Premium
Scales of spatial variability for surface ocean p CO 2 in the Gulf of Alaska and Bering Sea: Toward a sampling strategy
Author(s) -
Murphy Paulette P.,
Nojiri Yukihiro,
Harrison D. E.,
Larkin N. K.
Publication year - 2001
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1029/2000gl012375
Subject(s) - cruise , oceanography , longitude , environmental science , flux (metallurgy) , climatology , sampling (signal processing) , spatial variability , geology , structural basin , spatial ecology , latitude , geodesy , statistics , paleontology , ecology , materials science , mathematics , filter (signal processing) , computer science , metallurgy , computer vision , biology
Twenty‐four near‐exact repeat tracks (over a 48‐month period) of ocean surface p CO 2 data from the North Pacific offer an uprecedented opportunity to examine the space scales of p CO 2 in this region. Because previous basin scale studies of air‐sea carbon flux have shown that the field Of Δ p CO 2 (air‐sea disequilibrium) largely controls the field of flux (and atmospheric p CO 2 is relatively constant), knowing how to sample p CO 2 in seawater is a crucial element of the design of a basin scale carbon flux observing system. Unbiased (within 3 µatm) along‐track means for Δ p CO 2 can be obtained from measurements made every 40 km (∼ hourly for a ship traveling 20 knots). We find distinctly different characteristics of spatial variability in two open ocean regions, the Bering Sea and the Gulf of Alaska. The first zero crossing of the mean spatial autocorrelation function is 1.5° longitude in the Bering Sea and 3.5° in the Gulf of Alaska, when the cruise data are linearly detrended to remove the larger scale field. There is a strong seasonal variation in the zero crossing, with shorter scales in the summer. In near‐shore waters there is extreme variability, often with very small space scales and very large cruise‐to‐cruise differences. Neither the mean nor variability statistics from a single ship cruise data set appear likely to produce reliable information for designing an observing system.