Optimizing Bandwidth and Pulse Duration to Improve Detectability for Counting Small Fish at Short Range

Hydroacoustic sampling of juvenile salmonids passing into surface flow outlets (SFOs) at Bonneville Dam (BON) Powerhouse 1 (B1) or at The Dalles Dam is challenging.  Outlets are 6-m wide, 0.5-2 m deep, with a small hydraulic capture zone over the hydraulic control structure.  Outlets were designed as ice-and-trash sluiceways to pass debris, which precludes echo integration to estimate passage.  In 1990s, sampling was based on up-looking split beams mounted 6-10 m below an outlet, so distal ends of acoustic beams sampled ≤ 1/6 of outlet width.  Hydroacoustic counts at B1 in 1996 did not correlate with counts from up-looking cameras on the control gate.  Sampling SFOs with a DIDSON acoustic camera in 2002 revealed that fish were not entrained until they passed over the hydraulic control structure.  This hydraulic capture zone was downstream of most of the sample volume of an up-looking split beam, and detected fish were not committed to passing.  Multibeam sampling was effective but cost prohibitive for all 10 SFOs at these dams.  The same was true for optical cameras that also have range limitations.  In 2002, opposing side-looking deployments of split beams aimed across the hydraulic capture zone showed promise.  Video showed that most smolts passed into outlets tail first and were detected in side aspect.  Hydroacoustic counts were linearly correlated with camera counts in spring.  In summer, the correlation was linear up to about 2000 fish/h, and then camera counts increased exponentially relative to acoustic counts.  Many fish in schools during daylight could not be resolved with 20,000 Hz bandwidth (BW) and 0.2ms pulse width (PW) settings.  Smolts did not school or pass as much at night so echo-trace counting was acceptable then.  Resolvable range between adjacent targets is a function of PW and speed of sound.  At 18°C, range resolution is about 296mm at PW=0.4ms, 148mm at PW=0.2ms, 58mm at PW=0.08ms, and 15mm at PW=0.02ms.  Echosounder modifications allowed us to select a range of BW (20-100kHz) and PW (0.4-0.02ms).  Images of schools passing SFOs indicated that smolt spacing <50% of fish length was rare, making reasonable the settings of PW=0.08-0.2ms in spring and 0.06-0.08ms in summer.  Tank testing showed that radical shortening of range resolution relative to fish length is undesirable because multiple echoes may come from a single fish oriented about 45° to the acoustic beam axis.  Hydroacoustic counts at PW=0.08 were highly correlated with DIDSON counts at one BON SFO.