Partners and International Organizations
(English)
|
The Space Application Institute of JRC (SAI), (I); Sea and Sun Technology, (D); DJL Software Consultancy Ltd, (UK); University Le Havre, (F); Institute for Baltic Sea Research, (D); Finnish Institute for Marine Research, (FIN); National Environmental Research Institute, (DK)
|
Abstract
(English)
|
ABSTRACT
The final report summarizes the progress and the results of the EAWAG activities within the MITEC project. The results are sorted according to the work packages as listed in the Technical Annex.
WP 2000 - Algorithms / Data Analysis
Microstructure data, collected for turbulence analysis in natural waters, consist usually of a time series of temperature or velocity fluctuations (Ravens et al., 2000). The goal of microstructure data analysis is to resolve the variance of such turbulent fluctuations up to the smallest scales in order to quantify the level turbulent of mixing, usually expressed by dissipation of turbulent kinetic energy [W/kg]. The aim of this work package is the evaluation and implementation of the required data analysis procedures. The major effort has been put into the development of completely new micro-structure processing software, covering the full range from raw data scaling to fitting of spectra onto appropriate models, to statistical operation upon resulting dissipation profiles.
One major step ahead is the statistical estimation of the instruments stationary noise. That way the noise spectra does not have to follow a certain form in order to estimate an appropriate weight as a function of wave number domain for the fitting routines. Especially for high quality data, most of the processing does no longer require manual corrections if using this tool.
The newest development consists of the integration of the so-called inertial dissipation method into to microstructure data analysis package. This is much of practical interest, since the newest current meter technology, i.e. the so-called High-Resolution Acoustic Doppler Current Profiler (HR-ADCP) allows to resolved currents down to 0.1 mm/s (Gordon et al., 1999) and thereby provides an alterna-tive tool to determine turbulence levels down to 10-10 W/kg. For even weaker turbulence, the tempe-rature method remains unchallenged. In work package 5200 the collected data from different field measurements have been analyzed. Comparison of the output of the different methods is cur-rently in process with encouraging results (Jonas and Wüest, 2000).
WP 4000 - Software Development
Temperature microstructure: New challenging tech-niques could be made use of. The whole software package features data handling and toolboxes implemented in a strict object orien-tated design. A modern graphical user in-ter-face supports the con-venient and fast work with large sets of data. Currently temperature microstructure profiles were evaluated by fitting Batchelor's model spectra onto spectra of temperature segments, where the dissipation rate of TKE is one of two fit parameters. The most prominent output of these enhancements is the spectra's weighting function, which gives a criterion for weighting the individual points of the spectra while fitting (Jonas and Wüest, 2000). The segmentation strategy is another crucial step to unravel the turbulent information from microstructure data. Also long segments allow the detection of lower dissipation rates but on the other hand lead to a higher percentage of unfittable segments be-cause of inhomogeneous turbulence within the segment's range. If the system investigated does not display high advection or turbulent movement, low profiling speeds can mitigate these conflicts.
For the calculations of the spectra, the temperature profiles have been cut into segments of 512 values at a profiling speed of 0.082 m/s. Neighboring segments overlap 50%. The spectra have been estimated from data strings weighted with Blackman-type functions. With this setting we were able to distinguish between non-turbulent segments and low-turbulent segments with dissipation rates below 10-13 W/kg.
High Resolution Acoustic Doppler Profiler current fluctuations: From the HR-ADCP data, dissipation rates could also be evaluated by using the Inertial Dissipation Method. The time series is transformed into velocity information in space using an average advection velocity. Since the mean advection velocity fluctuated with time, the best results were obtained with rather short segments by first transforming time into space and then calculating a subsegment's mean spatial increment. In this context Welsh-averaging method was applicable because the spectra to be fitted are expected to have a linear form in the appropriate double-logarithmic domain. One step earlier, the ADP data were bin-averaged with time. This is a common technique to increase the accuracy of the velocity measurements, but at cost of the high-frequency part of the spectra.
As with the temperature microstructure data, the spectrum's weighting methods for the fitting was the most important part of developing new enhanced strategies for the estimation of dissipation values of TKE from the HR-ADP data. With subsegments of 64 points each, the spectra were long enough to allow cropping parts which have been disturbed by noise from the device. These frequency-dependent arguments were integrated into the spectra's weights together with wavenumber-dependent criteria based upon the domain and shape of the spectra.
WP 5200 - Lake Measurements
The small, wind-protected Soppensee, an ice-hole lake located in the Swiss Plateau, was selected in order to observe cooling-induced turbulence undisturbed by wind or large-scale effects During Sep-tember 20 - 30, 1998. The lake has a maximum depth of 27 m and an oval surface of about 800 by 400 m. During three nights the meteo conditions caused convection strong enough to deepen the top mixed layer by eroding the upper part of the thermocline. The restratified water column was mixed during these three nights all the way down to the bottom of the mixed layer, leading to a tem-perature decrease of 0.3 K. With average wind speed u5 at around 1m/s wave heights never exceeded a few cm. Several experiments have been carried out; stationary measurements of meteo-rological quantities and temperature and current measurements in the water column, verti-cal temperature and shear microstructure, and acoustic current profiler in the metalimnion. Additionally, SF6 was released, and CO2 fluxes measured, respectively, to quantify gas exchange and Con-ductivity-Temperature-Depth (CTD) transects in order to verify the occurrence of lateral effects induced by differential cooling.
A SeaBird temperature microstructure profiler was set up for a cable-constrained and uprising pro-filing mode, to ensure that the complete surface layer could be sampled. By paying out cable at a speed of 0.08 m/s the profiler took measurements whilst rising towards the surface. By returning it to the bottom starting position at the same speed, significant self-induced turbulence was avoided so that 15 minutes were found to be the optimum profiling interval. Noise-free bundles of profiles were of an excellent quality.
In addition, a Nortek high-resolution ADP was deployed at a depth of 7.3 m. Profiling upwards within a 4 m range, the instrument was set up to examine the processes at the bottom of the mixed layer. For 40 depth bins, one every 10 cm, all three velocity components were measured at an internal frequency of 0.5 Hz. The record of the vertical velocity component demonstrates the capability of the ADP to investigate amazingly detailed current patterns. During times of convective mixing an alternating pattern of strong vertical velocities passed over the ADP, as if plumes had advected by. The measurement uncertainty was estimated from a test-application at the site to be 0.2 mm/s.
We are currently in the final stage of data analysis and interpretation. Beside comparison with the different methods, the turbulent findings are also planed to be compared to further Large Eddy Simulations (Sander et al., 2000). Several publications are planed to be completed within the next 12 months.
|
