RIVET I 2012 Data Collections Coastal Imaging Lab
Oregon State University April 27 to May 20, 2012
CIL RIVET Collections RIVET required installation of a new Argus station in late April on the top of a 106’ high jack-‐up tower installed on the SW side of NRI. This document describes the Argus collection system and the data collected as well as how and where to access the data. Argus Cameras The Argus cameras are located in the jack-‐up tower at roughly xyz = [-‐169, -‐596, 34] in FRF coordinates. Six cameras allowed a roughly 200° field of view from alongshore to the SW to slightly into the inlet to the NE. Numbers and orientations of cameras are shown in Figure 1 below on a background image that is a merged, rectified time exposure image of the region in the RIVET coordinate system. Images can either be used individually or as merged products such as Figure 1.
Figure 1. Merged snapshot showing the fields of view of the six primary cameras. Cameras geometry control was established based on a set of ground control points (GCPs) whose locations were surveyed. Geometries were established at the beginning of the experiment and were checked qualitatively, especially to determine if the aim points changed when the tower was lowered then raised again in
C1
C2 C3
C4
C5
C6
response to impending storms or for camera maintenance. Camera resolution varies with distance from the camera as show in Figure 2. Time is synched to an NTP server continuously.
Figure 2. Local resolution maps in x and y. Data Locations and Naming Conventions Data are stored on the CIL ftp site, cil-‐ftp.coas.oregonstate.edu and can be downloaded using normal anonymous ftp (with a valid RFC822 format email address as the password). Data are sorted by site/year/camera/day/file. For example, images for C3 of April 29, 2012 would be located in /ftp/pub/rosie/2012/c3/120_Apr.29. Argus filenames using a long (and useful) naming convention. For example:
1335740400.Sun.Apr.29_23_00_00.GMT.2012.rosie.c3.timex.jpg is a time exposure image for C3 collected on April 29, 2012 at 23:00:00 GMT. All Argus data are saved in GMT (because we have stations in many time zones). Note that images also have timestamp and other information imprinted on the top and bottom borders. These are written in local standard time. Be careful to not confuse these (image stamp in EST, filename in GMT = EST+5). Also note that we never use daylight time, so you need to correct from GMT to daylight by subtracting 4 hours. Our primary time reference is epoch time, a computer standard that is the number of seconds since January 1, 1970 (GMT). This is the 10-‐digit leading number in the
filename. Routines to convert between epoch to matlab’s datenum are located in /ftp/pub/Experiments/SZO2010/commonMatlabCode. Five types of images are collected:
• Snapshot • Timex – average of 2Hz frames collected over ten-‐minute period • Var – standard deviation of same image sample • Bright – brightest that each pixel gets over the same period • Dark -‐ darkest that each pixel gets over the same period
In addition, daytimex images average all of the timex images for a day. Bright images can be useful for showing the map of where any wave broke during the 10 minute collect. Merged images like Figure 1, are automatically processed and are stored as camera cx using the same naming conventions but with an additional suffix ‘merge’. For example, 1335740402.Sun.Apr.29_23_00_02.GMT.2012.rosie.c3.bright.merge.png is a merged brightest image from April 29, 2300 GMT (1800 EST or 1900 EDT). These are png images and are saved in two sizes: merge images that span from x = -‐500: 1000, and y = -‐1300: 1000, so cover the entire inlet but at fuzzy resolution at large distances, and mergeSmall images that span from x = -‐500:500 and y = -‐1200:+100 at much better resolution. smallMerge images cover the south ebb tidal delta, western beach and part of the inlet. Note that the merged image data are also stored as .mat files so you can create your own figures. Pixel Time Series While image data synthesize data from a 10-‐minute collect into single images, pixel time series data sample and save the 2Hz time variability for 2048 samples (17+ minutes), but only for specific requested pixels. Data are generally called time stacks and files are saved under each camera as raster files with the suffix ‘.stack.ras’ (usually gzip compressed, so with an additional .gz suffix). Retrieval of data is somewhat complicated, but has been automated for RIVET I and results are stored under cx. For example, the file 1337471940.Sat,May.19_23_59:00.GMT.2012.rosie.cx.vbar2.mat is a matlab data file from May 19 that contains the pixel instrument vbar2 (a close-‐packed array of pixels from the middle of the inlet designed to allow estimation of currents – originally designed for longshore currents, hence the instrument name). All pixel instrument mat files contain the following variables
• T -‐ epoch time for each of the N samples in the time series • XYZ -‐ RIVET coordinates for each of the M pixels (z usually =0) • RAW -‐ N by M time stack of raw camera intensities • CAM -‐ camera number associated with each pixel (length M) • GAIN -‐ N gain values (for each frame) (usually constant for run) • SHUTTER -‐ N shutter exposure times
• CORRECTED -‐ image brightness corrected for gain and shutter. Corrected intensities are RAW divided by 10^(gain/20) * shutter, reversing the camera compensation for changing lighting conditions during the day. Pixel instrument collections during RIVET were of two kinds, cBathy arrays and vBar vector collections. Sampling evolved through the experiment as we tried different concepts, so some instruments were only collected during the latter part of sampling. cBathy arrays use the instrument name mBWx, where x = 1:4 (mBW is a historical name indicating a matrix collect (the ‘m’) of type Beach Wizard (BW)). Figure 3 shows the cBathy pixel array with dot symbols, hard to distinguish because they are so dense. The blue mBW1 pixels closest to the camera (the white partial circle at the top of the figure) have a spacing of 2.5 x 2.5 m, commensurate with the better pixel resolution at those short ranges, whereas the more distant pixels (mBW2-‐ red, mBW3-‐green, mBW4-‐black) have a 5 x 5 pixels spacing.
Figure 3. Pixel arrays for the cBathy array showing mBW1 (blue dots, closest to the cameras at the center top), mBW2 (red, toward the inlet channel), mBW 3 (green, the offshore pixels, and mBW4 (black to the SW covering the part of the open beach).
There are eventually four vbar arrays, one along the channel immediately offshore of the cameras (the site of the triPole and Jim Thomson’s met and in-‐situ fixed instruments) and three parallel transects down the throat of the inlet (Figure 4).
Figure 4. Merged snapshot image (camera boundaries apparent due to gain differences between cameras) and the analysis locations of vbar pixel time stacks along the main inlet channel. The red, green and blue transects correspond to instrument names vbar 2, 3 and 4, respectively while the experimental instrument in front of the cameras is vbar1. Note that the vbar pixel instruments have a dense spacing along transect (as close as is possible) while the cBathy array is more sparsely spaced but is usually dense enough for most purposes. Time series data from either array can be used as optical instruments to compare to any other data source. Both cBathy and vbar arrays were collected for research purposes to investigate new methods (under non-‐traditional conditions). Thus, production analyses are not current available for these instruments.
Collection Schedules Argus data were collected every half hour during RIVET I. Collection start times are staggered slightly due to a constraint in our archiving protocols. For each daylight hour, collections were as follows:
Minute after hour Collection 00, 30 image products (snap, timex, etc) 59, 29 pixel time stack data, mBW and vbar
The collection clock is synched continuously with an NTP time server. All collections should start at the top of the minute, but make have order ½ second delays. However, epoch times listed in files will always be correct. Matlab Files Several useful matlab files are located on /ftp/pub/Experiments/SZO2010/commonMatlabCode. These include epoch to matlab conversion and a short example routine to load and use an instrument. Problems: Contact Rob Holman, [email protected].