\documentclass[11pt]{article}
\usepackage[a4paper,margin=2cm,landscape]{geometry}
\usepackage{amsmath,amssymb,booktabs}
\usepackage[labelfont=bf,labelsep=space,justification=raggedright,singlelinecheck=false]{caption}
\newcommand{\cy}{\overline{c}^{\,y}}
\newcommand{\cmax}{c_{\max}}
\newcommand{\stk}[2]{\begin{tabular}[c]{@{}c@{}}#1\\{\scriptsize$[#2]$}\end{tabular}}
\pagestyle{empty}
\begin{document}
\setcounter{table}{3}
\begin{table}[h]\centering\footnotesize
\caption{Classical control for convective Prairie Grass (20 runs, 100 pairs, $\cy$): model with $\alpha=1$ and $K_z$ multiplied by a constant factor $f$, compared with the model with $\alpha=0.91$ and the original $K_z$. Columns 50--800~m: geometric mean of $c_p/c_o$ per arc. Last column: paired difference in FA2 ($\alpha=0.91$ minus the row), with 95\% block-bootstrap interval (10\,000 resamples, the experiment as the unit). Reference mesh $640\times160$.}\label{tab:control}
\setlength{\tabcolsep}{2.5pt}
\begin{tabular}{lccccccccccc}
\toprule
configuration & NMSE & Cor & FA2 & FB & FS & 50~m & 100~m & 200~m & 400~m & 800~m & \begin{tabular}[c]{@{}c@{}}paired\\$\Delta$FA2\end{tabular}\\
\midrule
$f=1.00$ & 0.076 & 0.978 & 0.72 & $-0.15$ & 0.14 & 0.94 & 1.24 & 1.62 & 2.15 & 3.26 & \stk{$+0.22$}{+0.14;+0.30}\\[2pt]
$f=1.20$ & 0.077 & 0.983 & 0.80 & $-0.01$ & 0.23 & 0.85 & 1.07 & 1.34 & 1.74 & 2.63 & \stk{$+0.14$}{+0.06;+0.23}\\[2pt]
$f=1.35$ & 0.116 & 0.986 & 0.84 & $+0.08$ & 0.30 & 0.80 & 0.97 & 1.18 & 1.52 & 2.28 & \stk{$+0.10$}{+0.03;+0.17}\\[2pt]
$f=1.50$ & 0.180 & 0.987 & 0.90 & $+0.17$ & 0.36 & 0.75 & 0.88 & 1.05 & 1.34 & 2.02 & \stk{$+0.04$}{0.00;+0.09}\\[2pt]
$f=1.70$ & 0.300 & 0.987 & 0.90 & $+0.27$ & 0.45 & 0.69 & 0.78 & 0.92 & 1.16 & 1.74 & \stk{$+0.04$}{0.00;+0.09}\\[2pt]
$f=1.90$ & 0.451 & 0.986 & 0.92 & $+0.37$ & 0.52 & 0.63 & 0.70 & 0.81 & 1.01 & 1.53 & \stk{$+0.02$}{-0.03;+0.07}\\[2pt]
\midrule
$\alpha=0.91$, $f=1$ & 0.104 & 0.975 & 0.94 & $+0.08$ & 0.22 & 0.81 & 1.06 & 1.19 & 1.14 & 0.91 & ---\\
\bottomrule
\end{tabular}
\end{table}
\end{document}
