CMTS Chlorophyll Validation

Is the interannual decline in CMTS (DMC-2) CTD-fluorescence top-50 m chlorophyll real, or an artifact of the instrumentation changing over the record?

Short answer: two separable things are true. Most of the apparent whole-record decline is an instrument artifact — fluorescence sensitivity differs ~5-fold between the CTDs used, and the raw record's −67% is dominated by the 2012/13 changeover. But a real decline did occur from 2009 to 2012: it is present in extracted water-bottle chlorophyll, within season, and does not involve the fluorometer at all.

FINDINGS.md is the summary of results. Methods, data sources and full caveats are in METHODS.md.

Run it

Open CMTS_chl_validation.Rproj in RStudio, then:

source("run_all.R")

Figures and tables land in output/. For a single shareable HTML, knit report.Rmd (or source("render_report.R")).

Packages: dplyr, tidyr, tibble, lubridate, ggplot2, mgcv, readxl, purrr; writexl for the archival export, rmarkdown for the report.

Layout

R/00_config.R    paths, parameters, shared helpers   <- edit paths here
R/01_load_data.R CTD / bottle / satellite loading, per-cast metrics
R/02_A2_*        bottle <-> CTD calibration, by instrument
R/03_B1_*        satellite seasonal + interannual GAM
R/05_C1_*        phaeopigment
R/06_C2_*        size fractions (with nesting QC)
R/08_seasonal_*  interannual change by season, per series
R/09_ctd_*       the bottle-anchored correction
R/10_bottlefire_* independent validation from Sea-Bird .bl fire records
R/11_instrument_* early vs recent window contrast, framed by instrument
R/12_export_*    full-column CTD workbook with units + metadata
R/99_summary.R   console summary
run_all.R        sources everything in order
report.Rmd       assembles it into one HTML

Module numbering has gaps — 04 and 07 were retired during the code audit. run_all.R is the authority on which modules run and in what order.

Data

Source data is not in this repository — it lives in OneDrive/Dropbox and is referenced by absolute path in R/00_config.R. Point PATHS at your copies:

Key	What
ctd	CTD_DMC_2_combined.csv — CTD profiles, 2007–2025 (downcast)
bottle	Uber Chl A 2026 combined.xlsx — extracted chl, 2009–2013 + 2020–2025
satellite	cmts_raw_1999_2025.csv — surface chl, 1998–2024
bottle_fires	bottle_fires_detailed.csv — from extract_bottle_fires.R

The main deliverable

output/CTD_corrected_full.xlsx — every column of the source CTD file plus the correction factors and corrected fluorescence, with units in the headers and metadata/notes sheets.

To apply the correction yourself: join output/CTD_fluor_correction_by_cast.csv to the CTD data by Date and multiply Fluorescence_mg_m3 by corr_best. Raw fluorescence must not be compared across instruments without it.