Or it can be configured with fixed cameras such as the installation shown below:

The Software: The software package included in the system is called VM95. It can be programmed for snap shots, averaged images, variance images and time stacks for any interval during any time
during day or night. Examples of these images are shown below:

VM95 can be set to use the ARGUS naming conventions for local storage and uploads.  ARGUS is a world-wide beach monitoring network developed and operated by the Coastal Imaging Laboratory (CIL) at Oregon State University, U.S.A (see Holman, R.A., and J. Stanley, The history, capabilities and future of Argus,  Coastal Engineering,  in review, 2005.)Other software features include the video server capability, complete remote control with fast video viewing, automatic creation of slide shows and movies, and the ability to upload to the Internet following any desired schedule.Digital Cameras:  Two consumer Olympus cameras and the digital SLR cameras from Canon can be used with the Biscuit system.  These cameras have enormous advantages over video cameras in that they provide much more detail (up to 20x) and much lower noise than video and high pixel count electronic cameras.  Canon SLR Cameras: Unlike the video cameras, the digital cameras cannot provide a stream of images, making them unsuitable for the average and variance computations until recently.. Due to the advent of gigabyte memory cards and faster electronics, useful average and variances images can be computed.  The SLR cameras from Canon can attain speeds of 3 images per second for a limited number of images (12 – 100 depending on model) and sustained rates from 0.7 -1.0 seconds per image (1 – 1.5 fps).  Images taken at these speeds are first all captured to the memory card, and then transferred to the PC.  Once in the PC, the average and variance images are computed and archived.  The process of retrieving and computing the images can take up to 10 minutes for a set of 100 images, depending on image size.  The Canon 350D is their most economical SLR camera..  It can take a burst of 12 images at 3 images per second and can shoot at a sustained rate of approximately 1 image per second.One advantage of SLR cameras is that they have interchangeable lenses.  This allows for the use of wide angle lenses yielding results far superior to those obtained with the consumer Olympus cameras (with built in lenses).  Also, image stability from shot to shot is better with SLR cameras than with the Olympus non SLR cameras, because the lens can be set in ‘manual’ mode and will never change.  Note that this means the zoom and focus cannot be changed through software.    Digital SLR cameras have a limited number of ‘snaps’ due to the mechanical shutter (video cameras use electronic shutters and don’t wear out).  This number appears to be anywhere from 100,000 to 500,000.  This needs to be kept in mind when designing the sampling schedule.   Consumer Olympus Cameras: In cases where image averaging and variance images are not required, but spatial coverage and detail is of high importance, the consumer Olympus cameras are more appropriate.  There are 2 models that are supported, the SP350 (8megs, 3x zoom) and the SP500 (6 megs, 10x zoom). VM95 software allows complete control of zoom, focus, and exposure. When combined with a pan tilt, they provide maximum spatial coverage: 360° horizontally and +/-90° vertically.  These cameras cost far less that the SLR cameras, yet provide extremely high quality imagery.  Sustained capture rates on these cameras is around 1 image every 4 seconds.  Field usage of the Biscuit System:   Examples of how scientist are using the Biscuit system are taken from three different groups of researchers: Jack Puleo at the Univ. of Delaware, Dan Hanes at the USGS, and Daniel Conley at the Saclant Research Center. Papers and publications are shown below for each of these groups:A University of Delaware study has yet to publish any results since their study just began, but the images that are being captured can be viewed on line at:http://sandcam.coastal.udel.edu

 
 Wave and current meter The underwater component of RESSORS is the in-situ wave and current meter (Figure 2). 

 Figure 2: Schematic diagram of the RESSORS wave-current meter during deployment.  The EMACS package which rests on the sea bottom contains batteries and the CPU unit for control of data collection from the current-pressure sensor, data analysis, and subsequent data transmission via Globalstar. This module is composed of a sensor, an EMACS unit and batteries housed in an underwater electronics package, and a sensor buoy which contains the satellite modem antenna.  The sensor used is a Nortek Vector 3-D velocimeter.  This sensor contains an acoustic-Doppler current velocimeter, as well as pressure, temperature, compass, and tilt sensors.  In typical operation, point measurements, which include 3 orthogonal components of geo-referenced fluid velocity, pressure and data quality flags, are collected for 12 minutes every collection period.  The approximately 4 Mbytes of data collected during this period are saved and subsequently processed in the EMACS unit. As a first processing step, the saved 8 Hz. data are cleaned as described by Elgar et al [2001a] and decimated to 2 Hz.  

In the subsequent step, mean quantities such as mean currents and water depth are calculated and the complete directional wave spectrum is estimated using a modified version of the DIWASP package [Johnson 2002].  The spectra estimation is performed utilizing the Iterated Maximum Likelihood Estimator [Pawka 1983].   Standard single value wave parameters such as significant wave height and peak period and direction are in-turn derived from the spectra.  In the processing step, the data stream is reduced by 3 orders of magnitude, passing from 4 Mbytes of data to less than a Kbyte.   The reduced data set is subsequently transmitted utilizing the satellite modem in the sensor buoy  Video Monitoring Package. The second component of RESSORS, the video monitoring system, is modeled after the ARGUS world-wide beach monitoring network developed and operated by the Coastal Imaging Laboratory (CIL) at Oregon State University, U.S.A [Holman & Stanley; 2005] and most of the software utilized for this component was generously provided by the CIL and their collaborators at NRL, Stennis Space Center. 

The camera system utilized in this version of RESSORS was developed by Erdman Video Systems (http://www.video-monitoring.com) and is composed of both a high resolution (2272 x 1704 pixel) digital still camera and a digital video camera (640 x 480 pixel) which can be sampled rapidly.  Both cameras are mounted on a fine-resolution pan-tilt device.  The camera is capable of collecting individual snap-shots, as well as long duration (>10 minutes) time-exposed images at data rates around 2 images.  The camera control software permits a remote operator to schedule image collection of specified types of images with desired camera orientation at arbitrary desired intervals.  Thus, for example, a collection schedule which assembles images, both time-exposed images and snapshots, spanning the entire beach, can be programmed to occur at sunrise, noon, and sunset every day.  The associated PC based controller performs camera control and image collection, creates the time-averaged images (thereby sharply reducing data transmission overhead) and transmits the resulting compressed images to the remote data collection center.  Due to the higher data transmission requirements of the video monitoring system, a Very Small Aperture Terminal (VSAT) satellite link was utilized to provide sufficient bandwidth for FTP of the images.  Two way satellite communication is available through this system thereby enabling operator intervention to request new data or to fine-tune existing image collection schedules. A real-time kinematic GPS survey is performed as part of the camera emplacement procedure.  This provides high accuracy (O(cm)) relative positions of several reference points in the camera field of view which are referred to as Ground Control Points (GCP).  The survey also serves to locate the absolute position of the camera installation.  This information is used to rectify and merge images from multiple views into a single geo-referenced image of the monitoring site.  Time exposed images of this type [Lippman & Holman; 1989] are used to extract additional information about morphological features of the submerged beach, dimensions of the surf-zone, and the location of nearshore currents.  Snapshot images also provide limited information about the direction of incident waves and breaker type.

Meteorological station The final RESSORS component is the beach meteorology unit in which a compact portable meteorological measurement station is paired with an EMACS unit and an Iridium satellite modem.    The onboard EMACS computer operates around the clock, sampling averaged temperature, relative humidity, barometric pressure, wind speed and direction every 30 seconds. Once per hour the data is compressed and transmitted to the remote data collection point via IRIDIUM.  Following successful transmission, the various components are reprogrammed for the subsequent cycle and the process repeats itself.    

3.  SYSTEM DEPLOYMENT & SAMPLE PRODUCTS As part of the MREA04 field trial which took place in Portuguese territorial waters during late March – early April 2004, RESSORS was deployed at Pinheiro da Cruz beach (Figure 3) from 4-10 April 2004.  Pinheiro da Cruz beach is a sandy-cliff stretch [Pires-Silva et al., 2002] bounded by the Sado Estuary to the north and Sines Harbor to the south.

Figure 3:  Diagram of the study area.  The shoreline is relatively straight and its orientation is approximately North-South.  The mean tide range is 2.75 m and a pronounced longshore bar at low tide is a persistent feature of the nearshore bathymetry.  The wave-current meter was deployed at 1.5 m depth at low tide on the landward flank of the first trough in the surf-zone.  The instrument was deployed with the velocity sensor element looking upwards and sensor is programmed to collect 12 minutes of data every two hours.  The data buoy cable was 12 m long and had a secondary sand screw mooring 7 m to the south of the sensor mooring in order to prevent cable fouling.  Following data analysis, the reduced data set of parametric values was transmitted to a remote data fusion center where automated computer scripts generated real-time products for the end user (Figure 4).

Figure 4:  Web figure which is automatically script generated in order to provide easy to assimilate end-user interface for wave-current sensor.  Present current and wave conditions are given graphically  (arrows) in the proper geographic location while a more complete past history of waves and currents is provided in graphs on the side bar.  A link for a text file containing the raw data in tabular form is also presented on the web page.   In order to provide additional measurements for sensor validation, two other pressure-current meters were deployed to the south of the RESSORS instrument (Figure 3).  These instruments were also Nortek velocimeters but they were deployed in an umbilical fashion in which cables for transmitting data and electrical power were strung between the instruments and a field laboratory.  These sensors were set to operate continuously at 8 Hz and data was recorded in the laboratory in 2 hour bursts.  The RESSORS sensor was labeled as sensor N4, the middle sensor was N6 and the southern most was N9. The video monitoring package was deployed on the bluff behind the beach at the location which provided the highest elevation with an unobstructed view of the beach.  This location was approximately 125 m to the East of N4 and 110 m to the north at an elevation of 20 above msl.  Pinheiro da Cruz is located outside the main footprint of the Eutelsat W10 satellite, and the satellite provider recommended a larger satellite dish (140 cm) then what is required in most other areas of Western Europe. The nominal bandwidth of the system was 256 kbps upload and 512 kbps download. Consistent tests however indicated significantly poorer performance, averaging 53 kbps up and 320 kbps download. This had a severe impact on the energy consumption characteristics of the system resulting in sharply reduced battery longevity. The video system was however quite stable and provided reliable connectivity throughout MREA04.  Sequences of 4 time-exposed images and 4 snapshots spanning the entire beach were programmed to be collected 3 times/day in the morning, at mid-day, and in the evening.   An example of a geo-referenced time exposed video image of  the study site which has been created from a 4 image sequence collected by the video monitoring system at low tide on 7 April 2004 is shown in Figure 5.  Sensor locations and the locations of objects used as GCPs have been plotted on the image.

Figure 5:  Geo-referenced image of the study area constructed by merging and  re-projecting 4 time exposed video images collected by the video monitoring system at 10:00 am on 7 April 2004.   The meteorological station was also placed on the bluff approximately 120 m E of N6 at an altitude of 20 m (sensor elevation 2 m above ground).  The station collected data continuously, calculated 30 s means and transmitted data to the data fusion center hourly.   Script generated web pages which were hosted at the data fusion center (Figure 6) provided real-time information for the end user.  This information can be used to assess local conditions, or as an input to local circulation models as well as NBC dispersion models.  Once again, data visualization products can be produced in the data fusion center and the data can be made available for utilization in simulation schemes.  

4.   DATA COMPARISON  The techniques utilized in the RESSORS system to measure nearshore environmental parameters have been utilized in myriad field and laboratory applications.  However the MREA04 exercise provided a unique opportunity to contrast the RESSORS implementation of these techniques against more traditional implementations as well as to observe the generality of single point sensors in the nearshore.  With that goal in mind, a plot of time series of tidal elevations and significant wave height for the RESSORS (N4) and the traditional sensors (N6 and N9) is presented in Figure 7

Figure 6:  Automatic script generated web graphic used to provide easy to assimilate end-user interface for the meteorological sensor.  Current weather conditions are provided in an easily accessible format at the top of the page and time series of previous observations are graphed below.  A link for a text file containing the raw data in tabular form is also presented on the web page.

Figure 7:  Plot of significant wave heights and mean sea-surface elevation changes as collected by all three wave current meters. The parameters for the traditional sensors have been calculated using 12 minute sections of the data stream which correspond to the collection periods of the RESSORS sensor.  As suggested by the figure, the rms difference in tidal elevations for all sensors is relatively low.  Using all 64 data points for which all sensors were submerged, an rms elevation difference among all sensors is 0.08 m which represents approximately 3% of the total tidal variation during the experiment.  This figure appears to be relatively independent of local water depth.  The significant wave height results show considerably more scatter and the mean rms variation in wave height estimation across all sensors is 0.10 m.  This figure represents over 15% of the mean measured wave height of 0.61 m.  It is however clear from the figure that the greatest differences occur during periods of lower water.   A comparison of the nearshore currents from the three sensors is presented in Figure 8. 

Figure 8:  Time series plots of (a.) mean water depth and significant wave height as collected by N4 and mean current vectors collected at N4 (b.), N6 (c.), and N9 (d.).  Arrows are relative to the shoreline so a vertical arrow pointing upwards represents a purely cross-shore current flowing to the west.  Examination of the figure suggests that there is little relation between the various currents measured by the different sensors.  This is supported quantitatively by the correlation analysis which is presented in Table 1.  

Figure 10:  Rectified and merged time exposed images of the MREA04 study site with current vectors from the 3 measurement sites superimposed.  The image in a). was collected at 10:00 on 7 April, 2004,  b). at 13:00 on 9 April 2004, and c). at 14:00 on 5 April, 2004.  The local depths at N4 were 1.7 m, 2.3 m, and 3.8 m respectively.

Flow at the northern most sensor (N4) indicates onshore flow from swash bores propagating shoreward over the crescentic bar.  Flow at the middle (N6) and southern (N9) show converging flow from return currents in the nearshore trough flowing seaward through the exit channel.      In Figure 10b it is seen that even at mid-tide (depth=2.3 m) the longshore current begins to dominate resulting in longshore coherence among all sensors even though there is still minor bathymetric influence on the landward components of the currents.  Finally, under high tide conditions (depth = 3.8 m) the residual currents observed in Figure 10c exhibit no apparent bathymetric influence. These observations provide additional support for the arguments presented in the previous paragraph.  Thus it appears clear that the RESSORS wave-current meter is providing reasonable measurements but that single point measurements of currents in the surf zone can not be expected to be representative when there is strong bathymetric control of flow conditions.   These observations also demonstrate how valuable RESSORS imagery can be to explain the results of other in-situ measurements.