韓国全国の洪水感受性マッピングにおける機械学習アルゴリズムの比較：未学習都市への空間的一般化
Comparing Machine Learning Algorithms for Nationwide Flood Susceptibility Mapping in South Korea: Spatial Generalization to an Unseen City (原題)
Bong-gyu Lee, K. Choo, Seungcheol Choi, Byung-sik Kim
🤖 gxceed AI 要約
日本語
韓国全国の100mグリッド洪水感受性マップを構築し、CatBoost・XGBoost・パッチ型CNNの空間的一般化性能を同一条件で比較した研究。釜山市を訓練から完全に除外し外部評価セットとした点が特徴的で、ROC-AUCは3モデルとも0.831〜0.835と同程度だが、CNNがPR-AUC・F1で最も高く偽陽性を抑制した。クラス不均衡下では複数指標の併読が必要と結論づけている。
English
Builds a nationwide 100 m flood susceptibility map for South Korea and compares CatBoost, XGBoost, and a patch-based CNN under identical inputs and spatial partitioning. Busan is fully held out as a geographic external test set; ROC-AUC is similar (0.831–0.835) but the CNN achieves the highest PR-AUC and F1, reducing false positives. Argues ROC-AUC, PR-AUC, and threshold metrics must be read together under class imbalance and unseen-city generalization.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本でも気候変動適応法やTCFDの物理的リスク評価において、洪水ハザードの高解像度マッピングは企業の拠点リスク開示と直結する。未学習地域への汎化性能の比較は、日本企業が外部ベンダーの洪水リスクモデルを選定・検証する際の評価軸として参考になる。
In the global GX context
Physical climate risk assessment under TCFD/ISSB increasingly relies on geospatial flood models, and this paper's emphasis on out-of-sample spatial generalization speaks directly to model validation standards for disclosure-grade hazard data. The metric-comparison lesson (ROC-AUC vs PR-AUC under imbalance) is relevant to how climate-risk vendors are benchmarked globally.
👥 読者別の含意
🔬研究者:空間ブロック分割と未学習都市での外部評価という設計は、気候リスクMLの汎化評価プロトコルとして参照価値がある。
🏢実務担当者:拠点の洪水リスク評価に外部モデルを採用する際、PR-AUCや未学習地域での性能を確認すべきという実務的示唆を与える。
🏛政策担当者:国レベルの洪水ハザードマップ整備において、機械学習モデルの外部検証と指標選択の枠組みづくりに示唆を与える。
📄 Abstract（原文）
Increasing heavy rainfall associated with climate change and the continued advance of urbanization have heightened the importance of spatially identifying flood-prone areas and responding proactively. This study constructed a nationwide 100 m grid flood susceptibility map and compared the spatial generalization performance of CatBoost, XGBoost, and a patch-based CNN under the same input data and spatial partitioning. Flood labels were constructed from 38,003 flood-trace polygons collected through the National Disaster and Safety Portal API. Eleven spatial variables related to topography, hydrology, land cover, vegetation, soil, and geology were used as inputs following quality inspection, correction, and reconstruction. Busan Metropolitan City was excluded entirely from training, early stopping, and model selection, and its 65,727 valid grid cells (natural flood rate, 3.50%) were used as a geographic external evaluation set. In non-Busan areas, 16 km spatial blocks and a 1.6 km buffer were applied to prevent spatial overlap between training and evaluation patches. In the Busan external evaluation, the ROC-AUC values of the three models were similar, ranging from 0.831 to 0.835, whereas the CNN achieved the highest PR-AUC (0.142) and F1-score (0.204)—4.1 times the random-prediction baseline, compared with 2.7 and 2.8 for CatBoost and XGBoost—indicating a reduction in false positives while comparable recall was maintained. The relative advantage of the models varied by evaluation metric, indicating the need to interpret ROC-AUC, PR-AUC, and fixed-threshold metrics together under conditions of natural class imbalance and when evaluating on a city unseen during training.
🔗 Provenance — このレコードを発見したソース
- semanticscholar https://doi.org/10.3390/w18212715first seen 2026-11-04 05:44:13 · last seen 2026-11-05 06:09:15
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録（無料・3キーワードまで）。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。