estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the...

23
Estimation of the bias of the minimum variance technique in the determination of magnetic clouds global quantities and orientation A.M. Gulisano , S. Dasso , C.H. Mandrini , P. D´ emoulin , Instituto de Astronom´ ıa y F´ ısica del Espacio, CONICET-UBA, CC. 67, Suc. 28, 1428 Buenos Aires, Argentina Departamento de F´ ısica, Facultad de Ciencias Exactas y Naturales, UBA, 1428 Buenos Aires, Argentina Observatoire de Paris, Meudon, LESIA, UMR 8109 (CNRS), F-92195 Meudon Cedex, France Abstract Magnetic clouds (MCs) are highly magnetized plasma structures that have a low proton temperature and a magnetic field vector that rotates when seen by a heliospheric observer. More than 25 years of observations of magnetic and plasma properties of MCs at 1 AU have provided significant knowledge of their magnetic structure. However, because in situ observations only give information along the trajectory of the spacecraft, their real 3D mag- netic configuration remains still partially unknown. We generate a set of synthetic clouds, exploring the space of parameters that represents the possible orientations and minimum distances of the satellite trajectory to the cloud axis, . The synthetic clouds have a lo- cal cylindrical symmetry and a linear force free magnetic configuration. From the analysis of synthetic clouds, we quantify the errors introduced in the determination of the orienta- tion/size (and, consequently, of the global magneto-hydrodynamic quantities) by the Mini- mum Variance method when is not zero. Key words: MHD and plasmas, Interplanetary space, Solar wind PACS: 95.30.Qd, 96.50.-e, 94.30.vf Corresponding author, e-mail: [email protected] Member of the Carrera del Investigador Cient´ ıfico, CONICET, Argentina Preprint submitted to Elsevier Science 9 October 2007

Upload: others

Post on 18-Mar-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

Estimation of the bias of the minimum variancetechnique in the determination of magnetic clouds

global quantities and orientation

A.M. Gulisano�����

, S. Dasso��� �����

, C.H. Mandrini�����

, P. Demoulin,

Instituto de Astronomıa y Fısica del Espacio, CONICET-UBA, CC. 67, Suc. 28,

1428 Buenos Aires, Argentina�Departamento de Fısica, Facultad de Ciencias Exactas y Naturales, UBA,

1428 Buenos Aires, Argentina�Observatoire de Paris, Meudon, LESIA, UMR 8109 (CNRS),

F-92195 Meudon Cedex, France

Abstract

Magnetic clouds (MCs) are highly magnetized plasma structures that have a low protontemperature and a magnetic field vector that rotates when seen by a heliospheric observer.More than 25 years of observations of magnetic and plasma properties of MCs at 1 AUhave provided significant knowledge of their magnetic structure. However, because in situobservations only give information along the trajectory of the spacecraft, their real 3D mag-netic configuration remains still partially unknown. We generate a set of synthetic clouds,exploring the space of parameters that represents the possible orientations and minimumdistances of the satellite trajectory to the cloud axis, . The synthetic clouds have a lo-cal cylindrical symmetry and a linear force free magnetic configuration. From the analysisof synthetic clouds, we quantify the errors introduced in the determination of the orienta-tion/size (and, consequently, of the global magneto-hydrodynamic quantities) by the Mini-mum Variance method when is not zero.

Key words: MHD and plasmas, Interplanetary space, Solar windPACS: 95.30.Qd, 96.50.-e, 94.30.vf

�Corresponding author, e-mail: [email protected]�Member of the Carrera del Investigador Cientıfico, CONICET, Argentina

Preprint submitted to Elsevier Science 9 October 2007

Page 2: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

1 Introduction

Magnetic clouds (MCs) are highly magnetized plasma structures that have a lowproton temperature and a magnetic field vector that rotates when seen by a helio-spheric observer.

Even though MCs have been studied for more than 25 years, there is no agreementabout their true magnetic configurations. This is mainly because the magnetic fielddata retrieved in situ by a spacecraft correspond only to the one dimensional cutalong its trajectory and, thus, it is necessary to make some assumptions to infer thecloud 3D structure from observations.

Different distributions of the magnetic field inside the MC have been considered byseveral authors. A cylindrical magnetic configuration with a linear force-free fieldwas proposed by Goldstein (1983), the so called Lundquist’s model (Lundquist,1950). This model has been considered as a relatively good approximation of thefield distribution inside MCs by several authors (e.g., Burlaga and Behannon, 1982;Lepping et al., 1990; Burlaga, 1995; Burlaga et al., 1998; Lynch et al., 2003; Dassoet al., 2005b; Lynch et al., 2005; Dasso et al., 2006). However, many other differentmodels have been also used to describe the magnetic structure of MCs. Some au-thors take a cylindrical shape for the cloud, but consider a non-linear force-free field(Farrugia et al., 1999). Mulligan et al. (1999); Hidalgo et al. (2000, 2002); and Cidet al. (2002) propose different models with non-force free fields. Non-cylindricalstatic models have been also applied to MCs (e.g., Hu and Sunnerup, 2001; Vandasand Romashets, 2003; Hu and Dasgupta, 2005).

From in situ observations and assumptions on the magnetic distribution inside theMC, it is possible to estimate some global magnetohydrodynamic quantities, suchas magnetic fluxes and helicity (e.g., Dasso et al., 2003; Mandrini et al., 2005;Gulisano et al., 2005; Attrill et al., 2006; Dasso et al., 2006). In order to obtain goodestimations of these quantities, it is necessary to find the correct MC orientation,improve the estimation of its size and components of � in the cloud frame.

The Minimum Variance method (MV) has been extensively used to find the orien-tation of structures in the interplanetary medium (see e.g., Sunnerup and Cahill,1967; Burlaga and Behannon, 1982). The Minimum Variance method applied to theobserved temporal series of the magnetic field can estimate quite well the orien-tation of the cloud axis, when the distance between the axis and the spacecrafttrajectory in the MC (the impact parameter, � ) is low with respect to the cloud ra-dius (see e.g., Klein and Burlaga, 1982; Bothmer and Schwenn, 1998; Gulisanoet al., 2005). The MV method has two main advantages with respect to other moresophisticated techniques that are also used to find the orientation of an MC: (1) itis relatively easy to apply, and (2) it makes a minimum number of assumptions onthe magnetic configuration. Unfortunately, when � is significant, the MV approach

2

Page 3: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

is no longer accurate and it provides an orientation which deviates from the realone. However, quantitative estimations of the errors in the cloud orientation givenby the MV technique for finite values of � have not been exhaustively done.

Some authors have used the MV technique to determine the orientation of MCs(e.g., Klein and Burlaga, 1982; Bothmer and Schwenn, 1998; Farrugia et al., 1999;Xiao et al., 2004; Gulisano et al., 2005). Other authors use the MV method to geta first order approximation for the MC orientation. Then, they use this estimationas a seed in a non-linear least squares fit of a magnetic model to the data. Thisapproach is expected to improve the cloud orientation and allows to find the valueof � (e.g., Lepping et al., 1990, 2003, 2006; Dasso et al., 2005b). However, wenotice that the solutions of this fitting can depend on the seed introduced as input.Moreover, these methods need important additional hypothesis to build a magneticmodel (e.g., Shimazu and Marubashi, 2000; Mulligan and Russell, 2001; Hidalgoet al., 2002; Cid et al., 2002; Hu and Sunnerup, 2002; Lynch et al., 2003; Hidalgo,2003). Conversely, the MV method just requires a well ordered large scale varianceof � in the three spatial directions. Several of these different sophisticated tech-niques have been compared analyzing the output of numerical simulations of MCs(Riley et al., 2004). These authors found that these techniques give orientations thatdiffer from the real ones by � ����������

in the latitude angle of the MC axis andby � ����

in the longitude angle, even for low values of ���� (where is the cloudradius). Larger differences in the MC orientation are found for larger values of ���� (see Tables I and II in Riley et al., 2004).

In order to estimate the error introduced by the Minimum Variance method, wegenerate a set of synthetic clouds, considering a cylindrical geometry and a linearforce free field (Lundquist’s model, Lundquist, 1950). The clouds in the set havedifferent orientations and � . We perform a Minimum Variance study in a similarway as the one applied to cloud observations from one spacecraft. The directionassociated to the intermediate eigenvalue gives the estimated direction of the MCaxis (see Section 2). Assuming a null impact parameter, we estimate the MC radius.Then, we fit a Lundquist’s model to the synthetic data to determine the physical pa-rameters (magnetic field strength and twist) keeping the orientation fixed (as givenby the MV method). This last step is done to estimate the propagation of the errorsin the axis orientation and ��� �

in the global quantities (magnetic fluxes and he-licity). We also provide an estimation of ���� from the data rotated to the frame ofthe MV eigenvectors. This permits us to correct the values of the global quantitieswhen ���� is finite. In Section 2, we describe briefly the Minimum Variance methodapplied to magnetic clouds, the physical model, and the global quantities. In Sec-tion 3, we explain how we generate the synthetic clouds. In Section 4, we comparethe results of our MV analysis with the original parameters that characterize thegenerated clouds. Finally, in Section 5, we summarize our findings and conclude.

3

Page 4: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

2 Minimum Variance Technique

2.1 The Method

The MV method finds the direction � in which the projection of a series of�

vectors has a minimum mean quadratic deviation and also provides the directionsof intermediate and maximum variance (e.g., Sonnerup and Scheible, 1998). Thismethod is very useful to determine the orientation of structures that present threeclearly distinguished variance directions. In particular, when this method is appliedto the magnetic field � , the mean quadratic deviation in a generic direction � is:

���� � ����� � � ������� ��� ��������� � (1)

where � � corresponds to each element ( � ) of the magnetic field series. The fieldmean value is:

� ��� � ����� � � � (2)

The MV method finds the direction where � �� is minimum under the constraint� � � � �. The Lagrange multipliers variational method can be used in this determi-

nation. � is the Lagrange multiplier in the following system of equations:

����! #" � �� � � � � � � � � � �%$ � �(3)

where & � �('%)*' �corresponds to the 3 components of � . After applying Eq. (3), the

resulting set of three equations can be written in matrix form as:

� �� �,+ �-� � �/. 0+��1 � �2� �3+ (4)

where:

. 0+ � �54 4 + � ���64 � �64 + � (5)

The indexes i, j represent the field components. The matrix . 0+ is symmetric withreal eigenvalues � � , � � and � - and orthogonal eigenvectors ( 78:9<;

, 7=>9?;, 7@A9?;

),which represent the directions of minimum, maximum and intermediate variationof the magnetic field (the symbol BCB#DEBFB on top of a variable means that it is a unit

4

Page 5: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

vector). The eigenvalues provide the corresponding variance � �� associated witheach direction. Thus, from the eigenvectors ( 78:9?;

, 7=>9?;, 7@A9?;

), it is possible toconstruct the rotation matrix � such that the components of the field in the MVframe of reference can be written as:

� 9?; ��� � � (6)

We will call4������

the field component that corresponds to 78:9<;(minimum va-

riance direction),4����

to that of the maximum variance direction, and4�� ���

tothat having the intermediate variance.

2.2 Cylindrical linear force free model for MCs

The local magnetic structure of MCs at 1 AU has been frequently modeled as acylindrical linear force free field (Lundquist, 1950) by several authors (see e.g.,Burlaga, 1988; Lepping et al., 1990; Lynch et al., 2003; Dasso et al., 2005b; Gulisanoet al., 2005). In this case, the magnetic field is given by:

� � 4�� 7��� 4�� 7� � 4�� 7� � 4������ � ����� � 7��� � � ����� � 7�"! (7)

where� � is the Bessel function of the first kind of order

�,4#�

is the strength of theaxial field at the cloud axis (i.e., at � �

�), and � is a constant, such that the twist

of the magnetic field lines near the cloud axis turns out to be:

$ � � $ � � � � � �)

(8)

The gauge-independent magnetic helicity per unit length (L) for this model is(Dasso et al., 2003):

% �& �('

4 �� �$ � � � �� � ) $ � � � � �� � ) $ � � ��)� � ) $ � � � � � ) $ � � $ � � ' (9)

where is the radius of the MC.

The magnetic flux across a surface transverse to the cloud axis (that is, with anormal along the 7� axis of the cylindrical structure) is:

*�+ � '4�� � � � ) $ � �$ � (10)

5

Page 6: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

The flux per unit length across a surface defined by the radial direction and the axisof the cloud (i. e., a surface with a normal pointing along the 7� direction) is (seee.g., Mandrini et al., 2005):

* �& �

4�� � � � �)� � ) $ � � !) $ � (11)

2.3 Minimum Variance method applied to magnetic clouds

The large and coherent rotation of the magnetic field vector observed by the space-craft when � � �

, allows us to associate: (1) the large scale maximum variancedirection to the azimuthal direction (variation of the observed component of thefield of the order of 2

4 �, see

4��in Eq. (7)), (2) the minimum variation to the

radial direction (the variance will be close to zero, see4 �

in Eq. (7)), and (3) theintermediate variance to the axial direction (variation of the order of

4��, see

4 +in

Eq. (7)) (see e.g., Bothmer and Schwenn, 1998).

However in Lundquist’s model, as well as in the observations of real clouds, themodulus of the magnetic field does not remain constant, being maximum near thecloud axis and minimum toward the cloud boundaries. Moreover, for MCs in ex-pansion and due to magnetic flux conservation in the expanding parcels of fluid,�

��

can decrease significantly while the spacecraft observes the cloud (see e.g.,Berdichevsky et al., 2003; Dasso et al., 2007). This decrease of

���with time is

called the ’aging’ effect since the in situ observations are done at a time, whichis more distant from the launch time as the spacecraft crosses the MC. This de-crease of

���can affect significantly the result of the MV method. However, the

relevant information to find the cloud orientation is in the rotation of the magneticfield. Thus, to decouple the variation of

���from the rotation, we apply the MV

technique to the normalized field vector series:� � � � � � � � � � � � � � � � .

Because MV does not give the positive sense of the variance directions, we choosethis sense for D����� so that it makes an acute angle with the Earth-Sun direction( D����� ). We also choose D ���� so that

4 �"���is positive at the cloud axis, and D�����

is closing the right handed system of coordinates.

3 Generation of synthetic clouds

To generate the set of synthetic MCs we use Lundquist’s model to represent theirmagnetic configuration. The input parameters for this model are:

4#� � )��nT,

� �) ��� �� (i.e., we set the MC boundary in coincidence with � equal to the first

zero of the Bessel function of order zero), and � � � � AU. The cloud velocity is

6

Page 7: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

set to���

km/s. This is only used to construct the magnetic field time series thatan observer located at the Lagrangian point L1 would measure in Geocentric SolarEcliptic (GSE) coordinates (the value of the velocity is not affecting the results).Once the magnetic field model and cloud velocity are fixed, the set of differentclouds is generated taking different values for � and different orientations with res-pect to the GSE system of coordinates. The set of cloud orientations are given by theangles: � (latitude angle, the angle between the cloud axis and the ecliptic plane),and � (longitude angle, the angle between the projection of the cloud axis on theecliptic plane and the Earth-Sun direction ( D��� � ), measured counterclockwise).

The intrinsic cloud reference system and the GSE system of coordinates can berelated using the following rotation matrices:

�������D�������� D��������� D ��������

�������� ����� ���� �

�������D��� �D�����D ����

�������� (12)

where:

��� ��������

��� � �"!$# � � �%�&� � �%� � � � !$# � �� �%�&� � !$# � � �

!'# � �"!$# � � !$# � � �%� � � ��� � �

�������� (13)

and

��� ��������

!$# ��( �%�&�)( �� �%� �*( !$# ��( �� � �

�������� � (14)

Without loosing generality we choose ( (the angle of an arbitrary rotation in theplane � D�+������� ' D�,������ � ) such that D��� � � D��������� � �

, that is:

-/. �*( � � -/. � ��%�&� � (15)

Thus, in this case, the spacecraft trajectory in the MC frame turns out to be:

� � � � ��0 � � � D� ��� � � D������1�' (16)

7

Page 8: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

with :

0 � � � � ��� ������� �� �) � � ������ � � � � � � ' (17)

where 0 � � � is the ’signed’ distance traveled by the spacecraft within the cloud, thisquantity is taken negative when the spacecraft is approaching the cloud axis andpositive when it is leaving the cloud center, and � � � � ��� � � � � is the total transittime (with

� �and

���the time at which the spacecraft enters and leaves the structure,

respectively).

The projection of the trajectory on the plane � D�������� ' D��� ���� � , which is the polarcoordinate � � � � , is:

� � � � �� � � � � � � ������ � � � � � � � � ����1�' � � ! (18)

where:

� � � � � � ���� � 0 � � � � !$# ��( �%�&� � !$# � � � �%�&�*( �%�&� � �� � � � ������� � � � constant � (19)

then, the angle ( �� ���� ) between the projection of the trajectory on the plane � D�+������� ' D������1�' �and the D�+� ����

satisfies:

�%�&� �� ���� � � � � � ������� � � � � � � � �� 0 � � � � � !$# �,( ��� � �"!'# � � � �%� �)( �%� � � � � � � � (20)

!$# � �� ���� � � � � � � ���� � � � � � � � 0 � � � � !$# ��( �%� � �"!'# � � � �%�&�)( �%�&� � �� 0 � � � � � !'# ��( ��� � �"!$# � � � ��� �)( ��� � � � � � � � (21)

The magnetic field time series inside the synthetic cloud is:

� � � � � 4�� �)� � � � � � � � D �����1�' �� 4�� � � � � � � � � � D�� ���� � � � ' (22)

where D�� ���� � � � is related to �%�&� �� ���� and !$# � �� ���� as:

D�� ���� � � � � � ��� � �� ���� � � � D�+����1�' �� !'# � �� ���� � � � D��� ���� � (23)

8

Page 9: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

Note that to compute � � � � we need � � � � and consequently 0 � � � and � � (see Eqs. (17)and (18)). To compute � � , we set

� � � �(or equivalently

� � � �) in Eqs. (17) and

(18), so that � � , then we find:

� � � ) � � � � �� ������ � !$# ��( �%�&� �"!$# � � � �%�&�)( �%�&� � � � (24)

To obtain the time series of � � � � from Eq. (22) we take ���� from 0 to 0.9 and thefollowing sets of orientations: � � ' � � � � ��� '��� � ��� ��'��� � � �,'��� � � ��,' � � �(� � ���,' �E)�� �(the axis of the MC from � � � to almost parallel to D ��� and almost perpendicularto the Sun-Earth line), � � ' � � � � � '%)��� � � ���' ���� � (the cloud axis almost on theecliptic plane and in the Sun-Earth direction), � � ' � � � � �,'���� � � ���,'%) � � (the cloudaxis almost on the ecliptic and perpendicular to the Sun-Earth line), and � � ' � � �� ����' ���� � .

4 Analysis of synthetic clouds

4.1 General Results

From the time series of � � � � (Eq. (22)), a given orientation and � , we generatethe corresponding time series (in the GSE components), which emulates the obser-vations of the spacecraft (i.e., the ’observations’ of the synthetic cloud). We thenapply the MV technique to the normalized field vectors (

� � � � � � � � � � � � � � � � ) ofthese synthetic time series.

We estimate the MC orientation: the latitude angle ( � ��� ) and the longitude angle( � ��� ), which for ���� �

will be different from the correct values of � and � chosenat the end of Section 3. From Eq. (24), using

� ������ and � � , we estimate the radius

( ��� ) assuming that � is null (since it is unknown).

Siscoe and Suey (1972) proposed a criterion to determine the anisotropy of a givennormalized vector series � , computing the eigenvalues ( � ) and the eigenvectorsof the matrix � + ����� � +�� . For a set of 100 vectors, as used in our case, theyfound a significance criterion to determine if the set presents significantly distinctspatial directions. This criterion is ��� � �(� � ��� � � ) and ������� �(� � � � � � � (a fullyrandom series of vectors satisfies it in 95% of the cases). For the vector series cor-responding to the MC field (

� ), the minimum variance method finds eigenvalues

and eigenvectors of the matrix . 0+ ��� � � +!� � �"� �� �"� +#� (see section 2.1).

Upper (lower) panel of Figure 1 left shows the ratio between the minimum and

9

Page 10: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

intermediate (maximum and intermediate) eigenvalues obtained from MV (using

. 0+ ) applied to the set of the generated synthetic MCs, as a function of ���� . Notethat, as expected, these eigenvalues ratios do not depend on the orientation of thecloud (because the eigenvalues are invariant under rotations). � � � stays lower than� � ������� � � � , while ������� stays above � � � � .Figure 1 right shows eigenvalues ratios using the matrix � 0+ (as done by Siscoeand Suey), where we take � � � � �

. Notice that the criterion of Siscoe and Suey isvery well satisfied for large values of ��� , and the eigenvalue separation becomeseven larger for increasing values of ���� . However, this only assures well distinctvariance directions, but it does not measure how good is the approximation of theMC axis by the intermediate eigenvalue direction. The criterion of Siscoe and Sueyis only a necessary condition, far from being sufficient for a good determination ofthe MC orientation.

4.2 Comparison with the expected cloud orientations

In Figure 2 we report the angle � between the generated cloud axis ( D ������ ) and the

one obtained by MV ( D ���� ), as a function of ���� , we plot only one curve since �does not depend on the MC orientation. It can be seen that the deviation from thereal generated orientation increases with ��� . For ��� � � � ) , we obtain � � ) �and when ���� reaches the extreme value � � � � , we get � � ��� �

. We remark thatfor real observations, cases with ��� � � � � will be more affected by the interactionbetween the MC and the solar wind (see e.g., Dasso et al., 2005a) with a conse-quent observation of � � � � significantly different from the ideal modeled magneticconfiguration; this will introduce biases which are not present in our synthetic MCset.

The deviation angle � is well represented by a quadratic curve (Figure 2) since thePCC (Pearson correlation coefficient) is 0.999. It is noteworthy that this deviation inthe MC orientation is not evident from the eigenvalues ratios (Figure 1) discussed inthe previous section. Despite the MV technique finds well distinguished (minimum,maximum, and intermediate) directions, when ���� is not small those directions donot correspond to D�+������ ' D�,������ ' D ��������

.

We compared the above results to those of an analysis using the time series � � � � , sowithout normalization, and we found worse orientations ( ��� )�� �

for ��� � � � �and � � � � � for ��� � � � � ). Thus, we conclude that to get better estimationsfor the cloud axis orientation it is convenient to apply the MV technique to thenormalized time series

� � � � .In Figure 3 (Figure 4) we report the deviation in the angles � ( � ), as � � � � � � ���( � � � � � � ��� ). We report these deviations, in addition to the deviation � , since

10

Page 11: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

� and � are frequently used in the literature to give the orientation of MCs.

There is a general trend to obtain larger deviations in � than in � , especially when� � � is large (this is an effect of the spherical coordinates used close to where � issingular, at the two poles). As expected,

� � � � and� � � � increase as � increases,

but with a different behavior for different cloud orientations, which is quantified inFigures 3 and 4. For all the synthetic MCs � ��� � � ; thus, MV tends to give lowervalues of the latitude angle as � increases.

When the cloud axis is close to the ecliptic plane and it is perpendicular to theSun-Earth direction, � is very well determined even for very large values of � (seesymbols � and � in Figure 3). However, clouds with their axis either nearly per-pendicular to the ecliptic (see symbols: � � � � ��� � , � � � � �� � , � � � �

� � , and� � � � �� � in Figure 3) or nearly parallel to Sun-Earth direction (see symbols: �

� � � ����� � , � � � � )���� � in Figure 3) give the largest errors, e.g., � � between � ) �and

��� �for ���� � � � .

For the cloud axis close to the Sun-Earth direction and low latitudes, � is welldetermined even for very large values of � (see symbol: � � � � ����� � in Figure4). For increasing values of the latitude, errors in � turn to be larger (see symbols:� � � � � � � , � � � � � � , � � � � �� � , and � � � � ��� � in Figure 4). We remark that thelargest error in the determination of � , when � is

� � �(cloud axis almost pointing to

the ecliptic north), is not meaningful because it comes from the small differencesin the determination of D �����1�'

(small � ) and, consequently, large deviations in itsprojection on the ecliptic plane (i.e., on � ). For latitudes lower than � � � � � weobtain a � � � ) ���

, even for ��� as large as 0.9.

4.3 Comparison with the expected cloud parameters and global magnitudes

To compare with the generated MCs, we fit the physical parameters of Lundquist’smodel to the � 9<;

components, using a non linear least-squares fitting, as des-cribed in Dasso et al. (2006). Then, from the fitted parameters, we calculate

%� �&

and the fluxes as discussed below.

We compute the mean value of4�����

, which is expected to be zero in the case� � �

, we depict� � 4������ � � � 4�� ��� versus ���� in the upper panel of Figure 5,

together with the quadratic regression curve that best fits the points. The deviation islarger when � increases. This deviation can provide a quantitative estimation of theimpact parameter by reporting the measured value of� ��� � � � 4 � ��� � � � 4�� ��� in the upper panel of Figure 5, or simply by using the

approximation � � � � ��� � � � � from the quadratic fit.

The magnetic field strength4 � ���

, found by fitting Lundquist’s model, is also af-fected by the error introduced by MV in the axis orientation. The middle panel of

11

Page 12: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

Figure 5 shows the difference � 4� � 4�� �<4�� ���. We also find a quadratic depen-

dence with ��� . The difference in the field strength can be significant, for example� � ��� for ���� � � � � .We also analyze the magnitude of the coherent rotation of the magnetic field asdefined by the angle � between the field vector components that lie on the maximumvariance plane (

4� ��� '�4�����), taken at the front and end boundaries of the MC. If

the right cloud orientation is used, � � ���� �for all values of � because all the

synthetic clouds satisfy4 + � �

at the MC boundary. However, with the orientationfound with the MV method,

���decreases as ���� increases due to a mix of field

components (lower panel of Figure 5). Thus, we find that the angle���varies from�����

(for � � �) to � � )���

in the extreme case when � � � � � (i.e. when thespacecraft crosses the cloud close to its periphery). This is still a large coherentrotation of the magnetic field.

Finally, in all panels of Figure 5 we plot only a unique curve to represent all thesynthetic clouds for a given ���� , since the values obtained do not depend on theMC orientation.

Figure 6 shows the radius obtained from MV (assuming � � �) in units of the

model radius ( � � � � AU) as a function of ���� ; thus, it quantifies the underes-timation of the real radius using ��� when � �� �

. The solid line in this figurecorresponds to

�� � � � ���� � � � , the value that would be obtained for the correct o-

rientations defined for the synthetic clouds, but with the (only) bias given when theunderestimation of the size is due to the finite value of � . This introduces an errorin the cloud radius lower than 30% for ���� � � � . As expected, the only caseshaving significant variations in the estimated radius because of errors in the orien-tations are those where the cloud axis is nearly parallel to the Sun-Earth direction:symbols � � � � ��� � ' � � ���� � � , � � � � ��� � ' � � ���� � � , and � � � � � � ' ��� )��� � � .Figure 7 shows the fitted values for � ( �

���). As expected, the value � � ) � AU

� �

(used for all the synthetic clouds) is recovered when ���� � �. The solid line

represents � � ) � AU� � � � � � � ���� � � , which corresponds to the value of � that

cancels�)� ��� � for the underestimated radius

� � � � � . Except for the same threeMCs that give the largest error in (axis oriented near the Sun-Earth direction), theobtained values of � differ from � � ) � AU

� �basically due to the underestimation

of the radius. The right panel of Figure 7 shows that, excluding the three casesmentioned before, there is a small trend to cancel this effect, because the fittedvalues of � are between � � ) � AU

� �and � � ) � AU

� � � � � � � � ���� � � � .For every synthetic cloud we compute the magnetic helicity (Eq. 9) from the modeland the fitted parameters to the Lundquist’s model, using the MV orientation (

% ���).

In Figure 8 we plot � % � % � ������

�as a function of ���� . The solid line cor-

responds to the helicity computed with the correct (values used to the produce thesynthetic clouds)

4�� � � )�� ��� � and � (=24AU� �

), but considering the underesti-

12

Page 13: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

mated radius as� � � � � . % ���

underestimates%

for two reasons: (1) the under-estimation of the radius and (2) the bias introduced in the fitted values of

4#� ���and

����

. The radius underestimation can be partially corrected by estimating the im-pact parameter from the observed

� � 4� ��� � � � 4�� ��� (upper panel of Figure 5),but an underestimation remains due to the error in the MC orientation (Figure 7).Indeed, from the middle panel of Figure 5 (underestimation of

4��for finite values

of � ) and Figure 7 (overestimation of � ), and the main dependence on these twoparameters (Eq. 9), it is expected that

%����will be further underestimated. If the

cloud size is corrected, the largest error is �����

for ���� � � � � . Thus, the largestsource of error (always an underestimation) in the magnetic helicity value of idealMCs is the influence of � on the estimation of the radius.

We also compute � *�+ � * + � � � � ����

�� � (Figure 9) and � * � � * � � �

� �����

��

� (Fi-gure 10). Both fluxes have the same sources of underestimation as the magnetichelicity above. From the above results, we conclude that the main reason for under-estimations of global quantities is the decrease of ��� , which can be corrected if �is computed. We also note that for the three global quantities, the largest deviationsappear in clouds oriented with their axes close to the Earth-Sun direction.

5 Summary and Conclusions

We study the bias introduced by the minimum variance (MV) technique in the de-termination of the orientation and size of magnetic clouds (MCs). We also analyzethe influence of this bias and fitted parameters on the estimation of their globalquantities (magnetic fluxes and helicity) for non-negligible values of the impactparameter � compared to the MC radius .

We explore this bias studying a set of synthetic cylindrical and linear force-freefield MCs. The physical parameters for this model are chosen to be the same for allclouds and the only differences in the set correspond to different orientations andimpact parameters (���� ).

MCs can present signatures of strong expansion (an effect not considered in oursynthetic set of MCs) (e.g., Shimazu and Marubashi, 2000; Berdichevsky et al.,2003; Nakwacki et al., 2005, 2006). However, the main source of error when usingMV for expanding MCs is introduced by the strong decrease of

���. This effect can

be corrected if MV is applied to the normalized time series:� � � � � � � � � � � � � � � � .

Moreover, Dasso et al. (2007) compared the orientation of an MC using the norma-lized MV technique with another technique that considers flux cancellation (Dassoet al., 2006). Both methods were in a good agreement. However, other effects canintroduce biases on the MC orientation and size. Examples are the deviation froma cylindrical symmetry, ambiguous MC boundaries, or the presence of a significantlevel of field fluctuations. Such effects are not studied in the present work. This will

13

Page 14: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

be the purpose of a future research.

We quantify, in function of � , the dependence of the increasing deviation ( � ) of theMC symmetry axis ( D ��������

), provided by MV applied to the normalized series ofvectors, from the generated one. We find a deviation of � � � �

for � � ����of the

cloud radius; even more, � remains being lower than)�� �

for � as high as����

ofthe MC radius. When MV is applied to � � � � we obtain larger deviations; thus, weconclude that application of MV to

� � � � gives better results.

The criterion discussed in Siscoe and Suey (1972), based on the quantification ofthe eigenvalue ratios, is a good estimator to determine if a given vector set cor-responds to well distinguished spatial directions. This criterion has been used inthe literature to validate the goodness of the orientations given by MV. However,we have found that even when this criterion is very well satisfied, the MV direc-tions can still deviate significantly from the main cloud axes. Thus, this criterionis not a good estimator of the quality of the cloud orientation obtained with theMV method. We conclude that this criterion should be taken with care, since tohave well distinguished variance directions is a necessary but far from sufficientcondition to assure that the orientation found is the real MC orientation.

A good determination of the latitude ( � ) and longitude ( � ) of the main axis of theMC is important to obtain a good estimation of the cloud size from its speed andrange of observed times. We have found that � is better determined for an MCwith its axis nearly on the ecliptic plane (i.e., � � � �

) and nearly perpendicular tothe Sun-Earth direction ( � � ���

or � � �����), even for values of � as large as

� � � � � (see Figure 3). For all the studied cases we have found typically � � lowerthan � � even for ���� as large as

� � � . The worse determination of � corresponds tothe MC axis nearly perpendicular to the ecliptic plane ( � � ��� � ), being this largeuncertainty in � a geometric amplification due to the description of the poles inspherical coordinates (the full range of � values are clustered around the poles).

One of the main unknown parameters of an MC is its radius, which can be sig-nificantly underestimated from observations when � is a significant fraction of theMC radius. This implies an underestimation of extensive global magnetohydrody-namical quantities in MCs, as the magnetic fluxes and helicity. The deviation fromzero of the mean value of the D����� magnetic field component (in the direction oflowest variance) can be used to obtain a first order estimation of � , directly fromobservations, as � � � � ��� � � � � for Lundquist’s MCs. This estimation of � letsus improve the estimation of the MC radius and, consequently, the estimations ofglobal quantities since the uncertainty in the cloud size is one of the main sourcesof error.

14

Page 15: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

Acknowledgments

This work was supported by the Argentinean grants: UBACyT X329 (UBA), PIP 6220(CONICET) and PICTs 03-12187, 03-14163, and 03-33370 (ANPCyT).A.M.G. is a fellow of ANPCyT. C.H.M. and P.D. thank CONICET (Argentina) and CNRS(France) for their cooperative science program (N

�20326).

References

Attrill, G., Nakwacki, M. S., Harra, L. K., van Driel-Gesztelyi, L., Mandrini, C. H., Dasso,S., Wang, J., 2006. Using the Evolution of Coronal Dimming Regions to Probe theGlobal Magnetic Field Topology. Solar Phys.,238, 117–139.

Berdichevsky, D. B., Lepping, R. P., Farrugia, C. J., 2003. Geometric considerations of theevolution of magnetic flux ropes. Phys. Rev. E 67 (3), 036405.

Bothmer, V., Schwenn, R., 1998. The structure and origin of magnetic clouds in the solarwind. Annales Geophysicae 16, 1–24.

Burlaga, L., Fitzenreiter, R., Lepping, R., Ogilvie, K., Szabo, A., Lazarus, A., Steinberg,J., Gloeckler, G., Howard, R., Michels, D., Farrugia, C., Lin, R. P., Larson, D. E., 1998.A magnetic cloud containing prominence material - January 1997. Journal of Geophys.Res.,103 (A1), 277–286.

Burlaga, L. F., 1988. Magnetic clouds and force-free fields with constant alpha. Journal ofGeophys. Res.,93 (12), 7217–7224.

Burlaga, L. F., 1995. Interplanetary magnetohydrodynamics. New York : Oxford UniversityPress, 1995.

Burlaga, L. F., Behannon, K. W., 1982. Magnetic clouds: Voyager Observations Between 2and 4 AU. Solar Physics 81, 181–192.

Cid, C., Hidalgo, M. A., Nieves-Chinchilla, T., Sequeiros, J., Vinas, A. F., 2002. Plasma andMagnetic Field Inside Magnetic Clouds: a Global Study. Solar Physics 207, 187–198.

Dasso, S., Gulisano, A. M., Mandrini, C. H., Demoulin, P., 2005a. Model-independent largescale magnetohydrodynamic quantities in magnetic clouds. Advances in Space Research35, 2172–2177.

Dasso, S., Mandrini, C. H., Demoulin, P., Farrugia, C. J., 2003. Magnetic helicity analysisof an interplanetary twisted flux tube. Journal of Geophys. Res.,108 (A10), 1362–1364.

Dasso, S., Mandrini, C. H., Demoulin, P., Luoni, 2006. A new model-independent methodto comute magnetic helicity in mangetic clouds. Astronomy and Astrophysics 455, 349–359.

Dasso, S., Mandrini, C. H., Demoulin, P., Luoni, M. L., Gulisano, A. M., 2005b. Largescale MHD properties of interplanetary magnetic clouds. Advances in Space Research35, 711–724.

Dasso, S., Nakwacki, M. S., Demoulin, P., Mandrini, C. H., 2007. Progressive transforma-tion of a flux rope to an ICME. Solar Physics, in press.

Farrugia, C. J., Janoo, L. A., Torbert, R. B., Quinn, J. M., Ogilvie, K. W., Lepping, R. P.,Fitzenreiter, R. J., Steinberg, J. T., Lazarus, A. J., Lin, R. P., Larson, D., Dasso, S.,Gratton, F. T., Lin, Y., Berdichevsky, D., 1999. A Uniform-Twist Magnetic Flux Ropein the Solar Wind. In: AIP Conf. Proc. 471: Solar Wind Nine. pp. 745–748.

15

Page 16: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

Goldstein, H., 1983. On the field configuration in magnetic clouds. In: Solar Wind Confer-ence. pp. 731.–733.

Gulisano, A. M., Dasso, S., Mandrini, C. H., Demoulin, P., 2005. Magnetic clouds: A sta-tistical study of magnetic helicity. Journal of Atmospheric and Terrestrial Physics, 67,1761–1766.

Hidalgo, M. A., 2003. A study of the expansion and distortion of the cross section of mag-netic clouds in the interplanetary medium. Journal of Geophys. Res.,108 (A8), 1320.

Hidalgo, M. A., Cid, C., Medina, J., Vinas, A. F., 2000. A new model for the topology ofmagnetic clouds in the solar wind. Solar Physics 194, 165–174.

Hidalgo, M. A., Cid, C., Medina, J., Vinas, A. F., Sequeiros, J., 2002. A non-force free ap-proach to the topology of magnetic clouds in the solar wind. J. Geophys. Res. 107 (A1),1002.

Hu, Q., Dasgupta, B., 2005. Calculation of magnetic helicity of cylindrical flux rope. Geo-phys. Res. Let. 32, L12109.

Hu, Q., Sunnerup, B. U. O., 2001. Reconstruction of magnetic flux ropes in the solar wind.Geophys. Res. Let. 28 (3), 467–470.

Hu, Q., Sunnerup, B. U. O., 2002. Reconstruction of magnetic flux ropes in the solar wind:Orientations and configurations. Journal of Geophys. Res.,107 (A7), 1142–1156.

Klein, L. W., Burlaga, L. F., 1982. Interplanetary magnetic clouds at 1 AU. Journal ofGeophys. Res.,87, 613–624.

Lepping, R. P., Berdichevsky, D. B., Ferguson, T. J., 2003. Estimated errors in magneticcloud model fit parameters with force-free cylindrically symmetric assumptions. Journalof Geophys. Res.,108 (A10), 1356–1375.

Lepping, R. P., Berdichevsky, D. B., Wu, C.-C., Szabo, A., Narock, T., Mariani, F., Lazarus,A. J., Quivers, A. J., 2006. A summary of WIND magnetic clouds for years 1995-2003:model-fitted parameters, associated errors and classifications. Annales Geophysicae 24,215–245.

Lepping, R. P., Burlaga, L. F., Jones, J. A., 1990. Magnetic field structure of interplanetarymagnetic clouds at 1 AU. Journal of Geophys. Res.,95 (14), 11957–11965.

Lundquist, S., 1950. Magnetohydrostatic fields. Ark. Fys. 2, 361–365.Lynch, B. J., Gruesbeck, J. R., Zurbuchen, T. H., Antiochos, S. K., 2005. Solar cycle-

dependent helicity transport by magnetic clouds. Journal of Geophys. Res.,110, A08107.Lynch, B. J., Zurbuchen, T. H., Fisk, L. A., Antiochos, S. K., 2003. Internal structure of

magnetic clouds: Plasma and composition. Journal of Geophys. Res.,108 (A6), 1239.Mandrini, C. H., Pohjolainen, S., Dasso, S., Green, L. M., Demoulin, P., Van Driel-

Gesztelyi, L., Copperwheat, C., Foley, C., 2005. Interplanetary flux rope ejected froman Z –ray bright point. The smallest magnetic cloud source–region ever observed. As-tronomy & Astrophysics 434, 725–740.

Mulligan, T., Russell, C. T., 2001. Multispacecraft modeling of the flux rope structureof interplanetary coronal mass ejections: Cylindrically symmetric versus nonsymmet-ric topologies. Journal of Geophys. Res.,106, 10581–10596.

Mulligan, T., Russell, C. T., Anderson, B. J., Lohr, D. A., Toth, B. A., Zanetti, L. J., Acuna,M. H., Lepping, R. P., Gosling, J. T., Luhmann, J. G., 1999. Flux Rope Modeling of anInterplanetary Coronal Mass Ejection Observed at WIND and NEAR. In: Solar WindNine, AIP Conference Proceedings. Vol. 471. p. 689.

Nakwacki, M. S., Dasso, S., Mandrini, C. H., Demoulin, P., 2005. Helicity analysis forexpanding magnetic clouds: A case study. Proc. Solar Wind 11 - SOHO 16, ESA SP-

16

Page 17: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

592, 629–632.Nakwacki, M. S., Dasso, S., Mandrini, C. H., Demoulin, P., 2006. Global Magnitudes in

Expanding Magnetic Clouds. In: Revista Mexicana de Astronomia y Astrofisica Confer-ence Series. p. 155.

Riley, P., Linker, J. A., Lionello, R., Mikic, Z., Odstrcil, D., Hidalgo, M. A., Cid, C., Hu,Q., Lepping, R. P., Lynch, B. J., Rees, A., 2004. Fitting flux ropes to a global MHDsolution: a comparison of techniques. Journal of Atmospheric and Terrestrial Physics66, 1321–1331.

Shimazu, H., Marubashi, K., 2000. New method for detecting interplanetary flux ropes.Journal of Geophys. Res.,105, 2365–2374.

Siscoe, G. L., Suey, R. W., 1972. Significance Criteria for Variance Matrix Applications.Journal of Geophys. Res.,77 (7), 1321–1322.

Sonnerup, B. U. O., Scheible, M., 1998. ISSI Science Report, Sr-001, Analysis methodsfor multispacecraft data. Kluwer Academic.

Sunnerup, B. U. O., Cahill, L. J., 1967. Magnetopause Structure and Attitude from Explorer12 Observations. Journal of Geophys. Res.,72 (1), 171–183.

Vandas, M., Romashets, E. P., 2003. A force-free field with constant alpha in an oblatecylinder: A generalization of the Lundquist solution. Astron. Astrophys.,398, 801–807.

Xiao, C. J., Pu, Z. Y., Ma, Z. W., Fu, S. Y., Huang, Z. Y., Zong, Q. G., 2004. Inferring offlux rope orientation with the minimum variance analysis technique. Journal of Geophys.Res.,109 (A18), 11218–11226.

0 0.2 0.4 0.6 0.8 1−2

0

2

4

6x 10

−4

p/R

λ min

/ λ in

t

0 0.2 0.4 0.6 0.8 10

5

10

15

p/R

λ max

/ λ in

t

0 0.2 0.4 0.6 0.8 10

0.05

0.1

p/R

λ min

/ λ in

t

0 0.2 0.4 0.6 0.8 10

5

10

15

p/R

λ max

/ λ in

t

Fig. 1. Left: Upper (lower) panel shows the ratio between the minimum and intermediate(maximum and intermediate) eigenvalues from MV (using �

0+) for the set of generated

MCs as a function of in units of the MC radius.Right: Upper (lower) panel shows the ratios but now using the matrix

� +, where we take

� ����� � for the set of generated MCs.

17

Page 18: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

0 0.2 0.4 0.6 0.8 10

2

4

6

8

10

12

14

16

18

20

a(p/R)2+b(p/R)+c: a =17 , b =3.2 , c =0.39

PCC =0.999 η

p/R

Fig. 2. Angle � between the generated cloud axis (�� �������

) and the one obtained by MV(squares) as a function of in units of the MC radius together with the quadratic regressioncurve (solid line).

18

Page 19: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

0 0.2 0.4 0.6 0.8 1

0

2

4

6

8

10

12

14

16

18

p/R

� �

Fig. 3. Bias from Minimum Variance in � ( ��� � ����� ��� ) in function of inunits of the radius. Different orientations of the synthetic clouds are represented by� ����� � � �� �� ���������� ��� ������������� ��� � � � ����! � ����������#" �%$ �����&����(' � �)������*&��� , + ��� �����&��� , ,� *�� ��� � ��� , - � �)� ��� $ �����. ��� � ���)� � � .

0 0.2 0.4 0.6 0.8 1−60

−50

−40

−30

−20

−10

0

10

20

p/R

� �

Fig. 4. Idem Figure 3 but showing the difference in ( �/ � 0�� ��� ).

19

Page 20: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.5

1

1.5

2

a(p/R)2+b(p/R)+c: a =1.6 , b =0.077 , c =0.053

PCC =0.994

p/R

| <B

XM

V

>|/B

0MV

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

5

10

15

20

a(p/R)2+b(p/R)+c: a =13 , b =4.1 , c =−0.25

PCC =0.999 |∆ B

0|

p/R

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1100

120

140

160

180

a(p/R)2+b(p/R)+c: a =−1.3e+02 , b =61 , c =1.8e+02

PCC =0.977

p/R

|γM

V|

Fig. 5. Upper panel: ����� � ����� ���� � ��� . Middle panel: �� � � � � ��� � ��� . Lowerpanel: absolute value of the rotation angle of � in the plane of maximum variance ( � ��� � ).All the quantities are plotted, in function of in units of the radius. Their quadratic regres-sion curves are added.

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

RM

V/R

p/R

Fig. 6. � ��� in units of the synthetic radius R (� ��� � AU) as a function of ��� . Solid

line corresponds to the small expected radius (because of the assumption � � ) computed

from the orientations of the synthetic MCs. Symbols are given as in Figure 3.

20

Page 21: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

0 0.5 10

50

100

150

200

250

300

α MV (

AU

−1 )

p/R0 0.5 1

24

26

28

30

32

34

36

38

40

42

p/R

Fig. 7. ����

as a function of ��� . Dashed line corresponds to the value set to generatethe synthetic clouds ( �

� � AU� �

). The symbols for different orientations are given as inFigure 3. Solid line shows � =24 AU

� � � � � � � � ��� � � � . The left panel shows the full set

of clouds, while the right panel shows a zoom.

21

Page 22: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

0 0.2 0.4 0.6 0.8 1

0

0.2

0.4

0.6

0.8

1

∆H/H

p/R

Fig. 8. Difference between the theoretical and estimated values (using MV) of the magnetichelicity, divided by the theoretical value, ��� ��� � �

� ������

� ��� , as a function of ��� . The symbols for different orientations are given as in Figure 3. Solid line corres-ponds to �

���computed with the values for � � and � used to generate the synthetic MCs,

� � � � ������� � and � (=24 AU� �

), but considering the radius� � � � � � � .

.

0 0.2 0.4 0.6 0.8 1

0

0.2

0.4

0.6

0.8

1

∆Fz/F

z

p/R

Fig. 9. Idem as Figure 8 but for �+

values.

22

Page 23: Estimation of the bias of the minimum variance technique ... · riance direction), 4to that of the maximum variance direction, and to that having the intermediate variance. 2.2 Cylindrical

0 0.2 0.4 0.6 0.8 1

0

0.2

0.4

0.6

0.8

1

∆Fφ/F

φ

p/R

Fig. 10. Idem as Figure 8 but for �� ��� values.

23