POSTER SESSION 4
10:30–12:30, Thursday, September 17
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
ABSTRACT 1161 | POSTER 140
ESTIMATING NET COMMUNITY PRODUCTION FROM HYPERSPECTRAL REMOTE SENSING REFLECTANCE IN OPTICALLY COMPLEX WATERS
Over long timescales, Net Community Production (NCP) approximates export production, providing insight into whether a region exports carbon to the deep ocean or re-emits it to the atmosphere. Previous satellite-based NCP estimates have generally been limited to case I waters and often rely on intermediary products rather than direct optical observations. Here, we present preliminary results showing that NCP can be estimated directly from hyperspectral remote sensing reflectance (Rrs), with potential application across both case I and II waters.
We collected paired in situ NCP and hyperspectral Rrs observations from the southeastern United States during 2025–2026 using the R/V Savannah as a ship of opportunity, producing the most extensive spatiotemporal NCP dataset yet assembled for this region. These observations were used to train a nine-member ensemble neural network (eNN) to predict NCP from Rrs. Initial in situ validation showed strong performance (R² = 0.81; RMSE = 6.1 mmol O2 m–2 d–1; N≈5,500). Direct transfer of this model to PACE matchups showed limited applicability, highlighting the challenge of translating in situ optical relationships to satellite retrievals. However, a matchup-trained eNN produced encouraging preliminary performance (R² = 0.97; SD ≈ ±15 mmol O2 m–2 d–1; N≈40), though the small sample size warrants caution.
When applied to PACE imagery from the study region, the preliminary algorithm produced spatial patterns consistent with expected nearshore-offshore NCP gradients and began to reveal regional spatiotemporal variability. Together, these results suggest a promising pathway toward direct satellite-based estimation of NCP in optically complex coastal and open-ocean waters.
Benjamin Lowin*, University of Georgia, [email protected], https://orcid.org/0000-0001-5826-7216
Sara Rivero-Calle, University of Georgia, [email protected]
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
