Non-parametric regression modelling of in situ fCO2 in the Southern Ocean

[摘要] ENGLISH ABSTRACT: The Southern Ocean is a complex system, where the relationship between CO2concentrations and its drivers varies intra- and inter-annually. Due to the lackof readily available in situ data in the Southern Ocean, a model approachwas required which could predict the CO2 concentration proxy variable, fCO2.This must be done using predictor variables available via remote measurementsto ensure the usefulness of the model in the future. These predictorvariables were sea surface temperature, log transformed chlorophyll-a concentration,mixed layer depth and at a later stage altimetry. Initial exploratoryanalysis indicated that a non-parametric approach to the model should betaken. A parametric multiple linear regression model was developed to use asa comparison to previous studies in the North Atlantic Ocean as well as tocompare with the results of the non-parametric approach. A non-parametrickernel regression model was then used to predict fCO2 andnally a combinationof the parametric and non-parametric regression models was developed,referred to as the mixed regression model. The results indicated, as expectedfrom exploratory analyses, that the non-parametric approach produced moreaccurate estimates based on an independent test data set. These more accurateestimates, however, were coupled withzeroestimates, caused by thecurse of dimensionality. It was also found that the inclusion of salinity (notavailable remotely) improved the model and therefore altimetry was chosento attempt to capture this e ect in the model. The mixed model displayedreduced errors as well as removing thezeroestimates and hence reducingthe variance of the error rates. The results indicated that the mixed modelis the best approach to use to predict fCO2 in the Southern Ocean and thataltimetry's inclusion did improve the prediction accuracy.

[发布机构] Stellenbosch University