Estimating total alkalinity for coastal ocean acidification monitoring at regional to continental scales in Australian coastal waters

Abstract

Abstract.Owing to a lack of resources, tools, and knowledge, the natural variability and distribution of Total Alkalinity (TA) has been poorly characterised in coastal waters globally, yet variability is known to be high in coastal regions due to the complex interactions of oceanographic, biotic, and terrestrially-influenced processes. This is a 5 particularly challenging task for the vast Australian coastline, however, it is also this vastness that demands attention in the face of ocean acidification (OA). Australian coastal waters have high biodiversity and endemism, and are home to large areas of coral reef, including the Great Barrier Reef, the largest coral reef system in the world. Ocean acidification threatens calcifying marine organisms by hindering calcification rates, Biogeosciences Discuss.,

Response to Anomynous Reviewer 2

“…In a region that exhibits the mixing of n end-members, a model with at least n-1 conservative explanatory variables must be employed in order to account for all conservative variability. Temperature is a conservative variable, and yes, it does not directly effect TA, but rather it indirectly is related to TA, when a cold water mass mixes with a warmer water mass, such as when upwelling occurs, which we know has seasonal variability and is not constant (Jiang et al 2014, Lee et al 2006, Millero et al 1998, Ingrosso et al 2016. Thus, the inclusion of two conservative parameters seems relevant to model the coastal region, which most often experiences the mixing of at least three water masses (river end-member, coastal end-member, offshore end-member, upwelling water mass).…”

“…By employing the use of AIC values we attempt to prevent over-fitting, as the calculation for AIC has a penalty term for the addition of explanatory variables. In statistical theory, this is equivalent to minimizing the cross-validation term (CV), a common term used to choose optimum models, and the term employed by Lee et al (2006). This is proven in Stone (1977).…”