Dataset: Kelp Forest Estimated Carbonate Parameters

In order to estimate other carbonate parameters from sensor pH data, we require an additional carbonate parameter. To accomplish this, we created a local empirical relationship to estimate total alkalinity (TA) based on discrete samples, and combined this with sensor data to calculate DIC, pCO2, and aragonite saturation state (ΩAr).

TA was estimated using a multiple linear regression (MLR) approach, with temperature (T) and salinity (S) as inputs (Alin et al., 2012; Carter et al., 2018). This approach has been demonstrated to be effective in the open ocean (Carter et al., 2018) and along the California Coast below 50 m depth (Alin et al., 2012). Takeshita et al. (2015) demonstrated that this MLR approach can be extended to nearshore environments and showed TA could be estimated to ± 6 μmol kg-1 (RMSE) inside a kelp forest in Southern California by applying an offset to the equations from Alin et al. (2012).

Here, we fit an equation of the form used in Alin et al. (2012) to the discrete samples (n = 271) collected in the kelp forest, and obtained the following equation:

TA est = alpha0 + alpha1 (T - Tr) + alpha2 (S - Sr) + alpha3 (T-Tr) x (S - Sr)

with a RMSE of ± 7.3 μmol kg-1. Tr and Sr are the mean temperature and salinity for the deployment. The estimated total alkalinity (TAest) was combined with sensor pH to estimate DIC, pCO2, and ΩAr using equilibrium constants from Lueker et al. (2000) and CO2SYS (van Heuven et al., 2011). For a range of pH from 7.56 to 8.30 the standard uncertainty for estimated carbonate system parameters was 8.98 to 13.43 μmol kg-1 DIC, 54.22 to 9.42 μatm pCO2, and 0.0501 to 0.223 ΩAr. These uncertainties were calculated using the ‘errors’ function in seacarb (Gattuso et al., 2020, version 3.2.13 - Accessed 6 Mar 2020) with inputs of TAest and measured pH, temperature and salinity over depth inside the kelp forest.