source: altifloat/doc/ocean_modelling/review/review.tex @ 219

\documentclass[a4paper,12pt]{article}

\usepackage[T1]{fontenc}
\usepackage{amsmath,amssymb,amsthm}
\usepackage{dsfont}\let\mathbb\mathds
\usepackage[english,frenchb]{babel}
\usepackage[all,dvips]{xy}
\usepackage[utf8]{inputenc}


\title{Answers to reviewers}
\author{Leila Issa \and Julien Brajard}


\begin{document}
\maketitle
\section{Reviewer 1}
\textit{Line 114: I see you included a footnote for NEMED, but you should also define this acronym in
the text. An expansion of this paragraph is needed; what are these 2 projects and what types of
data came out of them?}

Details about the two projects were added:
NEMED(Surface Circuclation in the NorthEastern MEDiteranean).
NEMED is an observational program to quantify the circulation and eddy variability in the Eastern Mediterranean Sea (EMS), with major attention to the eastern and northern regions of the Levantine sub-basin, using both in-situ and remotely sensed data.  Altifloat is a project gathering several mediterranean countries (France, Lebanon, Italy, Cyprus) to create a network of scientists interested in studying the surface circulation in Eastern Mediterrranean Sea. The project relies on lagrangian observations (drifter deployment) and methodological developments.\\

\textit{Line 152: Define the acronym CYCOFOS\_CYCOM.}

acronym were defined:
CYCOFOS (CYprus Coastal Ocean Forecasting and Observing System) - CYCOM (CYprus Coastal Ocean Model)\\


\textit{Line 209: Can the authors please provide more details in how the covariance is constructed? The diffusion filter method is used to model the correlation, but how do you define the model errors? Also, I am not seeing in eq. 9 how observation errors are implemented.}

Details were added on this subject:
The weights $\alpha_1$ and $\alpha_2$ correpond to the confidence given to each term in the cost function relatively to the observation error term. As the cost function is defined within a multiplicative constante, it is safe to set a standard confidence of 1 to the observation error term in the cost function, as long as we assumed that each observation errors is equal and decorrelated from the other. The relative confidence in the background $\alpha_1$ and to the non-divergent constraint $\alpha_2$ cannot be objectively determined. So they are determined empirically using twin experiments by chosing the weights that give the lower error a posteriori.\\

\textit{Line 213: How are $\alpha_1$ and $alpha_2$ determined?}

This was answered previously.\\

\textit{Lines 397-400: Is the simulated trajectory shown in figure 13? What are the mean and maximum position errors? Is this the difference of the center of the eddy between the background and corrected solutions? Please elaborate.
}

It is said in the text (section 5.2 L.????) that the maximum position error is 6.7 km with a mean of 0.96 km. It corresponds mainly to the background term in the cost function, and so, the error is not essentially on the position of the eddy but more on its dimension and intensity (as it shown in fig. 14).\\


\end{document}