Global Biogeochemical Cycles
Supporting Information for
Net community production and calcification from seven years of NOAA Station
Papa Mooring measurements
Andrea J. Fassbender,1,*,† Christopher L. Sabine,2 Meghan F. Cronin2
1
School of Oceanography, University of Washington, Seattle, Washington, USA
2
NOAA Pacific Marine Environmental Laboratory, Seattle, Washington, USA
†
Present address: NOAA Pacific Marine Environmental Laboratory, 7600 Sand Point
Way NE, Seattle, Washington, 98115, USA
*
andrea.fassbender@noaa.gov
Contents of this file
Text S1
Figures S1
Tables S1 to S2
Introduction
This document includes details about the assessment of errors for the budget analyses,
including information for variables that required data filling or more intensive error
computations.
1
Text S1. Error Assessment
S1.1 Continuity of Surface Salinity and Temperature
Errors in the temperature and salinity measurements are based primarily upon the
changes between pre- and post-deployment calibration drifts [Freitag et al., 1999]. In
addition, in situ calibrations are performed based upon gravitational stability
considerations [Freitag et al., 1999]. Numerous conductivity-temperature-depth sensors
have been deployed throughout the NOAA Station Papa mooring deployment history.
The largest instrumental errors for temperature and salinity sensors in the time series
history were applied to all deployments for simplicity. During time periods when surface
salinity and temperature measurements were missing, subsurface measurements (5 m, 10
m, and 20 m) were used to extrapolate to the surface. To determine an error for these
filled periods, daily climatologies of the difference between the subsurface and surface
measurements were developed from time periods when the surface sensors were
functioning. Matching year day climatology errors were used for each day that data were
filled.
Mixed layer depth errors were prescribed as one quarter of the vertical separation
between the sensors spanning the mixed layer depth. Higher spatial resolution glider
based estimates of the mixed layer depth have an error of ±2 m. Errors in the mixed layer
depth tendency rate (∂h ∂t-1) were calculated using standard propagation of errors,
assuming errors at each time step were not correlated.
S1.2 TA and DIC Estimates
Errors in calculated TA were derived using a Monte Carlo simulation in which the
salinity values and the linear regression coefficients were varied around their
uncertainties for 1000 computation iterations. The standard deviations of the 1000
resultant TA values at each time step were then used as the uncertainty estimates for TA.
Uncertainties for calculated DIC were determined using the same approach by varying
TA and in situ xCO2 values around their uncertainties. The standard deviations of the
1000 resultant DIC values at each time step, computed in CO2Sys, were then used as the
uncertainty estimates for DIC.
S1.3 Gas Exchange
The buoy CO2 and anemometer sensor errors were propagated through calculations
and when wind measurements were absent, gaps were filled with 0.25°, 6 hourly, crosscalibrated multi-platform ocean surface winds from NASA [Atlas et al., 2011]. To
calculate an error for these data fill periods, the mean difference between NASA winds
and buoy measured winds from April to December 2008 (spanning the two week gap in
anemometer data) was calculated. This difference was used as an estimate of the error in
the NASA winds used for filling. In addition to sensor errors that contribute to gas
exchange uncertainty, we include a 30% error for the gas transfer velocity after
Nightingale et al. [2000].
S1.4 Physical Processes
During periods when mixed-layer transport estimates were determined from ADCP
and current meter data, sensor errors were propagated through the averaging. During time
2
periods when in situ current data were missing and OSCAR surface currents were used to
fill in data gaps [Bonjean and Lagerloef, 2002], errors were based upon the difference
between the subsampled in situ data and OSCAR estimates computed over the
overlapping period for OSCAR and in situ currents.
Vertical velocity was prescribed a fixed 30% error and diffusivity errors were
computed for 5-day smoothed estimates through propagation of errors, following Cronin
et al. [2013]. For more detail on the calculation of diffusivity and its error at Station
Papa, see Cronin et al. [2015]. The standard error in the climatological monthly
diffusivity values were then computed as the root-mean-square of the errors for a given
month, reduced by the square root of the degrees of freedom, assumed here to be the
number of years in the time series for the given month.
Multiple linear regressions (MLR) were assessed using the stepwise function
in MATLAB R2014a (The MathWorks Inc., Natick, MA, 2000) to identify significant
correlations. The variance inflation factor (VIF) was used to assess coefficient error
inflation due collinearity between predictor variables where VIF>10 was used as a
threshold. Once ideal predictor variables were identified, the MATLAB robustfit function
was used to fit the MLR while decreasing the weight of potential outliers.
To estimate errors in the MLR derived monthly DIC and TA gradients, differences
between WOA 2013 optimally interpolated fields and the data used to construct the fields
within 10° latitude and 15° longitude of OSP were averaged at each depth level.
Approximately ~100 values went into each depth average, giving depth-dependent error
estimates for each predictor field in the OSP region. These errors along with standard
errors in the MLR regression coefficients were propagated through a Monte Carlo
simulation 1000 times to estimate the error in the predicted DIC and TA fields for each
month. The resultant errors were then propagated using standard methods through the
gradient and mixing computations.
S1.5 Evaporation and Precipitation
Horizontal and vertical salinity gradients were computed from WOA 2013 monthly
salinity fields. The method described in the previous section was used to determine depth
dependent errors in the WOA salinity field near OSP, which were then propagated
through the salinity mixing and EP calculations.
References:
Atlas, R., R. N. Hoffman, J. Ardizzone, S. M. Leidner, J. C. Jusem, D. K. Smith, and D.
Gombos (2011), A cross-calibrated, multiplatform ocean surface wind velocity
product for meteorological and oceanographic applications, Bull. Am. Meteorol.
Soc., 92(2), 157–174, doi:10.1175/2010BAMS2946.1.
Bonjean, F., and G. S. E. Lagerloef (2002), Diagnostic model and analysis of the surface
currents in the tropical Pacific Ocean, J. Phys. Oceanogr., 32, 2938–2954.
Freitag, H. P., M. E. McCarty, C. Nosse, R. Lukas, M. J. McPhaden, and M. F. Cronin
(1999), COARE Seacat data: Calibrations and quality control procedures, NOAA
3
Tech. Memo. ERL PMEL-115, NTIS: PB99-146144, NOAA/Pacific Marine
Environmental Laboratory, Seattle, WA, 88.
Nightingale, P. D., G. Malin, C. S. Law, A. J. Watson, S. Liss, I. Liddicoat, P. S. Liss, M.
I. Liddicoat, J. Boutin, and R. C. Upstill-Goddard (2000), In situ evaluation of
air‐sea gas exchange parameterizations using novel conservative and volatile tracers,
Global Biogeochem. Cycles, 14(1), 373–387.
Figure S1. Average monthly (a) DIC transformation rates for each process contributing
to the mixed layer DIC budget, including the sum (Total), and (b) mixed layer depth.
Month
1
κ
2 -1
m s
values
6.0E-04
κ
% Error
68
MLD
m
Error
5
SalEP
SalPhys
∂Sal ∂z-1
(Sala-Salh)/h
% Error
30
% Error
33
% Error
58
% Error
42
4
2
3
4
5
6
7
8
9
10
11
12
8.0E-04
1.5E-03
2.4E-03
6.0E-04
3.0E-04
1.0E-04
1.0E-04
1.0E-04
2.0E-04
3.0E-04
4.0E-04
69
100
39
48
24
18
16
17
19
21
27
5
5
4
3
2
2
2
2
3
5
5
32
41
39
37
67
33
19
15
16
18
22
33
38
37
36
68
57
45
37
34
31
30
95
100
100
100
100
100
43
100
100
94
100
46
83
100
100
100
100
35
100
100
88
100
Table S1. Diffusivity (κ) climatology and % errors for κ and each flux term in the
salinity budget as well as the vertical salinity gradients. Absolute errors given for the
mixed layer depth are prior to the monthly averaging of daily estimates.
5
6
Table S2. Errors for the calculated DIC and TA time series and % errors for each flux
term in the DIC and TA budgets as well as vertical gradients. For this table, µM is used
as shorthand for µmol kg-1. Errors given for DIC, TA, and DICGas are prior to the
monthly averaging of daily estimates.
7