Simulation Results and Performance Analysis of Phase Velocity Spectral Analysis Software (M-Transform) for Airglow Imaging Data
Abstract
Airglow imagers have been deployed in various locations to study the atmospheric gravity waves (AGWs) in the last few decades, which provide a huge amount of data. This ground-based imaging observation has been proven to be very effective to study energy and propagation characteristics of AGWs. However, the lack of sophisticated analysis methods prevented quantitative studies using huge data that have been collected over such long period of observation. Matsuda et al. [2014] developed a new phase velocity spectrum analysis method based on 3D FFT algorithm (M-transform) to address this issue. This algorithm can efficiently deal with extensive amounts of imaging data without bias and treat dynamical/physical effect of AGWs by precisely reflecting amplitude, area and lifetime of each AGW event. Based on Matsuda et al. [2014] method, we have developed a simple and user-friendly function to be used in IDL. The input of this program is a three dimensional array of a time series of 2-D image array where the wave parameters (e.g. horizontal wavelength (_h), wave period (_), phase speed (c), image resolution in space (dx, dy) and time (dt)) can be customized by the user in one-line command to execute the program. Various simulations by using artificial wave images as test data have been done in order to exhibit the new program's performance and the characteristics/interpretation of the spectral analysis. The simulation was done by using test data with different time resolution (dt=1 min, 3 min, 5 min and 7 min), image size (256x256, 512x512), duration (30 min, 60 min, 90 min, 120 min) and by changing the horizontal wavelength and wave periods with a fixed phase speed (40 m/s). In addition, by using observational data from Syowa station we conducted an investigation how waves with different horizontal wavelengths behave independently. We divided the horizontal wavelength into three categories: 5< _h <20 km, 20< _h <100 km and 5< _h <100 km. We found that propagation direction for small scale, close to ripple scales (5< _h <20 km), showed omni-directional feature, as previously suggested by Nakamura et al. [1999]. Beside the detail discussion on the simulation results and performance test, this presentation also aims to guide how to use the program within a practical computation time.
- Publication:
-
42nd COSPAR Scientific Assembly
- Pub Date:
- July 2018
- Bibcode:
- 2018cosp...42E2635P