cadair / asda

This is an updated code of the ASDA code in https://github.com/PyDL/ASDA

Credit:

Liu, J., Nelson, C, Erdelyi, R, Automated Swirl Detection Algorithm (ASDA) and Its Application to Simulation and Observational Data, ApJ, 872, 22, 2019 (https://iopscience.iop.org/article/10.3847/1538-4357/aabd34/meta)

This version is a revision by Norbert Gyenge and Jiajia Liu of version 1.0 in https://github.com/PyDL/ASDA This version is on average 3 times faster

ASDA

Automatic Swirl Detection Algorithms

System Requirements

OS Requirements

ASDA can be run on Windows, Mac OSX or Linux systems with Python 3 and the following dependencies installed.

Dependencies:

Python 3 with libraries including numpy, scipy, matplotlib, scikit-image, warnings, itertools
Python management softwares including Anaconda or Virtualenv are recommended

Hardware Requirements:

ASDA requires a standard computer with enough CPU and computation power depending on the dataset used.

Installation Guide:

ASDA is a stand-alone Python package, no installation is needed.

Description of Files (More information can be found in each file):

asda.py: Main programm of the implication of the swirl detection algorithms
utils.py: utility functions
demo.py: demo
data: this folder contains data need for running the demo

Instructions for Use:

Suppose you have vx and vy for the horizontal velocity field. These two arrays are in (y, x) order.

1. import neccessary libraries, import asda
2. Initialize the ASDA class lo = asda.ASDA_Calc(vx, vy)
   
3. calculate gamma1 and gamma2 values (see the reference) with gamma = lo.gamma_values(). The variable gamma will be a tuple, where gamma[..., 0] is gamma1 and gamma[..., 1] is gamma2.
   
4. perform the detection of vortices using center_edge = lo.center_edge(). The variable center_edge will be a dictionary with keywords "center", "edge", "points", and "peak". center is a list containing the pixel location of all swirls in the velocity field. edge is a list of the edges of all vortices. point is a list of all points within swirls. peak is a list of the peak gamma1 values of all swirls. radius is a list of the effective radii of all swirls. Note that center, points and edge are stored in the order of [x, y] and in units of pixels.
   
5. use ve, vr, vc, ia = lo.vortex_property(image=image) to calculate the expanding, rotating, center speeds of above vortices. ia is the average intensity from the observational image (if given) of all points within each vortex. Note that vc is stored in the order of [vx, vy].
   

Demo

A demo demo.py is available:

1. To run the demo, run python demo.py
2. The demo data consists 2 files: vxvy.npz stores the velocity field (vx, vy) determined by FLCT using SOT Ca II observations at 2007-03-05T05:48:06.737 and at 2007-03-05T05:48:06.737, and the observational data (data) at 2007-03-05T05:48:06.737. correct.npz contains the swirl detection result from ASDA version 1.0 (https://github.com/PyDL/ASDA) on exactly the same dataset.

Expected Running Time

On an Intel I7 4.20 GHz CPU: the first function in demo.py takes ~2 seconds, and the second function in demo.py takes ~ 27 seconds.