Oklahoma-calibrated chlorophyll/Secchi coefficient set
Source:R/okBATHTUB-package.R
dot-oklahoma_coefficients.RdReturns Oklahoma-specific empirical chlorophyll-a and Secchi depth regression coefficients calibrated from publicly available state lake monitoring data, layered on top of Walker (1985, 1996) BATHTUB Model 1 nutrient retention. Ecoregion-specific regressions are applied where calibration support is sufficient (n >= 15 observations, n >= 5 lakes, R^2 >= 0.25); a statewide pooled regression is used otherwise.
Calibration data
Calibration used 1998-2023 growing-season (May-October) surface grab
samples from publicly available state lake monitoring data, aggregated
to lake-station-year means requiring at least 3 samples per parameter,
then filtered to records with valid TP, chlorophyll-a, and Secchi
values (joint filter). The same filtered dataset feeds both the
Chl-a-from-TP and Secchi-from-Chl-a regressions, so per-ecoregion
n values are identical across the two fits. The statewide pooled
total (n=186 observations from 48 lakes) is larger than the sum of
the per-ecoregion fits because the pooled fit includes records from
ecoregions where the per-ecoregion fit did not meet the sample size
threshold (e.g. Ouachita Mountains). See
data-raw/CALIBRATION_README.md and
data-raw/ok_calibration_coefficients_v0111.csv in the package source
for full provenance.
Ecoregions with ecoregion-specific fits
Cross Timbers (Chl-a and Secchi)
Central Great Plains (Chl-a and Secchi)
All other ecoregions use the Oklahoma statewide pooled coefficients.
As of v0.1.11, ecoregions follow the standard EPA Level III
(US_L3CODE / US_L3NAME) vocabulary, assigned by spatial join against
the bundled ecoregion polygons rather than the bespoke scheme used through
v0.1.10. Only Cross Timbers and Central Great Plains met the minimum
thresholds (>= 5 complete-data lakes, >= 15 observations, R^2 >= 0.25);
all others fall back to the statewide pooled fit. See
data-raw/CALIBRATION_README.md for full context.