The significance of sampling biases on calculated monthly mean oceanic surface heat fluxes

Abstract

The possible influence of spatial gradients and seasonal and diurnal cycles in basic meteorological parameters on the accuracy of monthly five degree latitude–longitude mean ocean surface flux calculations is explored. A simple formalism for assessing the importance of these possible biases is outlined. Data for two sample eastern Pacific regions shows possibly significant biases in one or more variables. The effect of the bias on the 20-year means of the fluxes would appear to be less than 1 % of the variance or about 10% of the mean. The influence of the biases on individual monthly means could be far greater depending upon the number and distribution of available observations. It is suggested that in tropical and subtropical regions free of tropical storms about eleven observations is the minimum number of observations to assure the accuracy of individual monthly fluxes. DOI: 10.1111/j.2153-3490.1981.tb01745.x