\section{Fine Scale Surveys}\label{fine-scale-surveys}

This section will describe the manner in which surveys have been
conducted. Planning and executing these surveys in a controlled manner
ensures 100\% coverage of the area. The 12-hr MBARI operations limit for
ROV dives imposes a constraint on the survey coverage.

\subsection{Survey Design}\label{survey-design}

All surveys to date have been planned based on data collected previously
by MBARI's Dorado class mapping vehicles. These data, collected
typically by a 200-kHz multibeam sonar, yield maps with 1-m horizontal
resolution and 0.1-m vertical precision. From these maps, scientists at
MBARI identify areas of interest. In many instances, ROVs are used to
conduct video reconnaissance mission to validate these areas as targets
for the fine scale surveys. A site is primarily considered for these
specialized surveys because the area is either unique or likely to
experience a change. Following sections will discuss specific cases of
these targets.

Surveys are tailored to the site of interest, but generally follow a
``lawnmower'' style of motion, with the platform always moving forward,
turning before each line. Since the lighting around the cameras is
symmetrically spread, this has little impact on the stereo imaging data
when processed post-dive. In sites that are visited in repeated dives, a
simple marker may be dropped central to the survey area so that repeated
surveys can be more readily co-located. More optimally, a bottom feature
that is unlikely to move is identified near enough to serve the same
purpose as the marker.

Cross-track coverage is limited primarily by the field of view of the
lidar system, and as such lines are tightly spaced for only a small
amount of lidar overlap. The limiting factor to along-track speed is the
acquisition rates of the stereo cameras due to the recharge cycle of the
xenon strobe system.

\begin{figure*}[htbp]
\centering
\includegraphics[width=1.00000\textwidth]{./img/clams.png}
\caption{Three maps, collected simultaneously by multibeam sonar (left),
lidar (center) and stereo cameras (right). \label{fig:clams}}
\end{figure*}

\subsection{Station-Keeping Software}\label{station-keeping-software}

MBARI software engineers have collaborated with Stanford scientists to
aid in development of a station-keeping that we have employed for use in
these surveys. This software acts as a ``backseat driver'' to the ROV
control set, effectively replacing the joystick control with
computerized thruster control. What separates this software from similar
efforts \cite{Whitcomb98towardsprecision} is an interface that allows an
entire survey to be scripted by a set of relative movement instructions.
An origin within the survey is identified, and once the scripted survey
begins, all movements reference this point. Horizontal positions are
defined as waypoints. A constant altitude is specified, along with a
desired heading. Finally horizontal speed is defined so that along-track
resolution can be maintained.

These control features encompass a basic required of information to
conduct a rigorous survey at the desired scales of operation. Typically,
surveys are flown at 3-m altitude. Survey speeds are 0.2-m/s. Under
these parameters, a 100-m by 100-m box can be covered using 1.5-m line
spacing, and finished in approximately 10 hours of operation.

\subsection{Example Survey - Clam
Community}\label{example-survey---clam-community}

Once surveys are collected, the data are processed using MB System
\cite{mbsystem}. Within the Monterey Bay National Marine Sanctuary,
there is a site at 2850-m depth within the Monterey Canyon which houses
a community of chemosynthetic clams. Figure \ref{fig:clams} depicts a
survey conducted at this site. The three panes provide a comparison of
the capabilities of each system. On the left is the multibeam 5-cm
resolution data, and bathymetry is the main feature, with the clam
communities resolved as a texture.

The middle pane represents 1-cm resolution bathymetry collected using
the lidar. The clam community is better defined here, and individual
clams can be observed. The third pane is the photo-mosaic of the same
region. Clams are clearly visible, and provides ground-truthing to
targets we can identify and measure in the lidar. The map generated from
lidar data exhibits features that the multibeam sonar or the photography
have failed to capture, most notably the individual tracks left by the
clams as they move across the seafloor.