Artep, Inc., at Naval Research Laboratory, 4555 Overlook Avenue, SW, Code 7600A, Washington, DC 20375; and K. P.

Transcription

1 The Astrophysical Journal, 574: , 2002 July 20 # The American Astronomical Society. All rights reserved. Printed in U.S.A. A COMPARISON BETWEEN CORONAL EMISSION LINES FROM AN ISOTHERMAL SPECTRUM OBSERVED WITH THE CORONAL DIAGNOSTIC SPECTROMETER AND CHIANTI EMISSIVITIES E. Landi and U. Feldman Artep, Inc., at Naval Research Laboratory, 4555 Overlook Avenue, SW, Code 7600A, Washington, DC 20375; landi@poppeo.nrl.navy.mil and K. P. Dere Naval Research Laboratory, 4555 Overlook Avenue, SW, Code 7600A, Washington, DC Received 2001 December 20; accepted 2002 March 22 ABSTRACT The present paper compares off-disk spectral observations of the solar corona in the ranges and Å with theoretical emissivities calculated using the CHIANTI database. The observed spectra were recorded by the Coronal Diagnostic Spectrometer instrument on board the Solar and Heliospheric Observatory using the normal-incidence portion of the instrument. Using line-ratio techniques, we first measure the electron temperature and density in the emitting region, verifying that it is nearly isothermal. Next, we use an emission-measure analysis to compare measured spectral line intensities with predictions from the CHIANTI database. This comparison allows us to assess the quality of the CHIANTI data for the brightest coronal lines in the and Å spectral ranges. As a result, we are able to (1) select lines and ions for which the agreement between theory and observation is good, (2) identify a few lines that are blended, and (3) stress inconsistencies between a few lines and theory, thus showing where improvements to atomic data and transition probabilities are necessary. Subject headings: atomic data line: identification Sun: corona Sun: UV radiation 1. INTRODUCTION The CHIANTI database (Dere et al. 1997, 2001; Landi et al. 1999) aims to provide the solar and astrophysics communities with the best available atomic data necessary to analyze spectra emitted by abundant atomic species in solar and astrophysical plasmas. The atomic data contained in the database are applicable to a range of physical conditions but are most directly applicable to optically thin plasmas in collisional-ionization equilibrium. The CHIANTI database is ideally suited for measuring the physical properties of astrophysical plasmas by means of comparisons between measured spectral line intensities and theoretical calculations. The accuracy of the diagnostic results is crucially dependent on the quality of the atomic data used. Therefore, an important task of the CHIANTI project is to evaluate the accuracy of the data in the database. The accuracy of theoretical calculations can be best assessed from a direct comparison between the theoretical predictions and observations in well-controlled laboratory experiments. Unfortunately, it has been difficult to duplicate in well-diagnosed laboratory experiments conditions that exist in the astrophysical plasmas for which CHIANTI data are normally used. As a result, there is only a small number of verified theoretical atomic physics calculations for ions in the CHIANTI database. The few cases in which such comparisons exist involve ions found in low-ionization stages, typical of solar and stellar chromospheres and transition regions. In the cases involving plasmas with higher electron temperatures, T e ð1 2Þ10 6 K or more, few if any are available. As a result, most of the conclusions regarding the plasma properties in the solar corona that are reported in the literature are based on comparisons between 495 observations and mostly unverified theoretical calculations. Recently, Young, Landi, & Thomas (1998) carried out a comparison between CHIANTI predictions and observations of a solar active region obtained in 1989 from the Solar EUV Rocket Telescope and Spectrograph (SERTS; Thomas & Neupert 1994). Such a comparison involved mostly optically allowed lines in the Å spectral range. In general, a very good agreement was found, although in a few cases, Young et al. (1998) found inconsistencies that could not be attributed either to blending or to the instrument intensity calibration. Landi, Feldman, & Dere (2002) compared CHIANTI line emissivities with observations obtained with the Solar Ultraviolet Measurements of Emitted Radiation (SUMER; Wilhelm et al. 1995) instrument on board the Solar and Heliospheric Observatory (SOHO); these observations were associated with plasma high above the solar equatorial west limb. The comparison involved coronal lines (T K) and showed an overall good agreement with only a few exceptions. The Coronal Diagnostic Spectrometer (CDS) on board SOHO is an imaging spectrometer aimed at producing stigmatic spectra of selected regions of the solar spectrum in six spectral windows in the extreme-ultraviolet Å wavelength range. CDS consists of two separate spectrometers sharing the same telescope and operating separately. The Normal Incidence Spectrometer (NIS) observes in two spectral ranges (NIS 1: Å; NIS 2: Å); the Grazing Incidence Spectrometer (GIS) covers four spectral ranges: , , , and Å (full details of the CDS instrument can be found in Harrison et al. 1995). The aim of the present work is to compare CHIANTI emissivities with CDS-NIS observations of a quiet-sun region above the solar east limb. The choice of an

2 496 LANDI, FELDMAN, & DERE Vol. 574 off-disk region devoid of any visible plasma structure has been made in order to observe a spectrum emitted by a nearly isothermal solar region, so that the emission-measure method of analysis used by Landi et al. (2002) can be applied to the present data set. In x 2 we provide information on the spectra used in the present study and on the physical properties of the observed plasma. In x 3 we carry out traditional line-ratio diagnostics on the emitting plasma, while in x 4 we describe the results of the emission-measure analysis. In x 5 we assess the accuracy of the CHIANTI database, and in x 6 we include the summary of the analysis and the conclusions. 2. OBSERVATIONS In order to maximize the number of spectral lines available, the observational data used for the present analysis covered the full NIS spectrum. A field of view was chosen that included the solar limb and the lower corona in order to be able to select a quiet plasma region with nearly isothermal properties. Thus, the selected position needed to be free from active regions. The data used were taken on 1997 March 13. The center of the field of view had the heliocentric coordinates , 3>1, and only spectra from the central of the CDS slit were transmitted to the ground. The field of view had dimensions of For details, see Figures 1 and 2. Five rasters of the same region were taken, and each of them required an observing time of approximately 2 hr and 15 minutes, for a total observing time of 11 hr and 15 minutes. The data have been reduced, and cosmic rays have been removed, using standard routines available in the CDS software. Fig. 1. CDS field of view superposed on a He ii 304 image obtained with the SOHO EIT. The EIT 304 Å channel observes emission from plasma at around K. The slit position selected in the present study is also displayed as the line dividing the CDS field of view. Fig. 2. Intensity maps of the observed region from CDS, taken with the O v 629 and Mg x 624 lines. O v is formed at around 2: K and Mg x is formed at around 1: K. The black line indicates the position of the limb, as seen with the He i 584 line, and the white line indicates the slit position in the current study Selection of a Nearly Isothermal Region A nearly isothermal plasma can be selected by looking at a small portion of the corona observed above the limb in a quiet solar region. Observations sufficiently high above the limb (20 00 ) result in the cooler transition region and chromospheric plasmas being excluded from the line of sight. However, line intensity quickly decreases with distance above the limb because of the decrease in electron density. Therefore, when looking for an isothermal region, a compromise must be reached between finding a region sufficiently high above the limb and retaining a large number of strong spectral lines for analysis. Also, it is necessary to minimize the contribution of transition-region lines to the blending coronal lines. In order to avoid transition-region plasma, the intensity of the following transition-region lines was measured as a function of distance from the limb: He i 515.6, He i 522.2, He i 537.0, O iv 554.5, and O v The optimum region was selected to be where the intensities of these lines were as small as possible, while the signal-to-noise ratio for coronal lines was as high as possible. The optimum choice was solar X ¼ , i.e., above the limb. The slit position in this study is also displayed in Figures 1 and Temporal Variations of the Observed Region We followed two different approaches to check the temporal stability of the observed region during data acquisition. We first inspected EUV Imaging Telescope (EIT) fulldisk images of the Sun taken with approximately a minute cadence during the whole CDS observation. Full- Sun images available in the He ii 304 channel monitor the plasma at transition-region temperatures. These indicate whether any prominences are present in the field of view. The He ii images show that no chromospheric emission was present at the selected slit position. Also, a few EIT full-sun images obtained with the 171, 195, and 284 Å filters, sampling coronal temperatures, were taken during the CDS observation with a reduced cadence. They show that the coronal plasma in the CDS field of view was devoid of structures and was remarkably stable in time. The second approach consisted of measuring the line intensities of a few lines emitted by ions of the same element and determining the electron temperature from their intensity ratios. Such measurements are carried out under the

3 No. 1, 2002 COMPARISON BETWEEN CHIANTI AND CDS 497 TABLE 1 Line Intensity Ratios and Temperature Measurements as a Function of Time Mg x 624/Mg ix 368 Mg vii 367/Mg x 624 Mg vii 367/Mg ix 368 Observation Time (UT) Ratio log T (K) Ratio log T (K) Ratio log T (K) 11: : : : : Note. The time corresponds to the start of the observation for each raster; the uncertainty on each ratio is around 10%, and the uncertainty in the log T value is 0.01 dex. assumption of ionization equilibrium, by comparing observed line ratios with theoretical estimates calculated as a function of electron temperature using the CHIANTI database. These measurements have been taken for each of the five CDS exposures by averaging the spectrum along the slit and fitting a Gaussian line profile to each slit-averaged spectral line using the program described by Brooks et al. (1999). Fluxes were calibrated to ergs cm 2 s 1 sr 1 using the revised CDS calibration currently available in the CDS software. The measurements are reported in Table 1. The lines considered were emitted by Mg vii, Mgix, and Mg x: their ratios are remarkably constant and provide very consistent temperature measurements. The temperature seems to be very stable in time and gives us additional confidence that during the CDS observation time, no significant change occurred in the emitting plasma. Moreover, the temperature indicated by each ratio is the same, despite the fact that the temperatures of maximum abundance for each of the ions used are different (Mg vii: 6: K; Mg ix:1: K; Mg x:1: K, according to Mazzotta et al. 1998). This provides the first evidence that the selected region is nearly isothermal. Since the observed region showed no significant change during the whole range of the observations, we have taken the average of the five slit-averaged spectra at each pixel location in order to improve the signal-to-noise ratio. The final data set therefore consists of a single time- and slitaveraged one-dimensional spectrum. Spectral line intensities were measured by fitting Gaussian line profiles to the lines found in the one-dimensional spectrum. Spectral line identifications have been made on the basis of the CDS spectral atlas reported by Brooks et al. (1999). Fluxes were calibrated to ergs cm 2 s 1 sr 1 using the revised CDS calibration currently available in the CDS software. Line identifications and intensities are reported in Tables 2 and PLASMA DIAGNOSTICS USING LINE RATIOS To check whether the selected region is isothermal and to measure its electron temperature and density, we first carry out standard line-ratio diagnostics. The results are also necessary for the application of the emission-measure technique. In the present data set, lines from several consecutive stages of ionization of the same element are observed, and they are used to measure the electron temperature of the plasma. This technique relies on the assumption that the plasma is isothermal and in collisional-ionization equilibrium. Also, electron temperature is measured using available Si x and Si xi ratios of forbidden or intercombination lines with optically allowed lines. Electron density is TABLE 2 Measured Wavelengths and Intensities of Lines Observed in the NIS 1 Band ( Å) Wavelength (Å) Intensity (ergs cm 2 s 1 sr 1 ) Fe xi Mg viii Mg viii Si viii Mg viii Si viii Mg viii Mg vii Si viii Al viii Al viii Al x Mg viii Fe xii Mg viii Fe xi Si ix Si ix Fe x Fe xii Si x Fe xiii Fe xi Si ix Fe xii Fe xi Al vii Si x Fe xi Al vii Fe x Fe xi Mg vii Fe xii Mg vii Fe x Mg vii Mg ix Fe xi

4 498 LANDI, FELDMAN, & DERE Vol. 574 TABLE 3 Measured Wavelengths and Intensities of Lines Observed in the NIS 2 Band ( Å) Wavelength (Å) Intensity (ergs cm 2 s 1 sr 1 ) Ar viii Si xii Ar viii Al xi Ca x Al xi Ca x Si xi Ar vii Si xi Mg x Si x K ix Mg x Note. A few second-order lines from Si ix, Si x, Si xi, and Fe xiv were present in the spectrum, but because of their weakness, they have not been used in the present work, and so they have not been reported in this table. measured using density-sensitive intensity ratios from lines emitted by the same ion. Plasma diagnostic results are displayed in Tables 4 and 5 for temperature and density. The brightest line has been used for each ion. Fe xii and Si viii lines have not been used for temperature measurements because they are also strongly density sensitive; moreover, in the case of Fe xii, unresolved problems in theoretical emissivities have been found by Binello et al. (2001). Emissivities of the two strongly density-dependent Si x lines at and Å have been summed together, so that the resulting emissivity is only weakly density dependent. Uncertainties in the measured values are derived from uncertainties in the line TABLE 4 Electron Temperature Resulting from Line Intensity Ratios Wavelengths (A ) Observed Ratio log T (K) Al viii/al x / Al xi/al viii / Al xi/al x / Mg vii/mg viii / Mg vii/mg ix / Mg vii/mg x / Mg viii/mg ix / Mg viii/mg x / Mg x/mg ix / Si ix/si x /( ) Si xi/si ix / Si xii/si ix / Si xi/si x /( ) Si xii/si x /( ) Si xii/si xi / Fe x/fe xi / Fe xiii/fe x / Fe xiii/fe xi / Si x/si x /( ) Si xi/si xi / <6.38 TABLE 5 Electron Densities Derived from Line Intensity Ratios Wavelengths (A ) intensities; these are calculated considering the uncertainty in the CDS intensity calibration. In the case of the temperature measurements, the error bars are probably underestimated, since the uncertainty in the ionization and recombination rates used to calculate the ion population at equilibrium may provide a significant contribution to the uncertainty in the theoretical ratio. Table 4 shows that most of the temperature values fall in the range 5:97 < log T < 6:05, thus indicating that the plasma is nearly isothermal and providing a rather precise temperature measurement. The uncertainty in the Si xi 604/580 ratio is too large to provide a precise temperature measurement. It is interesting to note that the temperature provided by the intercombination/allowed line ratio Si x 621/( ) is close to the temperatures measured with the other line ratios: this indicates that the atomic data for this ion are rather accurate. The electron density values reported in Table 5 show agreement around log N e ¼ 8:50 (N e in cm 3 ), and this value is adopted throughout the rest of the present work. 4. PLASMA DIAGNOSTICS USING THE EMISSION-MEASURE ANALYSIS 4.1. Method of Analysis The emission-measure method of analysis was first introduced by Pottasch (1963) and has been revised and improved over the years (i.e., Harrison & Thompson 1991). A recent description of the method can be found in Landi et al. (2002); here we provide only a very brief outline. The intensity of an optically thin emission line observed at distance d can be written as I ji ¼ 1 4d 2 Z V Observed Ratio log N e (cm 3 ) Mg vii / :50 þ0:2 0:3 Si ix / Si x / G ji ðt; N e ÞNe 2 dv ; ð1þ where N e is the electron density, V is the emitting volume along the line of sight, and G ji ðt; N e Þ is the contribution function of the emitting line, usually dependent on both electron temperature and density. If the electron density (N e ) and temperature (T c ) are constant in the emitting volume V along the line of sight, we have I ji ¼ 1 4d 2 G jiðt c ; N e ÞhEMi; ð2þ where hemi ¼ R V N2 e dv ¼ Ne 2 V is the emission measure of the plasma. In this case, the emission measure can be directly evaluated as hemi ¼ 4d 2 I ji G ji ðt c ; N e Þ : ð3þ Under the assumption of constant N e and T c within the

5 No. 1, 2002 COMPARISON BETWEEN CHIANTI AND CDS 499 Fig. 3. Plot of I= G ji ðt; N e Þ functions for all lines emitted by ions not belonging to the Li, N, and Na isoelectronic sequences. The dashed lines represent the derived logðt DTÞand log½em DðEMÞŠ. emitting region, this quantity should be the same for all the observed lines. The diagnostic method consists of calculating the function hemðtþi, hemðtþi ¼ 4d 2 I ji G ji ðt; N e Þ ; h EMðT cþi ¼ hemi ; ð4þ defined as a function of electron temperature, using the observed intensities I ji of each observed line and a value of the electron density derived from line-ratio techniques. When all the hemðtþicurves are displayed in the same plot as a function of temperature, these curves should intersect at a common point ðt c ; hemiþ. Given the experimental uncertainties, this defines a narrow range in the T c -hemi space. Examples are given in Figures 3 5. In each of these figures, the crossing point and its uncertainties are determined as the region in which the largest number of the hemðtþicurves meet. The common crossing point directly determines the plasma temperature T c, hemi, and their uncertainties. If Fig. 4. Plot of I= G ji ðt; N e Þ functions for all lines emitted by N-like ions. The dashed lines represent logðt DTÞand log½em DðEMÞŠ. Fig. 5. Plot of I= G ji ðt; N e Þ functions for all lines emitted by the Nalike (top) and Li-like (bottom) ions. The solid lines represent logðt DTÞ and log½em DðEMÞŠin each figure. some curves do not cross the common intersection point, either the corresponding lines are blended by some other transition or the atomic data used to calculate their contribution function and hemðtþicurves are inaccurate. In case of systematic differences, calibration errors may be involved. Moreover, if all the lines of a given element miss the common crossing point and cross each other at some other value of hemðt c Þi, then the adopted abundance in the calculation of the contribution function is most likely incorrect, and its value can be modified according to the differences in the hemivalues. In the present work, the atomic data in the CHIANTI database was used to calculate relative level populations. References of the original papers are reported in Table 6. fractions from Mazzotta et al. (1998) and the element abundances of Feldman & Laming (2000) are used to derive the contribution function of each observed line. The Feldman & Laming (2000) photospheric abundances of the elements whose first ionization potential (FIP) is lower than 10 ev were increased by a factor of 3.5 to yield coronal abundances, as indicated by Feldman et al. (1999).

6 500 LANDI, FELDMAN, & DERE Vol. 574 TABLE 6 Table of s Used in the Present Study Sequence Radiative Data Reference Collisional Data Reference Sequence Radiative Data Reference Collisional Data Reference Li... Mg x 1, 2 1 N... Al vii Al xi 1, 2 1 Si viii Si xii 1 1 Na... Ar viii 22, Be... Mg ix 3 4,5 Kix Al x 6, 7 6 Ca x 22, Si xi 3, 5 6 Mg... Ar vii B... Mg viii 8, 9, 10 10, 11 Si... Fe xiii 3 33, 34 Si x 8, 9, 10, 12, P... Fe xii 24 25, 26 C... Mg vii 14, S... Fe xi 3 27, 28 Al viii 16, Cl... Fe x 3 29, 30, 31 Si ix 18 18, 19 References. (1) Zhang, Sampson, & Fontes 1990; (2) Martin et al. 1993; (3) P. R. Young 1996, 1998, 1999, 2000, unpublished calculations; (4) Keenan et al. 1986; (5) Sampson, Goett, & Clark 1984; (6) Zhang & Sampson 1992; (7) Interpolated from (2); (8) Zhang & Sampson 1996; (9) Bhatia, Feldman, & Seely 1986; (10) D. H. Sampson & H. L. Zhang 1995, unpublished calculations; (11) Zhang, Graziani, & Pradhan 1994; (12) Flower & Nussbaumer 1975; (13) Saha & Trefftz 1983; (14) Bhatia & Doschek 1995b; (15) Storey & Zeippen 2000; (16) Dankwort & Trefftz 1978; (17) Galavis, Mendoza, & Zeippen 1997; (18) Bhatia & Doschek 1993; (19) Aggarwal 1983; (20) Interpolated from (21); (21) Bhatia & Mason 1980; (22) Sampson, Zhang, & Fontes 1990; (23) Wiese, Smith, & Glennon 1966; (24) Binello, Mason, & Storey 1998a, 1998b; Binello et al. 2001; (25) Tayal, Henry, & Pradhan 1987; (26) Flower 1977; (27) Bhatia & Doschek 1996; (28) Gupta & Tayal 1999; (29) Bhatia & Doschek 1995a; (30) Pelan & Berrington 1995; (31) Malinovsky, Dubau, & Sahal-Brechot 1980; (32) Christensen, Norcross, & Pradhan 1986; (33) Gupta & Tayal 1998; (34) Fawcett & Mason TABLE 7 Coronal Abundances Adopted in the Present Study Element O... Ne... Na... Mg... Al... Si... S... Ar... K... Ca... Fe... log½nðxþ=nðhþš The elemental abundances used in this study are listed in Table 7. The adopted electron density is 3: cm 3 (log N e ¼ 8:5), as measured in x 3. In the present study, the scatter of the curves from a common crossing point has been taken as a measure for the quality of the atomic data and transition probabilities in the CHIANTI database and/or the ion fractional abundances of Mazzotta et al. (1998). Moreover, since the log of the electron temperature of the emitting plasma has been independently measured in x 3 to be 6:02 0:04 (the uncertainty is given by the standard deviation of all the measurements), we expect that the T c value of the common crossing point should agree with this value within the experimental uncertainties. As a by-product, the adopted elemental composition of the emitting plasma is also verified. Landi et al. (2002) applied this same method to SUMER spectra, which included several lines also observed in the NIS 2 section of the present data set. Therefore, the present study allows us to check the SUMER results and to extend the comparison to the optically allowed lines at wavelengths shorter than the SUMER wavelength range. Possible agreement between CDS optically allowed lines and lines also observed by SUMER and classified by Landi et al. (2002) as free of problems would suggest agreement between CDS lines and the other SUMER lines that also showed no problems. This would provide a link between CDS results for optically allowed lines and SUMER results for forbidden/ intercombination lines; however, in order to provide accurate comparison, it is necessary to use simultaneous observations in the CDS and SUMER wavelength range, which are not currently available Results The analysis of SUMER data has highlighted two main areas of disagreement from the results provided both by most of the ions and by independent measurements with other methods. One was represented by the lines emitted by ions in the nitrogen isoelectronic sequence: these lines indicated a higher temperature than all the others but yielded a similar value of the emission measure. The other was composed of the lines of the lithium and sodium isoelectronic sequence, which indicated a similar temperature to the most common value but provided emission-measure values lower by a factor of around 2 than the value provided by all the other lines. This fact was ascribed to inaccuracies in the ion fractions for these two sequences. In order to check the results obtained with SUMER, we have divided the CDS data set into three groups: N-like lines, Li-like and Na-like lines, and lines from all the other isoelectronic sequences. The diagnostic procedure has been applied to each of these three classes separately; the results are displayed in Figures 3 5. From these figures, the following lines have been removed, since they showed striking disagreement and caused unnecessary confusion: Fe xi Young et al. (1998) report this line as blended with an unidentified line. Fe xii Binello et al. (2001) report this line as blended with an unidentified line.

7 No. 1, 2002 COMPARISON BETWEEN CHIANTI AND CDS 501 Ar viii The intensity of this line is much higher than predicted, suggesting the presence of an unidentified blend. Ar vii This line is blended with a second-order Si ix transition. Fe xi This line is blended with a Mg vi multiplet. Al viii This line is blended with a Cr xiii transition. Fe xi The observed intensity of this line is higher than predicted. This behavior is unclear, as no blends are present for this line, and no problems were found by Young et al. (1998) for this line. Measurements of temperature and emission measure are made by visually identifying the region in which all curves meet; the extremes of this region are used to calculate the average temperature and its uncertainties. Figure 3 includes all the lines emitted by ions not belonging to the Li, N, and Na sequences. It shows that despite a certain amount of scatter from a few lines (mainly Mg vii lines, the coldest in the data set), a common crossing region can be defined that includes all the displayed lines. This result shows that the plasma can be considered as isothermal. The electron temperature indicated by the crossing region ranges between log T ¼ 6:00 6:05 ½ð1:0 1:1Þ 10 6 KŠ, in agreement with line-ratio results. The log emission measure determined from the crossing point is log EM ¼ 43:35 0:15. Results from the N-like lines, displayed in Figure 4, are more uncertain, as lines from only two ions (Al vii and Si viii) are available; moreover, their emission-measure curves are both steep at the temperature of the emitting plasma, so their crossing point is more uncertain. However, Figure 4 indicates a crossing point with log T ¼ 6:10 0:03 and log EM ¼ 43:45 0:15. Figure 5 displays the results for the Na-like and Li-like lines. The crossing point of the Na-like lines, indicated by the solid lines in Figure 5 (top), is very ill defined, since each line exhibits a rather flat curve at coronal temperatures. The crossing point can be identified in the very broad region log T ¼ 5:85 6:07, log EM ¼ 43:05 43:60. It is interesting to note that in the temperature range log T ¼ 6:00 6:05, indicated by the ions not belonging to the Li, N, and Na sequences, the Na-like lines indicate an emission-measure value of log EM ¼ 43:35 0:15, in agreement with the estimate of the former lines. The crossing point for the Li-like lines, displayed in Figure 5 (bottom), indicates a significantly lower emission measure and a slightly higher temperature: log T ¼ 6:03 6:08, log EM ¼ 43:05 0: DISCUSSION 5.1. Lithium and Sodium Isoelectronic Sequence Figure 5 (bottom) confirms the results obtained using the SUMER instrument. Landi et al. (2002) reported that Lilike lines (in part, the same lines used in the present work) indicated an emission-measure value lower by a factor of around 2 than the value obtained with the rest of the lines. The same factor is confirmed in the present work. The fact that the crossing point is well defined indicates that the cause of the discrepancy is to be found in the ion fractions of the lines rather than in the excitation and deexcitation rates. This is for two main reasons: (1) the two lines of each ion of the sequence agree with each other, so that the collisional or radiative rates for the populations of the upper 2 P levels of the Li-like doublet seem to have no problem throughout the sequence; (2) the fact that the whole isoelectronic sequence is self-consistent but indicates a different crossing point suggests that some ionization or recombination rate is inaccurate for the whole sequence, biasing the ion-fraction calculations for all the ions in the same direction, thus resulting in an apparent higher ion population than observed. The Na-like lines instead behave differently from the findings of Landi et al. (2002) with SUMER. Even though the lines used are the same, the present results provide an emission-measure value in close agreement with the measurement from all the other ions. It is to be noted, however, that this value is rather uncertain, as the crossing point reported in Figure 5 (top) is ill defined. A broader wavelength range might have allowed us to observe hotter lines such as Ti xii and Cr xiv that would have helped in defining the common crossing point. However, the meaning of the agreement found in the present work, in contrast to the disagreement found by Landi et al. (2002), is not clear. The presence of residual cold plasma in the field of view is unlikely to be the cause of this discrepancy, as it would also have affected the results for ions of other sequences, which on the contrary show no problems. The only possible explanation seems to be related to the different nature of the emitting region analyzed in the two studies or to the slightly different temperature found in the two emitting plasmas. It is interesting to note that even if the curves in Figure 5 (top) do not provide a definite crossing point, they nevertheless agree with each other, thus giving confidence that no blends or inaccuracies in the collisional excitation rates are present Nitrogen Isoelectronic Sequence Only two N-like ions are present in the CDS-NIS wavelength range that can be used for the present study, and this makes the conclusions somewhat more uncertain. However, the present work confirms the problems found with the N- like lines by Landi et al. (2002): the temperature indicated by the crossing point is too high. In the latter work, this problem was due to the forbidden transitions within the ground configuration; the present work confirms the disagreement using optically allowed lines. The origin of this disagreement may be due to a number of problems. In the calculation of line emissivities with the CHIANTI database, Version 3.02, the atomic model includes only the lowest 13 levels. Proton rates and photoexcitation from the background photospheric radiation are neglected, and resonances in the collision strengths for transitions within the ground configuration have been included only for Si viii. Recently, Bhatia & Landi (2002) have carried out extensive distorted-wave calculations for collision strengths for transitions within the lowest 72 Si viii levels and have also provided proton excitation rates. Comparison between line emissivities calculated with this extended data set and those with the CHIANTI, Version 3.02 data set have given the following results: 1. Photoexcitation effects are negligible. 2. Proton rates have only a small effect on the population of the ground levels and a negligible one on the population of excited levels.

8 502 LANDI, FELDMAN, & DERE Vol Cascade effects from n ¼ 3 levels (included in Bhatia & Landi 2002) are negligible for densities greater than 10 8 cm Effects of resonances are important, giving rise to differences of up to 50% in the emissivities coming from the first excited configuration, and higher for the forbidden lines within the ground configuration. Considering these results, we argue that the discrepancy that is present in the N-like sequence is due to the neglect of the resonances in the collision excitation rates for transitions within the ground configuration. However, to date, collisional data for N-like ions including resonances have been made available in the literature for Si viii and S x only, so that it is not possible to confirm this. It is highly recommended that new calculations for ions in the nitrogen isoelectronic sequence be carried out in the future Plasma Properties The results of the present analysis allow us to draw the following conclusions on the physical parameters of the emitting plasma: 1. The electron temperature of the plasma is log T e ¼ 6:03 0:03 (T in K). 2. The emission measure of the emitting plasma is log EM ¼ 43:35 0:15, given by all lines not belonging to the Li-, N-, and Na-like sequences. It is very interesting to note that the temperature values obtained from the emission-measure technique are in good agreement with estimates provided by the line-intensity ratios used in x 3. This gives confidence both in the reliability of the diagnostic techniques used and in the quality of the adopted atomic data. Temperature diagnostics have been carried out using line ratios from consecutive stages of ionization of the same element and therefore represent a crude check of the quality of the ion fractions for each of the elements used. The agreement found between the results obtained in xx 3 and 4 supports the accuracy of the adopted ion fractions, From the value of the emission measure, it is possible to give a rough estimate of the average length occupied by the emitting plasma along the line of sight. The emission measure can be approximated as Z hemi ¼ Ne 2 dv N2 e AL ; ð5þ V where A is the area viewed by a single CDS pixel, N e is the average plasma electron density, and L is the average lineof-sight length. Taking into account the size of a single CDS pixel of 2>032 1>68 (slit 1), where 1 00 ¼ 7: cm, and using N e ¼ 3: cm 3, we have A ¼ 1: cm 2 ; L ¼ 1: cm ¼ 0:19 R : The size of the plasma s line-of-sight length is comparable to the solar radius. We have tried to compare this value with visual estimates carried out on images of the emitting region taken with EIT, the Ultraviolet Coronagraph and Spectrometer (Kohl et al. 1995), and the Large Angle and Spectrometric Coronagraph Experiment (Brueckner et al. 1995) in EUV and visible light. However, the Sun was very quiet ð6þ and unstructured when the observations were taken. The value of L corresponds to the path length of the line of integration through a homogeneous corona. The close agreement of lines emitted by ions of different elements within each group of lines shows that the elemental abundances used throughout the whole of this work are self-consistent. However, only low-fip lines (with the only exception of a single Ar viii line) have been used in this work, so the amount of the FIP effect present in this data set cannot be accurately determined. 6. CONCLUSIONS In the present work, we have carried out a comparison between CDS observations of coronal lines from an off-disk quiet solar spectrum and CHIANTI emissivities. The comparison has allowed us to check the quality of the CHIANTI atomic data in the and Å wavelength ranges and to measure the basic properties of the emitting plasma. The diagnostic results have also been compared with those obtained using line-intensity ratios. Overall, a good agreement is found between lines of different ions and different isoelectronic sequences, thus giving confidence in the quality of the data for these wavelength ranges. Our results confirm, for the Å wavelength range, the results of Young et al. (1998). Discrepancies have been found for a few lines, which can normally be explained with the presence of blends. The ions of the Li-like, N-like, and Na-like isoelectronic sequences show a peculiar behavior. Results for the Li-like ions agree with the findings of Landi et al. (2002): these ions provide a too low emission measure, and this discrepancy has been attributed to inaccuracies in the ionization and recombination rates. Results for the Na-like ions are more uncertain, but seem to be in broad agreement with the ions of all the other sequences, contrary to the results of Landi et al. (2002). N-like lines indicate a too high electron temperature; we speculate that this problem may be due to the effects of resonances in the collision rates for transitions within the ground configuration for the N-like ions, largely unaccounted for in CHIANTI, Version The emission measure and electron temperature of the emitting plasma were measured using an emission-measure technique. The temperature thus measured is in good agreement with the independent measurements obtained from line ratios. The agreement of the results from these two independent diagnostic techniques can be an indication of the TABLE 8 EUV Lines Recommended for Coronal Diagnostics Wavelengths (Å) Mg vii , , , Mg viii , , , , Mg ix Al viii Si ix , , Si x , , Si xi , Fe x , , Fe xi , , , Fe xiii

9 No. 1, 2002 COMPARISON BETWEEN CHIANTI AND CDS 503 accuracy of the ion fractions adopted in the present work and an indirect confirmation of the validity of the assumption of ionization equilibrium in the emitting plasma. The emission-measure value determined by the diagnostic technique has allowed the measurement of the line-of-sight extent of the emitting plasma. The elemental composition of the emitting plasma has been checked for the low-fip elements only, confirming the relative values of the abundance of Mg, Al, Si, K, Ca, and Fe. This work provides a list of lines from a number of different elements and ions that can be confidently used for plasma diagnostic purposes. This list is reported in Table 8 and includes most of the lines not belonging to the Li, N, and Na isoelectronic sequences. Lines from each of these three sequences, although self-consistent, should be treated with caution, as the emission-measure analysis has shown some problems. This work has also shown that the emission-measure diagnostic technique has proved to be an excellent method of analysis for isothermal plasmas. U. F. acknowledges support from the NRL/ONR Solar Magnetism and Earth s Environment Accelerated Research Initiative and NASA grants. The work of E. L. and K. P. D. was supported by a grant from NASA s Applied Informations System Research Program (AISRP). The authors wish to thank the referee for his very helpful comments and suggestions. Aggarwal, K. M. 1983, J. Phys. B, 16, L59 Bhatia, A. K., & Doscheck, G. A. 1993, At. Data Nucl. Data Tables, 55, a, At. Data Nucl. Data Tables, 60, b, At. Data Nucl. Data Tables, 60, , At. Data Nucl. Data Tables, 64, 183 Bhatia, A. K., Feldman, U., & Seely, J. F. 1986, At. Data Nucl. Data Tables, 35, 319 Bhatia, A. K., & Landi, E. 2002, At. Data Nucl. Data Tables, in press Bhatia, A. K., & Mason, H. E. 1980, MNRAS, 190, 925 Binello, A. M., Landi, E., Mason, H. E., Storey, P. J., & Brosius, J. W. 2001, A&A, 370, 1071 Binello, A. M., Mason, H. E., & Storey, P. J. 1998a, A&AS, 127, b, A&AS, 131, 153 Brooks, D. H., et al. 1999, A&A, 347, 277 Brueckner, G. E., et al. 1995, Sol. Phys., 162, 357 Christensen, B. R., Norcross, D. W., & Pradhan, A. K. 1986, Phys. Rev. A, 34, 4704 Dankwort, W., & Trefftz, E. 1978, A&A, 65, 93 Dere, K. P., Landi, E., Mason, H. E., Monsignori Fossi, B. C., & Young, P. R. 1997, A&AS, 125, 149 Dere, K. P., Landi, E., Young, P. R., & Del Zanna, G. 2001, ApJS, 134, 331 Fawcett, B. C., & Mason, H. E. 1989, At. Data Nucl. Data Tables, 43, 245 Feldman, U., Doschek, G. A., Schuhle, U., & Wilhelm, K. 1999, ApJ, 518, 500 Feldman, U., & Laming, J. M. 2000, Phys. Scr., 61, 222 Flower, D. R. 1977, A&A, 54, 163 Flower, D. R., & Nussbaumer, H. 1975, A&A, 45, 349 Galavis, M. E., Mendoza, C., & Zeippen, C. J. 1997, A&AS, 123, 159 Gupta, G. P., & Tayal, S. S. 1998, ApJ, 506, , ApJ, 510, 1078 Harrison, R. A., & Thompson, A. M. 1991, RAL Technical Report RAL (Oxfordshire: Rutherford Appleton Lab.) REFERENCES Harrison, R. A., et al. 1995, Sol. Phys., 162, 233 Keenan, F. P., Berrington, K. A., Burke, P. G., Dufton, P. L., & Kingston, A. E. 1986, Phys. Scr., 34, 216 Kohl, J. L., et al. 1995, Sol. Phys., 162, 313 Landi, E., Feldman, U., & Dere, K. P. 2002, ApJS, 139, 281 Landi, E., Landini, M., Dere, K. P., Young, P. R., & Mason, H. E. 1999, A&AS, 135, 339 Malinovsky, M., Dubau, J., & Sahal-Brechot, S. 1980, ApJ, 235, 665 Martin, I., Karwowski, J., Diercksen, G. H. F., & Barrientos, C. 1993, A&AS, 100, 595 Mazzotta, P., Mazzitelli, G., Colafrancesco, S., & Vittorio, N. 1998, A&AS, 133, 403 Pelan, J. C., & Berrington, K. A. 1995, A&AS, 110, 209 Pottasch, S. R. 1963, ApJ, 137, 945 Saha, H. P., & Trefftz, E. 1983, Sol. Phys., 87, 233 Sampson, D. H., Goett, S. J., & Clark, R. E. H. 1984, At. Data Nucl. Data Tables, 30, 125 Sampson, D. H., Zhang, H. L., & Fontes, C. J. 1990, At. Data Nucl. Data Tables, 44, 209 Storey, P. J., & Zeippen, C. J. 2000, MNRAS, 312, 813 Tayal, S. S., Henry, R. J. W., & Pradhan, A. K. 1987, ApJ, 319, 951 Thomas, R. J., & Neupert, W. 1994, ApJS, 91, 461 Wiese, W. L., Smith, M. W., & Glennon, B. M. 1966, Atomic Transition Probabilities, Vol. I: Hydrogen through Neon: A Critical Data Compilation (Washington, DC: US Gov. Printing Off.) Wilhelm, K., et al. 1995, Sol. Phys., 162, 189 Young, P. R., Landi, E., & Thomas, R. J. 1998, A&A, 329, 291 Zhang, H. L., Graziani, M., & Pradhan, A. K. 1994, A&A, 283, 319 Zhang, H. L., & Sampson, D. H. 1992, At. Data Nucl. Data Tables, 52, , At. Data Nucl. Data Tables, 63, 275 Zhang, H. L., Sampson, D. H., & Fontes, C. J. 1990, At. Data Nucl. Data Tables, 44, 31