Evaluating the cross-lake transferability limits of machine learning models for Sentinel-3 inland water chlorophyll-a retrieval

Chudi Wu

doi:10.1016/j.ecoinf.2026.103969

Evaluating the cross-lake transferability limits of machine learning models for Sentinel-3 inland water chlorophyll-a retrieval

2026-08-04

0

OA

AI

C

Chudi Wu

Z

Zhi Chen *

DOI：10.1016/j.ecoinf.2026.103969

Original

Original request for help

Save

Abstract

En 中文

• Sentinel-3 OLCI machine learning Chla models were tested on a held-out target lake. • Class-specific gains were method-dependent and did not transfer to Lake Huron. • Failures persisted for in-domain, spectrally matched Lake Huron samples. • Shifts in OAC covariation and spectral ambiguity explain the transfer limit. • Cross-lake retrieval needs domain adaptation, not pooled training alone.

Keywords:

Chlorophyll-a

Sentinel-3 OLCI

Machine learning

Optical water type

Cross-lake transferability

Evaluating the cross-lake transferability limits of machine learning models for Sentinel-3 inland water chlorophyll-a retrieval

Abstract

Journal

Organization

Cited Papers

Citing Papers