ALIS data analysis

The primary raw-data output from ALIS is multi-station image data in FITS format (Section 2.2.5). Currently, the main part of the data-analysis software for ALIS includes programs for correcting the images, calibrations against the star background, auroral tomography and triangulation, spectroscopic and absolute measurements, etc. Most of the programs are written in Matlab and C, but some work was also carried out in IDL, NOAO/IRAF as well as other common image-processing environments. The main authors of the current data-analysis software for ALIS are Björn Gustavsson (see Table 6.1) and Peter Rydesäter (see Table 6.2).

Table 6.1: Summary of image processing tools provided by Björn Gustavsson at http://www.irf.se/~bjorn/Starcal/. They provide a small and self-contained set of functions for camera calibration as well as single and multi-viewpoint image data analysis. These tools are primarily intended for middle and upper atmospheric physics as well as for auroral research. (Quoted from: http://www.irf.se/~bjorn/Starcal/Documentation/ see also Gustavsson [2000].)

Table 6.2: Summary of data analysis tools written by Peter Rydesäter, available at http://petrydpc.itm.mh.se/tools/. See also Rydesäter [2001] for more information.

Carl-Fredrik Enell has developed additional tools for the analysis of PSC data [PSCWorks, see Enell, 2002, Appendix B]. A combination of the data-analysis tools mentioned was utilised to analyse nearly all of the observations summarised in this chapter.

Investigations of new data-analysis methods for ALIS

As ALIS produces large amounts of data, studies have been carried out to investigate various new and partly automated approaches to the problems of image processing, pattern recognition and image classification. Some references to published works on ALIS are presented below, however, this field is to a large extent highly experimental in nature and therefore these results are not included in the standard analysis of ALIS data.

To classify auroral images is not a straight-forward task. As noted by Brändström et al. [1998] it is sometimes difficult, even for an experienced observer, to discriminate between for example diffuse aurora and aurora behind thin clouds (Figure 6.1).

Figure 6.1: An example of the difficulties of auroral image classification. In this image, clouds, stars, diffuse and black aurora are seen. [After Figure 8 in Brändström et al., 1998]

Early on in the ALIS project, an auroral classification scheme was suggested (by Åke Steen, see Table 6.3).

Table 6.3: An auroral classification scheme (after a proposal by Å. Steen) used by the manual classification program for ALIS data. (``Dark aurora'' stands for a class of phenomena commonly known as ``black aurora'')

Auroral classification scheme
Main	Secondary	Temporal	Vortex
morphological	morphological	feature	feature
feature	feature
	Non-diffuse	Active	Spiral (E.g. WTS,
	Diffuse	Quiet	omega, torch)
	Multiple (2,3,...)	Pulsating	Fold
Auroral arc	Rayed		Curl
one dimension is	Striated
significantly larger	Corona
than the other	Dark aurora
	Enhanced
	Partly clouds
	Intensity
	Non-diffuse	Active	Fold
	Diffuse	Quiet	Curl
	Multiple (2,3,...)	Pulsating
	Rayed
Fragmental auroral	Striated
structure	Corona
	Dark aurora
	Enhanced
	Partly clouds
	Intensity
	Dark aurora	Pulsating
Diffuse aurora	Partly cloudy
covering a larger region	Intensity
Unidentified aurora

A classification program (developed by Petrus Hyvönen and later Mats Luspa) was written. The program was used to manually, but relatively quickly, classify the images according to the classification scheme in Table 6.3. This classification was very useful but time-consuming due to the large number of images from many stations. The quality of the image-classification was also dependent on the operator. This classification method was therefore abandoned, and investigations started to find automated methods of image classification.

In an exploratory study carried out by Eide et al. [1997]; Waldemark et al. [1997] Pulse Coupled Neural Networks (PCNN) were evaluated as a preprocessor for classifying image-data from ALIS. The PCNN is a biologically inspired neural network based on findings in the visual cortex of small mammals. The algorithm has been successfully applied in the field of mammography [Kinser and Lindblad, 1997] which, to some extent, possesses similar image processing and classification problems as encountered in auroral imaging. A three-step procedure for automatically classifying auroral images is outlined: (1) use a PCNN for image segmentation, (2) post-process the resulting data, either by using Singular Value Decomposition (SVD) or by applying a second PCNN for feature extraction, (3) carry out the actual classification using a traditional neural network. This work was followed up by Rydesäter et al. [1998] who concentrated on the problem of how to find auroral arcs, and on their location within the images. For this purpose an arc detection algorithm was constructed and tested. Initial attempts were made to enhance the algorithm with a Radial Basis Neural Network (RBNN). It was concluded that the arc detection algorithm gave a robust detection of auroral arcs and their direction. Preliminary results from the application of a RBNN was promising.

Alpatov et al. [2000] applied self-organising neural networks to ALIS data, for the detection of polar stratospheric clouds.

A different approach to the problem of automatic recognition of auroral forms is studied by Pudovkin et al. [1998]. Here a more traditional method, based on the analysis of isolines of auroral luminosity shapes is utilised. Classical forms such as ellipses, spirals and folds appear to be confidently retrieved. However, Pudovkin et al. [1998] notes that the entire field of auroral classification seems to be at an early stage: ``There are needs for extensive and purposeful studies of the forms peculiar to certain geophysical conditions, with the nature of the physical process within the auroral plasma and their characteristic time and space scales being taken into account.''

Rydesäter [2001] presents a survey of possible methods of implementing a Selective Imaging Technique (SIT) functionality in ALIS, as proposed by Steen et al. [1997a]. (This would be an algorithm or a device that automatically controls ALIS, or advises the operator to acquire data of specific interest when favourable conditions occur.)

A study on the effects of lossy compression of auroral image data is presented in Rydesäter et al. [2003]. At least for the images studied, it appears to be possible to compress images to approximately 5-12% of their original size without increasing errors.

Another topic is the study of the fractal dimension of auroral features. An initial study was carried out by Alpatov et al. [1996b]; Alpatov et al. [1996a]. It was concluded that fractal analysis methods can be used to segment auroral images into aurora, stars, clouds and background. However it remains to be seen if segmentation of the auroral features themselves can be related to physical processes responsible for the aurora.

Despite the fact that these studies provided promising preliminary results, much work remain before these methods can be applied to auroral imaging on a regular basis.

Apart from the work related to ALIS, many other groups are working in this field. An example of this is the very promising work by Syrjäsuo [2001, and references therein] on auroral detection/classification by automated image analysis.