Aquatic Sciences Meeting, Albuquerque 2001

SS36 Dealing With Scales in Aquatic Ecology: Structure and Function in Aquatic Ecosystems (Spatial and Temporal Connections)

Date: Tuesday, February 13, 2001, Time: 11:30:00 AM

Location: San Miguel

Fisher, K, E, Cornell University, Ithaca, USA, kef10@cornell.edu

Cowen, E, A, Cornell University, Ithaca, USA, eac20@cornell.edu

Pasour, V, , Cornell University, Ithaca, USA, pasour@cam.cornell.edu

WAVELET VARIANCE AND FRACTAL INTERPOLATION TECHNIQUES FOR CHARACTERIZING PATCH STRUCTURES

Along-track data were gathered for acoustic backscatter, fluorescence, salinity, and temperature in the surface waters around Georges Bank during six U.S. GLOBEC Broadscale cruises. Wavelet variance analysis of these measurements shows that horizontal patchiness of both plankton populations and hydrographic variables exhibits structure across all measured scales, from 17 meters to 29 kilometers. The wavelet transform, W[x,a_n], gives the spectral component of the along-track 1D surveys obtained in January, March, and June in both 1998 and 1999: x is position along the track and a_n is the wavelet scale. The spectral power is VW, the variance of W[x,a_n]. If VW ~ (a_n)^b, then the data series is self-affine --there is no inherent scale in the spectral power as a function of wavelength. A white noise (uncorrelated data) results in b = 0; a Brownian motion (strongly correlated data) results in b = 2. For comparison, Fourier power spectra have been computed as well. These analyses show that biological variance structure differs substantially from physical variance structure in many instances. Thus, it is expected that patchiness of biological variables, controlled by a combination of behavior and environment, both differs in character and influences function of ecosystems in a way distinct from patchiness of purely physical variables. In addition, biological and physical controls differ in character at various scales, in different hydrographic regimes, and with changes in season. Interpreting patterns of variance requires the context of the length scales determining the degree of correlation in the control processes. Fractal interpolation is compared to linear techniques as a strategy for constructing 2D fields for each measured variable from trackline data. Fractal interpolation has the potential of incorporating information from observed length scales and associated variance structures into the 2D fields. In turn, the implications for quantitative interpretation of the resulting fractal and linear 2D fields are considered.

Return to This Session's Schedule · Complete Session Listing ·

Aquatic Sciences Meeting, Albuquerque 2001

Quicklinks

Quicklinks