All Figures

$\begin{figure} \par\includegraphics[width=5.1cm,clip]{5951fg1A.eps}\par\vspace*{3mm} \includegraphics[width=5.1cm]{5951fg1B.eps}\end{figure}$ Figure 1: Central disk detail of a CaIIK image from OAR PSPT archive and corresponding mask obtained with the second identification method described in the text. Features with areas lower and larger than 2000 Mm² are shown in gray and black colors respectively.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=6.5cm,clip]{5951fig2.eps} \end{figure}$ Figure 2: Fractal dimension d1 versus area of bright features identified on calcium images ( ${\sim } 2\hbox {$^{\prime \prime }$ }$ /pixel). Full circles: summers 2000-2005 OAR-PSPT. Open triangles: summer 2005 MLSO-PSPT. Full triangles: summer 2005 OAR-PSPT. d1 increases fast with object size at area smaller than 2000 Mm². For larger areas, a plateau is observed for summer 2005 OAR and MLSO data, and a slow rise on the 2000-2005 OAR dataset.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=6.5cm,clip]{5951fig3.eps} \end{figure}$ Figure 3: Temporal variation of the fractal dimension d1 versus area for features identified on OAR-PSPT calcium images. The bar on the left represents the largest error bar, obtained for the largest areas for year 2004. At area smaller than about 1000 Mm² all the curves overlap, while differences (not clearly correlated with solar cycle) are observed at the largest areas.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=6.5cm,clip]{5951fig4.eps} \end{figure}$ Figure 4: Temporal variation of the fractal dimension D versus area for selected OAR-PSPT calcium images and for different area range. Error bars in the case of fits performed on the whole dataset (circles) or at smallest objects (triangles) are smaller then the symbol size. Results obtained for the largest area are in good agreement with results obtained by d1 estimator (Fig. 3).
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[height=6.8 cm,width=5.7cm,clip]{5951fig5.eps} \end{figure}$ Figure 5: Fractal dimension d1 ( top) and D ( bottom) estimated for circles as function of object size and minimum area threshold respectively. Results obtained with external sides (crosses) and 8-contiguous point (triangles) perimeter finding algorithm are shown. Error bars are smaller then symbol size. Note that the horizontal scales differ for the two panels.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[height=3.6cm,width=6.5cm,clip]{5951fig6.eps} \end{figure}$ Figure 6: d1 evaluated for von Koch snowflakes of level 6. Like non fractal objects and real data, d1 increases with object size and reaches a plateau at areas $\ge$ 1000 pixel². The plateau value, about 1.34, is an overestimate of the snowflake fractal dimension (see text).
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=7.3cm,clip]{5951fig7.eps} \end{figure}$ Figure 7: Facular fractal dimension estimated on the two different full resolution MLSO quality sets described in the text. When the estimation is carried out on images less affected by seeing degradation, the measured fractal dimension is higher.
Open with DEXTER

In the text

$\begin{figure} \par\includegraphics[width=6.9cm,clip]{5951fig8.eps} \end{figure}$ Figure A.1: Perimeter (in units of pixel) and Area (in units of pixels square) in logarithmic scale of detected structures on OAR PSPT data taken during summer 2002. Continuous line is the fit to the whole set of data (D=1.35). Points at area larger than about 1000 pixels square are better approximated by a higher slope line. Horizontal line is the area window width over which d1 is estimated.

In the text

$\begin{figure} \par\includegraphics[width=13.8cm,clip]{5951fig9.eps} \end{figure}$ Figure A.2: Left: perimeter versus area in logarithmic scale of snowflakes of levels 2, 4, and 6. Because of pixelization, these structures exhibit a fractal scaling only within certain area ranges, the bounds depending on the snowflake level. Right: d1 versus area evaluated with different window sizes for snowflakes of level 6. Peaks obtained with the small window (open circles) are due to the steep variations visible in plot on the left. Peaks are not detected with a larger window (full circles). The area range over which d1 is almost constant is the range over which the simulated images are fractal.

In the text

$\begin{figure} \par\includegraphics[width=5.1cm,clip]{5951fg1A.eps}\par\vspace*{3mm} \includegraphics[width=5.1cm]{5951fg1B.eps}\end{figure}$	Figure 1: Central disk detail of a CaIIK image from OAR PSPT archive and corresponding mask obtained with the second identification method described in the text. Features with areas lower and larger than 2000 Mm² are shown in gray and black colors respectively.
Open with DEXTER

$\begin{figure} \par\includegraphics[width=6.5cm,clip]{5951fig3.eps} \end{figure}$	Figure 3: Temporal variation of the fractal dimension d1 versus area for features identified on OAR-PSPT calcium images. The bar on the left represents the largest error bar, obtained for the largest areas for year 2004. At area smaller than about 1000 Mm² all the curves overlap, while differences (not clearly correlated with solar cycle) are observed at the largest areas.
Open with DEXTER

$\begin{figure} \par\includegraphics[width=6.5cm,clip]{5951fig4.eps} \end{figure}$	Figure 4: Temporal variation of the fractal dimension D versus area for selected OAR-PSPT calcium images and for different area range. Error bars in the case of fits performed on the whole dataset (circles) or at smallest objects (triangles) are smaller then the symbol size. Results obtained for the largest area are in good agreement with results obtained by d1 estimator (Fig. 3).
Open with DEXTER

$\begin{figure} \par\includegraphics[height=6.8 cm,width=5.7cm,clip]{5951fig5.eps} \end{figure}$	Figure 5: Fractal dimension d1 ( top) and D ( bottom) estimated for circles as function of object size and minimum area threshold respectively. Results obtained with external sides (crosses) and 8-contiguous point (triangles) perimeter finding algorithm are shown. Error bars are smaller then symbol size. Note that the horizontal scales differ for the two panels.
Open with DEXTER

$\begin{figure} \par\includegraphics[height=3.6cm,width=6.5cm,clip]{5951fig6.eps} \end{figure}$	Figure 6: d1 evaluated for von Koch snowflakes of level 6. Like non fractal objects and real data, d1 increases with object size and reaches a plateau at areas $\ge$ 1000 pixel². The plateau value, about 1.34, is an overestimate of the snowflake fractal dimension (see text).
Open with DEXTER

$\begin{figure} \par\includegraphics[width=7.3cm,clip]{5951fig7.eps} \end{figure}$	Figure 7: Facular fractal dimension estimated on the two different full resolution MLSO quality sets described in the text. When the estimation is carried out on images less affected by seeing degradation, the measured fractal dimension is higher.
Open with DEXTER