Freshwater ecosystems support biodiversity and provide essential ecosystem services. In Korea, the Water Environment Information System monitors these ecosystems using separate biological and physicochemical indicators. Complex interactions occur among diverse biological taxa and physicochemical conditions. Thus, integrating heterogeneous monitoring data is crucial for accurately assessing ecosystem health. However, differences in data characteristics between the indicators present significant integration challenges. Given the scale and heterogeneity of the monitoring data, advanced analytical techniques are necessary to detect interactions among variables. This study aimed to identify key correlations among biological and physicochemical indicators by clustering similar variables and removing noise using the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) algorithm, followed by Spearman’s rank correlation coefficient and maximal information coefficient (MIC) analyses. HDBSCAN effectively eliminated noise indicators and grouped biological and physicochemical indicators into clusters based on shared characteristics, thereby enhancing the interpretability of the correlation analysis. Spearman analysis showed strong associations among biological indicators, particularly among species with similar ecological traits. MIC analysis further detected nonlinear associations between ecological assessment indices and specific biological species, which also reflected similar ecological characteristics. These findings are significant in that the comprehensive analysis of existing monitoring data revealed relationships within biological and physicochemical indicators while preserving the original purpose and function of each monitoring network. This study is expected to serve as a foundational resource for freshwater environmental monitoring and the development of effective management strategies.

Abstract
1. 서 론
2. 재료 및 방법
    2.1. 데이터 수집
    2.2. 데이터 전처리
    2.3. 비지도 클러스터링
    2.4. 상관관계 및 비선형 관계 분석
3. 결과 및 고찰
    3.1. 모니터링 데이터 전처리
    3.2. 지표 간 HDBSCAN 클러스터링 결과
    3.3. 지표 간 상관관계 분석 결과
    3.4. 지표 간 비선형 관계 분석 결과
    3.5. 시사점 및 제언
4. 결 론
적 요
Declaration of Competing Interest
사 사
REFERENCES

• 문민호(상명대학교 융 합공과대학 생명공학전공) | Min-Ho Mun (Department of Biotechnology, Sangmyung University, Seoul 03016, Republic of Korea)
• 안형은(상명대학교 융 합공과대학 생명공학전공) | Hyung-Eun An (Department of Biotechnology, Sangmyung University, Seoul 03016, Republic of Korea)
• 백종원(상명대학교 융 합공과대학 생명공학전공) | Jong-Won Baek (Department of Biotechnology, Sangmyung University, Seoul 03016, Republic of Korea)
• 한승민(상명대학교 융 합공과대학 생명공학전공) | Seung-Min Han (Department of Biotechnology, Sangmyung University, Seoul 03016, Republic of Korea)
• 김성욱(상명대학교 융 합공과대학 생명공학전공) | Sung-Wook Kim (Department of Biotechnology, Sangmyung University, Seoul 03016, Republic of Korea)
• 김창배(상명대학교 융 합공과대학 생명공학전공) | Chang-Bae Kim (Department of Biotechnology, Sangmyung University, Seoul 03016, Republic of Korea) Corresponding author
• 이미경(상명대학교 융 합공과대학 휴먼지능정보공학전공) | Meijing Li (Human Centered AI Major, Sangmyung University, Seoul 03016, Republic of Korea)
• 김동근(상명대학교 융 합공과대학 휴먼지능정보공학전공) | Dong-Keun Kim (Human Centered AI Major, Sangmyung University, Seoul 03016, Republic of Korea)