|
ADMINISTRATIVE COMMITTEE |
||||
|
|
||||
|
5. |
CONSIDER EXPENDITURE OF BUDGETED
FUNDS TO PURCHASE HARDWARE AND SOFTWARE FOR THE VIDEO SURVEILLANCE SYSTEM AT
THE |
|||
|
|
||||
|
Meeting Date: |
October 4, 2007 |
Budgeted: |
FY 2007-08 |
|
|
|
||||
|
From: |
David A. Berger, |
Program |
Capital Asset Purchases |
|
|
|
General Manager |
Line Item No.: |
Account No. 24-9130 |
|
|
|
||||
|
Prepared By: |
|
Cost Estimate: |
$ 12,235 |
|
|
|
Kevan Urquhart/ |
|
|
|
|
|
Inder-Mohan S. Osahan |
|
|
|
|
|
||||
General Counsel Approval: N/A |
||||
Committee Recommendation: The Administrative Committee considered this item on October 4, 2007 and recommended ____________. |
||||
|
CEQA Compliance: N/A |
||||
SUMMARY: The automatic counter at the San Clemente Dam fish ladder counts fish that slide over the metallic comb connected to a data logger. The counter’s mechanism requires that a fish touch the counter to be able to make it move, which then triggers a switch to increment the count. It appears that every year some fish have jumped over the counter, or local mammals have triggered false counts. Staff have observed this in person, and on video. To have a method of double checking the fish counts at the automatic counter, staff installed a video camera to digitally record all activity at the fish counter during the migration season. There is not enough staff time to review all of the video footage collected each year. The Board will consider this request to improve the accuracy and quality control of steelhead counts at the San Clemente Dam fish ladder by expending the requested funds to automate the analysis of the video footage collected annually. The benefits of doing so include:
1) Accounting for steelhead
that jump over the mechanical counter, and would otherwise not be counted.
2) Correcting for
false positives due to other animals or debris triggering the counter.
3) Gathering data on
the type and condition of steelhead passing the dam.
4) Preventing data
loss during times when the mechanical counter is broken or malfunctioning.
5) Extending the
seasonal duration of passage counts when flow levels are too low to operate the
mechanical counter.
RECOMMENDATION: Staff recommends that the Board approve expenditure of the funds in the not-to-exceed amount of $13,485 to purchase and install the new video processing system and software. Of this amount, $1,250 will be reimbursed through a donation by the Carmel River Watershed Conservancy, for a net cost of $12,235. The Administrative Committee considered this matter at its October 4, 2007 meeting and voted _____ to recommend _________.
BACKGROUND: On February 24, 1994, the Monterey Peninsula Water Management District (MPWMD) completed installation of an automated fish counting device on the San Clemente Dam Fish Ladder. Over time, MPWMD staff observed that adult fish could jump over the counter, without triggering the counting device.
In January 2004, a digital surveillance video camera was installed in conjunction with the fish counter. The primary purpose of the camera was to act as a quality control for the automated fish counter. The video camera records continuously for about three to six months, typically December 15th through April 15th of each year. The data are digitally archived to an external hard drive attached to the system at San Clemente Dam. Any fish that crossed the counter, whether it triggered the counter to increment the count or jumped over it would be caught on digital video. These data enable staff to get a complete and more accurate count of the fish migrating past the San Clemente Dam. In addition to providing an accurate count of fish jumping over the counter (Exhibit 5-B), the video also enables staff to get a visual observation of every fish passing the counter. This will help determine if the fish are resident or sea-run, the sex of the fish, size, and general health condition. Review of the video also helps eliminate false positive counts from the counter that are due to local mammals visiting the device and triggering it. We found eight missed fish and four false positives by reviewing less than 12% of the video for 2007. Though this is not a statistically valid, random subsample of the video from 2007, if you apply these numbers to the rest of the as yet unviewed video, we may have had as much as a 33 fish error in the count for 2007. This could amount to being off on the annual estimate by approximately 15%. The funds we are requesting to automate analysis of the video will save many days of staff time, and enable us to efficiently and accurately review all of the video footage collected for the presence of fish passing up the San Clemente Dam Ladder.
In addition, during low flow years we are forced to remove the mechanical counting device to avoid impeding the final weeks of the spawning immigration. In 2007 this forced us to shut down ladder counts on April 3rd, instead of continuing the counts for as long as they were needed. The digital video camera also serves as a duplicate counter during periods when the primary mechanical device fails. This duplication allowed us to account for fish passage counts during the week in 2007 when the mechanical counter malfunctioned, yet we were still able to detect and enumerate 16 steelhead passing San Clemente Dam that would have otherwise been missed. However, this last effort alone cost us approximately three weeks of staff time to manually review the digital video recordings for just that week.
The video recorder utilized since January 2007 has collected about 2 TB (terabytes) of video per year. To ensure seeing any missed fish jump over the mechanical counter in the fish ladder, staff has to view the entire period of video recorded. This is an excessive amount of staff time that we cannot afford to allocate, equivalent to posting a person at the counter 24/7 for three to six months. It is also visually tiring for staff to keep on staring at the same picture for hours at a time, undermining the accuracy of that approach.
In order to find a way to automatically post-process the video footage collected, staff worked with Ojo Technologies and came up with a software solution that would automate this process using a video-server. The video collected on archive media will be filtered using the Object Video software which captures only the frames that have fish jumping over the counter. The software is similar to face recognition software used for surveillance at airports and other public places. The software vendor has modified the software to accommodate the MPWMD’s needs. Staff has seen a demonstration of the software and would like to deploy it this year. Staff did try and view some of the existing video themselves, but it is not an efficient use of staff time, nor is it feasible to allocate between 200 – 400 days of our limited staff resources to this sole purpose, each year.
Improving the accuracy of the
fish counts may eventually aid in the de-listing of steelhead in the
As requested by the Administrative Committee in July 2007 contacts were made with the National Marine Fisheries Service [Santa Rosa Office] (NMFS), Department of Fish and Game’s Central Region in Fresno (DFG-CR), Carmel River Steelhead Association, Carmel River Watershed Conservancy (CRWC), and California American Water (CAW) to request that these groups provide letters of support and preferably co-funding for this project. So far, we have not received a formal response to our request for funds from anyone but the Carmel River Watershed Conservancy (Exhibit 5-C), but will likely receive letters of support from CAW, DFG-CR, and NMFS.
5-A Fish Ladder Video Server Components
5-B Pictures of Fish Jumping Over the Mechanical Counter
5-C CRWC Letter of Support and Donation
U:\staff\word\committees\Admin\2007\20071004\05\item5.doc