ecol, Vvi, Stefanikova 3, Bratislava 81499, Slovakia
[2] Slovak Acad Sci, Inst Informat, Vvi, Dubravska Cesta 9, Bratislava 84507, Slovakia
[3] Natl Agr & Food Ctr NPPC, Hlohovecka 2, Luzianky 95141, Slovakia
关键词
multitemporal classification; Google Earth Engine; within-season crop mapping; domain adaptation; agricultural monitoring; crop monitoring; IMAGE TIME-SERIES; LANDSAT; DATASET; SCALE;
D O I
10.3390/rs15133414
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Multitemporal crop classification approaches have demonstrated high performance within a given season. However, cross-season and cross-region crop classification presents a unique transferability challenge. This study addresses this challenge by adopting a domain generalization approach, e.g., by training models on multiple seasons to improve generalization to new, unseen target years. We utilize a comprehensive five-year Sentinel-2 dataset over different agricultural regions in Slovakia and a diverse crop scheme (eight crop classes). We evaluate the performance of different machine learning classification algorithms, including random forests, support vector machines, quadratic discriminant analysis, and neural networks. Our main findings reveal that the transferability of models across years differs between regions, with the Danubian lowlands demonstrating better performance (overall accuracies ranging from 91.5% in 2022 to 94.3% in 2020) compared to eastern Slovakia (overall accuracies ranging from 85% in 2022 to 91.9% in 2020). Quadratic discriminant analysis, support vector machines, and neural networks consistently demonstrated high performance across diverse transferability scenarios. The random forest algorithm was less reliable in generalizing across different scenarios, particularly when there was a significant deviation in the distribution of unseen domains. This finding underscores the importance of employing a multi-classifier analysis. Rapeseed, grasslands, and sugar beet consistently show stable transferability across seasons. We observe that all periods play a crucial role in the classification process, with July being the most important and August the least important. Acceptable performance can be achieved as early as June, with only slight improvements towards the end of the season. Finally, employing a multi-classifier approach allows for parcel-level confidence determination, enhancing the reliability of crop distribution maps by assuming higher confidence when multiple classifiers yield similar results. To enhance spatiotemporal generalization, our study proposes a two-step approach: (1) determine the optimal spatial domain to accurately represent crop type distribution; and (2) apply interannual training to capture variability across years. This approach helps account for various factors, such as different crop rotation practices, diverse observational quality, and local climate-driven patterns, leading to more accurate and reliable crop classification models for nationwide agricultural monitoring.
引用
收藏
页数:35
相关论文
共 49 条
  • [1] Introducing ARTMO's Machine-Learning Classification Algorithms Toolbox: Application to Plant-Type Detection in a Semi-Steppe Iranian Landscape
    Aghababaei, Masoumeh
    Ebrahimi, Ataollah
    Naghipour, Ali Asghar
    Asadi, Esmaeil
    Perez-Suay, Adrian
    Morata, Miguel
    Garcia, Jose Luis
    Rivera Caicedo, Juan Pablo
    Verrelst, Jochem
    [J]. REMOTE SENSING, 2022, 14 (18)
  • [2] Phenology-based sample generation for supervised crop type classification
    Belgiu, Mariana
    Bijker, Wietske
    Csillik, Ovidiu
    Stein, Alfred
    [J]. INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 2021, 95
  • [3] Random forests
    Breiman, L
    [J]. MACHINE LEARNING, 2001, 45 (01) : 5 - 32
  • [4] Bull AD, 2011, J MACH LEARN RES, V12, P2879
  • [5] A high-performance and in-season classification system of field-level crop types using time-series Landsat data and a machine learning approach
    Cai, Yaping
    Guan, Kaiyu
    Peng, Jian
    Wang, Shaowen
    Seifert, Christopher
    Wardlow, Brian
    Li, Zhan
    [J]. REMOTE SENSING OF ENVIRONMENT, 2018, 210 : 35 - 47
  • [6] The Harmonized Landsat and Sentinel-2 surface reflectance data set
    Claverie, Martin
    Ju, Junchang
    Masek, Jeffrey G.
    Dungan, Jennifer L.
    Vermote, Eric F.
    Roger, Jean-Claude
    Skakun, Sergii V.
    Justice, Christopher
    [J]. REMOTE SENSING OF ENVIRONMENT, 2018, 219 : 145 - 161
  • [7] Fisher A, 2019, J MACH LEARN RES, V20
  • [8] Seasonal Composite Landsat TM/ETM plus Images Using the Medoid (a Multi-Dimensional Median)
    Flood, Neil
    [J]. REMOTE SENSING, 2013, 5 (12): : 6481 - 6500
  • [9] Status of land cover classification accuracy assessment
    Foody, GM
    [J]. REMOTE SENSING OF ENVIRONMENT, 2002, 80 (01) : 185 - 201
  • [10] Phenology-adaptive pixel-based compositing using optical earth observation imagery
    Frantz, David
    Roeder, Achim
    Stellmes, Marion
    Hill, Joachim
    [J]. REMOTE SENSING OF ENVIRONMENT, 2017, 190 : 331 - 347