STIX counts - GOES flux Relationship

What is the relationship between peak counts in STIX flares and GOES class/flux?

c.f. Lastufka et al 2020b

Get data

• STIX flare list from STIX data center (stixdcpy)
• HEK events from HEK database (sunpy Fido)

Prepare data

• Light travel time correction for STIX flare list times
• Normalize counts to counts @ 1 AU from Sun

Additional quantities

• is HEK flare visible from Solar Orbiter?
• rotated coordinates of flare location (AIA frame to Solar Orbiter frame)
• difference between event peak times as recorded by AIA and STIX
• convert GOES class (str) to GOES flux (float)

Selection

• HEK flares visible from Solar Orbiter
• Absolute time difference between peak time observed less than 10 minutes
• if more than one HEK result, keep the one from SSW Latest Events because that's the one that also gives a GOES class associated with it
  • to consider: Flare Finder gives GOES flux instead of class, might be more useful?
  • need to check which one gives a result in more cases
• optional: STIX peak counts threshold

Current limitations

• no CFL locations for STIX flares; would be helpful to double-check joint visibility
• STIX peak counts/s energy dependent? need to check

#ldf['AU_to_Earth']=np.sqrt(np.multiply(ldf.solo_x-1,ldf.solo_x-1)+np.multiply(ldf.solo_y,ldf.solo_y)+np.multiply(ldf.solo_z,ldf.solo_z))
#want no correction if concentric with Earth orbit... ie if 1 AU from sun
#so for distance correction use 1/(1-solo_r)**2
ldf.describe()

	solo_x	solo_y	solo_z	AU_to_Earth
count	701.000000	701.000000	701.000000	701.000000
mean	0.088660	-0.030678	0.009766	1.047430
std	0.569704	0.571013	0.039378	0.621496
min	-0.622433	-0.935519	-0.056593	0.011632
25%	-0.453305	-0.547169	-0.017508	0.319418
50%	-0.124245	-0.049534	0.008084	1.383826
75%	0.695613	0.566451	0.022782	1.519672
max	1.009807	0.804849	0.099672	1.787906

Flare property distributions

Total Flares: 2864

Jointly Visible: 1228

Jointly Visible, with GOES class: 963

Flare Locations

Counts- flux relationship

Linear fit to log-log data (least squares)

$\displaystyle GOES\_flux = 0.665\ log_{10}(peak\_counts)+-8.083$

$\displaystyle R:0.844, p:1.571e-262$

O(2) polynomial fit to log-log data (least squares)

$\displaystyle GOES\_flux = -0.033\ log_{10}(peak\_counts)^{2}+0.879\ log_{10}(peak\_counts)+ -8.414$

sum of squared residuals: 64.661

Residuals

Observed GOES class - GOES class calculated using linear relationship

log10 (counts)	1 $\sigma$	N flares
< 3	0.273	707
[3,4)	0.224	198
> 4	0.196	58

Selecting only events with peak_counts above a certain threshold

What threshold in peak counts gives a fit the strongest correlation (R-value)?

def find_best_r(df, threshold_range):
    largest_r=0
    for t in threshold_range:
        dfc=df.where(dfgc.peak_counts_corrected > t).dropna(how='all')
        rel2=linregress(np.log10(dfc.peak_counts_corrected.values),np.log10(dfc.GOES_flux.values))
        if rel2.rvalue > largest_r:
            largest_r=rel2.rvalue
            best_results=rel2
            threshold=t
            data=dfc
    return best_results, largest_r,threshold, data

res2,bigr,thresh,dfc=find_best_r(dfgc,np.arange(100,2000,50))

Peak Counts Treshold: 1100 counts

Number of Flares: 241

$\displaystyle GOES\_flux = 0.628\ log_{10}(peak\_counts)+-7.954$

$\displaystyle R:0.853, p:2.373e-69$

Flares not jointly visible - calculate GOES flux