What does this research mean for the field?

XGBoost achieves the highest performance among machine-learning and deep-learning models for parcel-level crop classification in small, irregular fields using structured Sentinel-2 satellite image time-series data. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

Evaluate a parcel-level crop classification framework for small and irregular agricultural fields using structured Sentinel-2 satellite image time-series data.

August 23, 2026Scientific ReportsOpen Access

Parcel-level crop classification in small and irregular fields using structured Sentinel-2 data and clustering based analysis

View Full Paper

Ask AI

Bookmark

Authors

JGJaejun GouSeoul National University SJSeongju JangSeoul National University JPJinseok ParkSeoul National University

Discussion

Member takes

Implication

Comparative modeling study demonstrates effective parcel-level crop classification from structured satellite time-series in irregular farmlands, highlighting potential for regional agricultural...

Key Points

Evaluate a parcel-level crop classification framework for small and irregular agricultural fields using structured Sentinel-2 satellite image time-series data.
Extracted 6-month Sentinel-2 imagery structured into a 2D time-by-band matrix (10-day average reflectance) alongside drone-based ground truth surveys across four South Korean regions covering 20 crop types.
Evaluated ten machine-learning and deep-learning models, applying post-hoc SHAP analysis and K-means++ clustering to explore crop spectral-temporal characteristics and class similarities.
XGBoost achieved the highest classification performance with a Micro F1 of 0.7175, Macro F1 of 0.5474, and Kappa of 0.6746.
The CBAM deep-learning model achieved the second-highest performance with a Micro F1 of 0.6892, Macro F1 of 0.5512, and Kappa of 0.6470.
K-means++ clustering and SHAP analysis confirmed that spectro-temporal similarity during overlapping growing seasons directly accounts for classification errors among specific crops.

Cite This Study

Gou et al. (2026) studied this question.

synapsesocial.com/papers/6a8aad167677a341144455b4 https://doi.org/10.1038/s41598-026-68078-5

View Full Paper

Ask AI

Bookmark

Implication

Comparative modeling study demonstrates effective parcel-level crop classification from structured satellite time-series in irregular farmlands, highlighting potential for regional agricultural...

Parcel-level crop classification in small and irregular fields using structured Sentinel-2 data and clustering based analysis

Discussion

Member takes

Implication

Key Points

Cite This Study

Parcel-level crop classification in small and irregular fields using structured Sentinel-2 data and clustering based analysis

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider