662 



Fishery Bulletin 102(4) 



meteorology and, incidentally, nighttime light sources 

 (Croft, 1978; Elvidge et al., 1997, 2001b). Nighttime 

 light detection by satellites has proven useful for vari- 

 ous environmental questions, such as identifying the 

 extent of forest fires (Elvidge et al., 2001a) and the 

 effects of urban lighting on sea turtle nest selection 

 and hatchling survivorship (Salmon et al., 2000). Com- 

 pilation of light data for the global light-fishing squid 

 fleet has contributed to examinations of the fishery's 

 ecosystem impacts (Rodhouse et al., 2001). For local 

 squid fisheries, the locations of boat lights over space 

 and time are particularly valuable in cases where na- 

 tional boundaries pose constraints on the collection of 

 effort data (e.g., Illex argentinus in the southwestern 

 Atlantic, Waluda et al., 2002). 



In this study, we used boat lights to quantify the spa- 

 tial and temporal patterns of market squid fishing ac- 

 tivity in the Southern California Bight over the period 

 1992-2000. The bight has come to represent the great 

 majority of squid landings off California (Vojkovich, 

 1998; Butler et al., 1999; CDFG, 2000). An important 

 component of our study is ground-truthing work that 

 validates the feasibility of using light data as a measure 

 of fishing effort. This estimate for fishing effort enables 

 us to present novel landings-per-unit-of-effort (LPUE) 

 data for the market squid. A companion paper analyzes 

 the light detection properties of the DMSP-OLS satellites 

 over the Southern California Bight (Elvidge et al. 1 ). 



Materials and methods 



Light detection by satellites 



The DMSP is a polar orbiting satellite system that 

 acquires daytime and nighttime data during each orbit. 

 The OLS is an oscillating scan radiometer designed for 

 cloud imaging. A full technical description of image 

 acquisition by the DMSP-OLS system, and the subse- 

 quent processing of images, appears in a companion 

 paper (Elvidge et al. 1 ). Briefly, the DMSP-OLS acquired 

 nighttime data for over 2200 satellite orbits over the 

 Southern California Bight (i.e., 117° to 122° W, 32°30' 

 to 34°30'N) between 26 April 1992 and 4 April 2001. 

 Four different satellites were employed during this time. 

 Three overlapped in operation dates, producing mul- 

 tiple images for some dates. On all dates, images were 

 acquired between 18:30 and 22:00 Pacific Standard Time 

 (PST), with 20:21 PST being the average time. The satel- 

 lite images were processed into geo-referenced images of 

 boat lights and clouds. This process involved superimpos- 

 ing a field of grid cells onto the satellite image, which 

 quantified the satellite's "field of view," the extent of 

 detected clouds, and the area available for light detec- 

 tion. Image pixels of lights were taken directly from the 



1 Elvidge, C. D., J. Safran, M. R. Maxwell, K. E. Baugh, A. 

 Henry, and J. R. Hunter. Unpubl. data. Satellite based 

 indices of lightboat fishing effort. 



satellite image. Pixels were identified as lights by their 

 visible band digital number. The images were subjected 

 to quality-control procedures to correct for atmospheric 

 noise and to eliminate images overly contaminated by 

 solar glare, sunlight, heavy lunar illumination, or those 

 containing missing data. Fixed sources of lights, such as 

 city lights along the southern California coast, the city of 

 Avalon (Santa Catalina Island), off-shore oil platforms, 

 and naval installations, were masked from the light 

 detection algorithm. 



Data deliveries were irregular during 1992, result- 

 ing in gaps in the early part of the time series. For 

 1992-98, only data collected during the dark half of 

 the lunar cycle were available. To control for lunar il- 

 lumination throughout the time series, we restricted 

 analysis of fishery data to images for which lunar il- 

 lumination was less than 0.02 lux (lumens per square 

 meter). Images for analysis were evaluated against ad- 

 ditional criteria. For a given image, we calculated the 

 number of total grid cells that were not used for light 

 detection because of glare, missing data, or the mask- 

 ing of known nonboat lights. If the resulting number of 

 grid cells left available for light detection was at least 

 50% of the original number of cells, we retained the 

 image for analysis. Cloud coverage can obscure light 

 sources 1 ; therefore we used only images from nights 

 when clouds covered less than 25% of the grid cells 

 available for light detection. For nights with multiple 

 acceptable images, we averaged the percent cloud cover- 

 age and the number of detected light pixels. 



Ground-truthing: aerial observations of boat activity 



To determine the relationship between detected light 

 pixels and the number of squid fishing vessels on the 

 water, 35 aerial surveys were conducted from 10 June 

 1999 to 18 May 2000. Each survey took place in a Cessna 

 337 Skymaster flown at an average altitude of 1160 m 

 above sea level. The path of each survey covered the 

 main areas of squid fishing activity within the South- 

 ern California Bight (Fig. 1): from San Diego, over the 

 Channel Islands, to Point Conception, and back down the 

 coastline to San Diego. Each survey took approximately 

 four hours to complete, occurring between 18:00 h and 

 midnight PST. These times encompassed the time that 

 the DMSP-OLS satellites were over the bight. The 35 

 surveys produced 26 nights of usable data. Survey data 

 were discarded if satellite images were unavailable, if 

 flights were aborted because of weather, or if heavy fog 

 obscured boat visibility. We note that, for this ground- 

 truthing work, we did not restrict our analysis to nights 

 with lunar illumination of less than 0.02 lux. Rather, 

 we used all of the acceptable 26 nights, and quantified 

 lunar illumination as a proportion of the moon's phase, 

 where 0.00 denoted a new moon and 1.00 denoted a full 

 moon. 



All vessels on the water were identified by using Fuji- 

 non 10x50 gyroscopic binoculars, and the GPS positions 

 of all vessels were recorded. Vessel type was identified 

 as either a nonsquid vessel or as a squid fishing vessel. 



