A flare sensitive 3 h solar flux radio index for space weather applications

Transcription

1 SPACE WEATHER, VOL. 9,, doi: /2010sw000585, 2011 A flare sensitive 3 h solar flux radio index for space weather applications Ariel O. Acebal 1,2 and Jan J. Sojka 1 Received 8 March 2010; revised 1 April 2011; accepted 20 April 2011; published 21 July [1] Many space physics models use the F10.7 as their input for solar activity. The F10.7 is a daily index derived from solar radio measurements taken at 2800 MHz, excluding activity from solar flares. In this paper, we compute a 3 h composite index, similar, in part, to the F10.7, using solar radio observations taken at 2695 MHz (11.1 cm) by the United States Air Force s Radio Solar Telescope Network. This index, called the F11.1 index, is similar to the F10.7. But unlike the F10.7 index, which is measured three times each day, at 1700, 2000, and 23 UT, F11.1 consists of eight measurements each day, uniformly distributed over 24 h. These 3 h intervals are aligned in UT with the planetary geomagnetic index Kp s time intervals. Each interval provides an F11.1 value that minimizes solar flare radio emission data. This composite index also provides two additional pieces of quantitative information that the F10.7 does not provide. The first is a factor, ranging from 0 to 1, indicating how representative the single F11.1 value is of this entire 3 h period (representation accuracy parameter). The second is a measure of how much of the 3 h interval can be classified as solar disturbed or as having a flare in progress (duration parameter). These aspects together have relevance for ionospheric modeling/ specification for solar conditions in which significant change can occur over a 24 h period. Citation: Acebal, A. O., and J. J. Sojka (2011), A flare sensitive 3 h solar flux radio index for space weather applications, Space Weather, 9,, doi: /2010sw Introduction [2] Solar UV radiation below 200 nm is the primary source for heating the thermosphere, creating the ionosphere and driving diurnal cycles of winds and chemistry [Woods et al., 1998]. Variations in this energy range alter the energy, dynamics, and chemistry of the atmosphere. Unfortunately, solar UV radiation is completely absorbed by the thermosphere. This limits the observations of solar UV radiation since it can only be observed with space based sensors. In order to overcome this limitation, scientists have looked for different proxies for solar UV radiation. For instance, soft X rays from the GOES satellites, are often used by space weather forecasters to infer changes in the ionosphere. Unfortunately, however, these wavelengths only impact the lower ionosphere. Furthermore, recent losses of soft X ray sensors could reduce the reliability of this data source. The most commonly used proxy for solar UV radiation is the F10.7 index [Tapping, 1987; Tapping and Harvey, 1994]. 1 Center for Atmospheric and Space Sciences, Utah State University, Logan, Utah, USA. 2 Air Force Institute of Technology, Wright Patterson Air Force Base, Dayton, Ohio, USA. [3] In 1947, the National Research Council of Canada began taking daily flux density measurements of the Sun at 2800 MHz (l = 10.7 cm). These measurements have been made at the Algonquin Radio Observatory near Ottawa, Canada from 1947 until 31 May The observing program was then moved to the Dominion Radio Astrophysical Observatory, near Pencticton, British Columbia where it continues to operate today. A detailed discussion of the observing technique and the calibration of the equipment can be found in articles written by Tapping and Charrois [1994] and Tanaka et al. [1973]. This resultant measurement is not an average of the Sun s daily variation, but a snapshot of the Sun at the time the measurement is taken. The 10.7 cm flux density measurement is commonly referred to as the F 10.7 or F10 and is a combination of solar components. Solar radio emission falls into three main components: a burst component, produced by flares and other transient phenomena; a slowly varying component, which originates in the chromosphere and corona over active regions, varying slowly with the evolution and number of active centers; and the quiet Sun component, which forms a constant base level that is only observed when activity is very low. The components of solar radio emission are discussed in depth by Kundu [1965]. Both then and now, corrections must be made to the flux value to remove any burst activity [Tapping and Copyright 2011 by the American Geophysical Union 1of12

2 Figure 1. Map of RSTN sites. The Radio Solar Telescope Network is operated by the United States Air Force. It consists of four observatories located in Palehua, HI (PHFF); Learmonth, Australia (APLM); San Vito, Italy (LISS); and Sagamore Hill, MA (K7OL). Each observatory monitors eight discrete frequencies at a 1 s cadence. Charrois, 1994]. This daily index is called a flux density. The measurement unit for flux density is watts per square meter per Hz or a Jansky. The intensity of solar radio observations is very small, usually around wm 2 Hz 1. This value is called a solar flux unit or SFU and is used to report solar radio emission values as well as this index. [4] The correlation between sunspot and solar radio emissions in the centimeter wavelength range was discovered independently by Covington and Lehany [Tapping and DeTracey, 1990], and it was not long before the F10.7 became a proxy for solar activity. Later research showed that the F10.7 could be used to model UV and EUV [Bossy, 1983]. Although high temporal resolution solar radio data is available online from a variety of sources, including the Radio Solar Telescope Network (RSTN) 1 s data and Pencticton s F s data, many space weather models and applications only use the official daily value of the F10.7 index available from the Space Weather Prediction Center. Since this index is a daily value, models either update the solar input once a day or use a linear interpolation between days to adjust the indices throughout the model run. [5] However, since the Sun is not static throughout the day, if a model runs in real time, this daily variability is not adequately captured by the F10.7 index. In order to overcome these issues, we propose a 3 h index computed with real time radio observations from solar radio data. This index would be similar to the F10.7 (in that flare effects would be excluded), but it would provide information regarding how accurately the index represents the current level of solar activity, which would include flare information. Section 2 discusses the physics behind radio emissions in the centimeter wavelength range; section 3 covers the radio data set and the equipment used to measure it; section 4 presents the method used to compute the index; section 5 discusses these F11.1 index results; and finally section 6 provides the summary which includes the relevance to ionospheric modeling. 2. The Radio Sun [6] The Sun is an average star with a radius of km, mass of kg with a distance of about 150 million kilometers from Earth. It is primarily com- Table 1. Actual Frequencies Monitored by the RSTN Sites Frequency (MHz) Wavelength (cm) Palehua (MHz) Learmonth (MHz) San Vito (MHz) Sagamore Hill (MHz) , , , , , , , , , , , , , , , , , , , , , , , , , of12

3 Figure 2. June observing times for each RSTN observatory. Because of the longitudes of the observatories, some observatories observing time windows overlap, giving multiple observations of the solar radio flux and events. Figure 3. During the December solstice in the Northern Hemisphere, the overlap among the observatories observing times decreases, due to the nonequatorial latitude of the sites. However, there is always coverage from at least one observatory. 3of12

4 Figure 4. Here 13 July 2005 as seen by all four solar radio observatories. On this particular day, there were three significant radio bursts: around 0300 UT, 1500 UT, and 2100 UT. (a) Each burst was observed by at least two observatories. (b) Thebaselinediffersfromoneobservatoryto another. This is due to the calibrations performed at each observatory. The data for this plot was quality controlled with several different algorithms described in this paper. posed of hydrogen and helium but has traces of many other elements. The spectrum of the Sun resembles that of a blackbody at a temperature of 5,800 K. A blackbody at this temperature emits primarily in the visible and infrared wavelength range. This is evident in the solar energy received at the top of the Earth s atmosphere. This energy is called the solar constant and is around 1360 W m 2. Most of this energy is in the infrared and visible range 4of12

5 Figure 5. Each 3 h period is analyzed for each observatory. Two values are computed. For the first value all observations are summed for the 3 h period (black and red trace). The second value is the summation of all values except those greater than average plus two standard deviations (black trace). This second value excludes most of the activity associated with solar flares (shown on the red trace). while less than 0.1% falls in the range of X rays, EUV, and radio. Despite its name, the solar constant is not constant. In addition, not all wavelength ranges change in a similar fashion. The visible and infrared ranges show little variation regardless of timescale but the X ray and radio ranges do show significant variation at all timescales. The 11 year solar sunspot cycle exhibits some of the longest timescale variations. On a shorter time scale, the 27 day solar rotation period also changes the solar constant. These changes are more rapid but not as long lasting and are associated with the emergence, evolution, and disappearance of active regions. According to Lean and Frohlich [1998] the 11 year variations of the solar constant are around 0.1%, while the 27 day variations are in the order of a few tenths of a percent. Even solar flares cause very small, short term variations in the solar constant. Flares are violent events in the solar atmosphere that are associated with short duration, minutes to hours energy releases of up to J. [7] The solar radio spectrum spans six orders of magnitude and varies according to the three time scales previously discussed. These three types of solar radio emissions are (1) the quiet Sun, (2) the slowly varying component, and (3) bursts. The background or quiet Sun component varies in intensity over a timescale of years and originates all over the solar disk. The difference between solar maximum and solar minimum could be as high as 300% at 2695 MHz. The slowly varying component is mainly due to enhanced thermal radiation from active regions. This component varies in a monthly timescale as new regions are formed, remain on the Sun for several solar rotations and eventually dissipate. The variation for this component could be from as high as a 25% change during solar minimum to 50% during solar maximum. The burst component is associated with flares. These bursts are usually located within active regions and could last from a few minutes to several hours depending on the size of the flare. The changes observed in the radio spectrum depend on the radio frequency observed and the size of the flare and could show as large as a two order of magnitude increase. The mechanisms responsible for these different types of radio emissions are bremsstrählung, magnetobremsstrählung, plasma emissions, and electron cyclotron maser. Details of these mechanisms can be found in the work by Dulk [1985], Lee and Gary [2000], Pacholczyk [1970], Raulin and Pacini [2005], and Treumann [2006]. 3. Radio Solar Telescope Network [8] The United States Air Force (USAF) currently operates four solar radio observatories that span the globe. These four locations are shown on a map of the world in Figure 1. The observatories monitor eight discrete 5of12

6 Figure 6. Three hour index for 13 July 2005 for all four observatories. Notice that the values are not the same for different observatories for the same 3 h period. This is due to the baseline established for each observatory during the observatory s calibration procedures. The symbol key is as follows: open square for Learmonth; solid square for Palehua; open circle for Sagamore; and solid circle for San Vito. frequencies and two frequency ranges 24 h a day, 365 days a year, using a 1 s time resolution. The radio observatories form the Radio Solar Telescope Network (RSTN) and are identified by a four letter location indicator code, assigned by the International Civil Aviation Organization (ICAO). The ICAO codes for the RSTN sites are PHFF for Palehua, Hawaii; APLM for Learmonth, Australia; LISS for San Vito, Italy; and K7OL for Sagamore Hill, Massachusetts. For clarity, the sites will be referenced by location rather than ICAO (e.g., Learmonth instead of APLM or Australia). The discrete frequencies vary slightly between the observatories to account for local conditions at each site (Table 1). Because of their geographic locations, some of the observatories overlap during their daily observation period. Figures 2 and 3 show the overlap periods during the June and December solstices. This overlap provides a check on what is, in effect, a continuous observation of the Sun, but from four independent observatories in sequences. 4. Three Hour Index Methodology [9] The four RSTN stations provide a one second data product in solar flux units (SFU) for each of their eight frequencies. Each RSTN station carries out a daily calibration and in addition, an intercalibration is carried out between the stations. Appendix A provides a summary of the RSTN calibration procedures. Before the 3 h index can be computed, the raw one second data requires an additional cleanup procedure that is based upon first hand work with the RSTN data product. The following steps are applied to all data. [10] 1. The first and last 20 min period of each station s daily observations are removed. (Manual tracking operations during sunrise and sunset are prone to varying effectiveness.) [11] 2. Data drop outs of less than 5 s are removed. (Manual movement of antennas lead to such dropouts.) [12] 3. Spikes of less than 10 s duration are removed. (Most common form of local RFI.) [13] 4. During calibration periods, identified as zero SFU, the adjacent 10 s period of data is removed. (Start and end of calibration period involves manual adjustments.) [14] Note that three of these four procedures could be redundant if a fully automated tracking system were adopted. Figure 4a shows an example of the accepted one second data from the four stations over a 24 h period. These are superimposed in Figure 4b. In this overlay, it is evident that the baselines for the four stations are still not identical. This difference is, however, within the specification for the RSTN product. For the 3 h product, this is removed by adopting one of the stations as the baseline. 6of12

7 Figure 7. To remove the effects caused by the different baselines, one observatory is considered the ground truth, and the other three are scaled accordingly. The 03UT index values are computed for Palehua and Learmonth. The values of Learmonth are scaled so that they are the same as the values for Palehua. The dashed line is the ideal result where both observatories have the same value. The solid line is a linear least squares fit of the data. [15] In order to make the F11.1 index similar to the daily F10.7 index, the flare related radio bursts have to be excluded from the computations. Unfortunately, there is no standard method to carry out this procedure. The bursts themselves range from short lived intense fluxes to enhanced fluxes that extend over hours. Hence, the procedure adopted here is intended to capture and, therefore, remove the strongest short lived flares. Using all observations recorded by each observatory in 2005, the average value and standard deviation for the 2695 MHZ observations are computed for each site. Then any observation whose value exceeds the average value plus two standard deviations is excluded from the F11.1 computation on the basis that these are flare related enhancements. Figure 5 shows a 3 h period on 13 July 2005 observed by the Palehua Solar Observatory. The data shown in black are those below the 2 standard deviation threshold and are used for the F11.1 index computation. These values in the 3 h UT window are averaged and become the F11.1 value. [16] The next step is to quantify the effectiveness of the F11.1 index. Since the F11.1 index, like the daily F10.7 index, does not include the effects of readily identifiable solar flares, the value may not be a true representation of what flux is actually emitted by the Sun. The F11.1 SFU represents the average flux during the 3 h interval after the flare flux is removed. In this way it is somewhat similar to F10.7. However, it does not represent the effective flux impinging on the Earth during this time. The representation accuracy factor is the ratio of the F11.1 flux integrated over the 3 h period divided by the total flux. Using Figure 5 as an example, the integrated F11.1 flux is 1,849,038 and that for the total flux is 2,031,699. Their ratio is 0.91 which is the representation accuracy for the F11.1 SFU during that period, i.e., the F11.1 index only accounts for 91% of the radio flux during that 3 h period. In general, if no data is excluded in the computation of the F11.1 index (i.e., there are no observations greater than the average plus two standard deviations), the ratio is 1. On the other hand, if the entire 3 h period is influenced by a solar flare, the ratio would be zero. In the cases where the ratio is not 1, another parameter is computed. This parameter measures how long the F11.1 index is contaminated by flare activity within that 3 h window as a percentage of the 3 h. Using the previous example, the red trace lasted 888 s. Therefore, the F11.1 index was contaminated for 8% of the time period. Figure 6 shows the F11.1 index for all four observatories on 13 July In many cases, the values between the observatories are not the same (see Appendix A). Since the baseline for each observatory is different and that baseline can change during an observing day as well as change from one day to the next, the F11.1 index is slightly different between the observatories. This issue of different baselines was addressed by defining Palehua as the standard observatory. Then the following procedure was adopted to normalize the other stations to this standard. The best overlap time between Palehua and Learmonth is 0300 UT; their SFU values at this time are compared in Figure 7. The relationship between these two sets of SFU values is linear. Next a linear least squares fit ratio was computed between Palehua and Learmonth s 03UT values. This ratio was then used to scale all of Learmonth s values. The procedure was repeated between Learmonth and San Vito, and San Vito and Sagamore Hill. This procedure resulted in all F11.1 values being scaled to Palehua s values. The results of the linear least squares fit are shown in Table 2. All the slopes are close to one, as they should be in an ideal case. On a few days this rescaling procedure identifies a station as having too large a baseline offset. Under these conditions, Table 2. Results of the Linear Least Squares Fit for the Calibrations Between the Observatories That Observed the Sun at the Same Time Observatory Pair Slope Offset RMS Palehua Learmonth Learmonth San Vito San Vito Sagamore Hill of12

8 Table 3. Weighted Values for the RSTN Observatories 00Z 03Z 06Z 09Z 12Z 15Z 18Z 21Z Palehua Learmonth San Vito Sagamore Hill a rescaling parameter is devised for that specific station s F11.1 index. [17] The final step in the creation of the F11.1 index is devising a way to determine which observatory s value should be used as the valid 3 h index when there is overlap between the observatories. To solve this issue, for each observatory, the F11.1 value closest to the observatory s local noon is assigned a rank of 1. The F11.1 values immediately before and after this one is given a rank of 2 and so on. Table 3 shows these rank values for each station. To create the F11.1 time series, the computed value with the lowest ranking in that 3 h interval is selected. If some stations demonstrate the same ranking for some time periods, then the value for the station that is past its local noon is selected. For example, both Palehua and Learmonth s 03UT computation have a ranking of 2. However, Palehua s value falls after its local noon, while Learmonth sfallsbefore its local noon. Thus, Palehua s value was selected for the time series. Note this procedure generates a daily time sequence of eight, F11.1 values that represent values from the individual stations and not an average over available stations. 5. Results [18] Figure 8 shows two time series of the F11.1 index for the second half of The black dots are the scaled time series relative to Palehua. The open circles are the raw time series. In either case, the solar rotations can clearly be seen. Furthermore, the SFU differences in the two time series are less than the daily variability shown on the eight, 3 h values. This confirms that the difference in station absolute calibration is not the dominant source of variability in this data set. Note a few stations F11.1 values do fall outside this trend and as discussed in the prior section were rescaled for that specific day. [19] Figure 9 shows the representation accuracy and duration parameters for the F11.1 index over the same period as Figure 8. If the representation accuracy (black circle) is not 1.0, then that particular time period shows some flare activity. The corresponding duration parameter would then be larger than 0.0, indicating the fraction (0.0 to 1.0) of time during the previous 3 h that demonstrates flare activity. [20] Figure 10 shows GOES 10 X ray data, F10.7 index, F11.1 index, representation accuracy, and the duration parameter for the period 5 through 9 September This Figure 8. The raw and scaled time series for the 3 h index for the second half of The scaled F11.1 index value is a way to account for the different baselines for each observatory. All values are scaled to Palehua values. 8of12

9 f11 3-hr index Duration Accuracy /01 08/01 09/01 10/01 11/01 12/01 UT Time (month/day) Figure 9. In addition to the 3 h index, two other values are computed. These two values are (bottom) the representation accuracy of the F11.1 index, and if the index is inaccurate, (top) an estimate of how long the inaccuracy lasted. Figure 10. Solar activity from 5 September 2005 to 10 September The top panel shows the official daily F10.7 value obtained from the Space Weather Prediction Center; the next three panels show the F11.1 index, its representation accuracy, and the duration index. The last panel shows the GOES soft X rays. The observed flares in the GOES panel are clearly seen in the F11.1 index. 9of12

10 Figure 11. F11.1 versus F10.7 comparison during quiet solar conditions. The dashed line is the ideal result where both indices have the same value. The solid line is a linear least squares fit of the data. period contains several flares recorded by the GOES 10 X ray data. This graph raises several issues regarding radio bursts and flares. Similar intensity X ray flares do not result in similar F11.1 index values. Also, not every X ray flare has a response in the radio spectrum. Note that each day has a sequence of eight sets of values for the three F11.1 products and that they can be quite variable over a 24 h period. The F10.7 index daily value (top panel in Figure 10), does not, by design, account for any of these dynamic changes. On 7 September, a major flare begins at 1717 UT, (see bottom panel of Figure 10). The F11.1 index increases from a preflare value of 98 to over 120 SFUs, i.e., a 20% increase in 6 h, before returning to preflare values. The F10.7 index does not provide this dynamic 20% boost for one of two reasons: (1) that the flare occurs when the observatory is in darkness, and (2) the F10.7 algorithms specifically remove these flare enhancements. Although the F11.1 algorithms attempt to remove flares, clearly in this example, Figure 10, flare signatures are still present in the F11.1 and their full effect is captured by the associated representation accuracy and duration indices. [21] In situations where there is a need to account for the increased radio emissions due to a flare (i.e., include the flare in the index), dividing the F11.1 index by the representation accuracy yields the integrated SFU value for that 3 h period. [22] The F11.1 index was divided into flare and nonflare solar conditions by using a representation accuracy threshold of 0.9 or greater as nonflare solar values and values below this threshold as flare conditions. Figure 11 shows a comparison of the values for the nonflare conditions to the corresponding F10.7 values. The solid line is the least squares fit of the data and it shows the F11.1 values diverging at higher values. This comparison is between two different, although similar, radio wavelengths namely 10.7 and 11.1 cm. Hence, there is not an expectation that they are exactly correlated. However, the trend is reasonably linear with a slope slightly greater than 1. But of greater interest is that the diagram is populated with very distinct vertical lines of eight values. These correspond to the eight 3 h values of F11.1 in a given day plotted against that day s particular F10.7. If no F11.1 variability existed during a day, the eight points would all be superimposed on each other. That vertical streaks are present provides a measure of daily variability, i.e., the length of these streaks. [23] Figure 12 shows the comparison between F11.1 and F10.7 during active conditions. Although most of the flare effects were removed from the F11.1 index, the values were greater than the daily F10.7 (whose objective is the full removal of flares). The fact that the F11.1 is significantly larger than the F10.7 is not only due to these two data sets being measured at different wavelengths but also that in this first version of the F11.1 index, a flare threshold based on a yearly average and 2 standard deviations has been adopted. This method of Figure 12. F11.1 versus F10.7 comparison during active solar conditions. 10 of 12

11 Table 4. Ionospheric Impact of 20% Variability in Solar Ionization Radiation Proxy SFU N m F 2 (cm 3 ) TEC ( m 2 ) h m F 2 (km) Baseline , % increase simulation , Increase , Percent increase 20% 26.0% 29.5% thresholding does not remove flares completely, especially ones that have longer duration with slow rise and fall times. 6. Summary [24] The daily F10.7 index has been computed for over fifty years using radio observations taken at 2800 MHz and it is often the only solar input that goes into space physics models. The F11.1 is uniformly sampled over each day, whereas F10.7 is not. F11.1 is computed every 3 h and provides a measure of representation accuracy and duration resulting from the exclusion of solar flares. These three parameters (F11.1, representation accuracy, and duration factors) along with the faster cadence, could be an improved proxy for space physics models. [25] In ionospheric modeling, the time constants for photoionization and subsequent plasma production, loss and diffusion range from less than a second to many minutes. Hence, if the solar ionizing fluxes are changing significantly during a few hours due to flares and also overall background change, the dayside ionosphere will respond. Clearly, the radio fluxes at 10.7 and 11.1 cm do not cause ionization, however, the 10.7 cm flux has for over four decades been found to be a very good proxy for the ionizing EUV fluxes. Therefore, since F10.7 and F11.1 fluxes have been found to be satisfactorily related, (Figure 11), the F11.1 could also be used as a solar EUV proxy. Given its two attributes: (1) it is specified every 3 h and (2) using its representation accuracy to rescale F11.1 to provide the full 11.1 cm flux, the EUV flux is obtained and a significant step forward in solar input to ionospheric models is achieved. [26] For example, in Figure 10 the 3 h F11.1 flux on 7 September 2005 showed over a 20% change during the day, thus indicating that the solar EUV flux was not constant, but varied by at least that much. Two simulations of the ionosphere above Bear Lake, Utah (42.0 N, E) using the Utah State University (USU) Time Dependent Ionospheric Model (TDIM) were carried out for the September 2005 day. The first simulation used a proxy index value of 90 SFU and the second a 20% higher value, 108 SFU. These two simulations were for the same quiet geomagnetic conditions. The objective of these simulations was to determine the change in operationally useful ionospheric parameters: the peak density (N m F 2 ); peak height of the F layer (h m F 2 ); and the total electron content (TEC). Results of theses two simulations for local noon are given in Table 4. All three parameters have changed. The peak density and TEC have increased by 26% and 29.5% respectively in response to the 20% increase in the solar proxy. The increase in the peak height of about 4 km is relevant for radio communication that depends not only on the density of the F layer but also its height. [27] This new F11.1 suite of indices still needs refinement. The algorithm to exclude solar flares is very rudimentary. Figure 12 clearly shows that some flare effects remain in the computed index. Short spikes that do not affect the baseline are not difficult to filter out, but events with long decays, or gradual rise and fall events following flares are not excluded from these computations. The yearly average and two standard deviations were suitable in this six month study. However, the yearly average may not be the best approach for a real time index or for larger studies. A running average may be a better approach and will be tested in the near future. The method to create the F11.1 time series is static, i.e., the same observatory is used every day for a particular time period. In reality, maintenance, calibration, or operational issues could corrupt the data used to compute the index. In those cases, it would be better to use a different observatory. The scaling between observatories accounted for some of these issues, but it could and should be handled more rigorously. [28] In summary, the results showed that the F11.1 index tracks the solar rotations and that this suite of three indices provides a measure of representation accuracy and duration relative to the exclusion of solar flares. The next step would be to see how this index can be used as the solar input in ionospheric physics models. Currently, as indicated in the body of this text, many ionospheric physics models use the daily F10.7 index as a proxy solar input for the model. This F10.7 index does not change throughout the day and does not include compensation for the effects of flares. Another key task will be to compare the F11.1 suite of indices with the better than 1 min 24/7 solar EUV spectral measurements being made by the Extreme Ultraviolet Variability Experiment (EVE) instrument on the NASA Solar Dynamics Observatory (SDO). Appendix A [29] The one second RSTN data used in this study have undergone calibration to generate the Solar Flux Unit (SFU) values. This occurs in house prior to release of the data product. A full description of these calibration is beyond the purpose of this study, but an overall appreciation of the calibration steps is in order. Each RSTN site is responsible for its own daily calibration to establish the 11 of 12

12 solar flux unit conversion from electronic signal voltage. This calibration is either performed before sunrise or at local noon. Step one in the daily calibration is to move the antenna 2 h west of the Sun on its daily track. This provides a cold sky reading. Then a calibration signal is turned on. The calibration signal is produced by a solidstate device that generates a RF noise at a known and constant output. This calibration signal is needed to compensate for the radiometer drifts and hence is used to provide absolute and linearity calibration. The cold sky reading is then used as the zero solar flux and all signals are referenced to this voltage. For the data sets used in this study the resetting of the antenna to track the Sun was carried out manually. Once the Sun was reacquired, automated tracking was used to follow the Sun. [30] Each day the station analyst inspects the days recalibrated SFU for consistency with prior days and with the other three RSTN stations. Any value that is deemed to be more than 10% different from the others is flagged and recalibrated. All values within a 10% threshold are flagged good while the others are flagged uncertain. This level of quality control and calibration meets the Air Force RSTN needs for the one second solar radio product. A more extensive discussion of these calibration procedures is given by Acebal [2008]. [31] For scientific usage it is probable that a more detailed calibration procedure would be needed. However, for this exploratory study investigating the continuous monitoring of the Sun at radio wavelength these data are adequate. [32] Acknowledgments. This research was supported by NSF grants ATM and AGS and NASA grant NNG05GJ48G to Utah State University. Ariel Acebal was sponsored by the Air Force Institute of Technology. The views expressed in this paper are those of the authors and do not necessarily reflect the official policy or position of the Air Force, the Department of Defense, or the U.S. Government. The authors wish to thank one of the referees for the insights provided in contributing to this paper. References Acebal, A. (2008), Extending F10.7 s time resolution to capture solar flare phenomena, Ph.D. dissertation, 186 pp., Utah State Univ., Logan. Bossy, L. (1983), Solar indices and solar UV irradiances, Planet. Space Sci., 31, , doi: / (83) Dulk, G. A. (1985), Radio emission from the Sun and stars, Annu. Rev. Astron. Astrophys., 23, , doi: /annurev.aa Kundu, M. R. (1965), Solar Radio Astronomy, 660 pp., Interscience, New York. Lean, J., and C. Frohlich (1998), Solar total irradiance variations, in Synoptic Solar Physics, edited by K. S. Balasubramaniam, Astron. Soc. Pac. Conf. Ser., 140, Lee, J., and D. E. Gary (2000), Solar microwave bursts and injection pitch angle distribution of flare electrons, Astrophys. J., 543, , doi: / Pacholczyk, A. G. (1970), Radio Astrophysics: Nonthermal Processes in Galactic and Extragalactic Sources, Freeman, San Francisco, Calif. Raulin, J. P., and A. A. Pacini (2005), Solar radio emissions, Adv. Space Res., 35, , doi: /j.asr Tanaka, H., et al. (1973), Absolute calibration of solar radio flux density in the microwave region, Sol. Phys., 29, , doi: / BF Tapping, K. F. (1987), Recent solar radio astronomy at centimeter wavelengths: The temporal variability of the 10.7 cm flux, J. Geophys. Res., 92(D1), , doi: /jd092id01p Tapping, K. F., and D. P. Charrois (1994), Limits to the accuracy of the 10.7 cm flux, Sol. Phys., 150, , doi: /bf Tapping, K. F., and B. DeTracey (1990), The origin of the 10.7 cm flux, Sol. Phys., 127, , doi: /bf Tapping, K. F., and K. L. Harvey (1994), Slowly varying microwave emissions from the solar corona, in The Sun as a Variable Star: Proceedings of IAU Colloquium 143, editedbyj.m.papetal.,pp , Cambridge Univ. Press, Cambridge, U. K. Treumann, R. A. (2006), The electron cyclotron maser for astrophysical application, Astron. Astrophys. Rev., 13, , doi: / s y. Woods, T. N., et al. (1998), TIMED solar EUV experiment, Proc. SPIE, 3442, , doi: / A. O. Acebal, Air Force Institute of Technology, Wright Patterson Air Force Base, Dayton, OH , USA. J. J. Sojka, Center for Atmospheric and Space Sciences, Utah State University, 4405 Old Main Hill, Logan, UT , USA. (jan.sojka@usu.edu) 12 of 12