Italian Magnetic Network and magnetic reference fields at 2010 . 0

The Istituto Nazionale di Geofisica e Vulcanologia (INGV) has systematically undertaken the task of making measurements of the Earth's magnetic field in Italy. By tradition and because of the elongated geometric shape of our peninsula and islands, in Italy, a grid of more than 110 points, called the first order repeat stations, with an average spacing around 55-60 km, is in operation. Over this grid the measurements are repeated regularly, every 5 years. A survey of 131 repeat stations of the Italian Magnetic Network (including 2 observatories, 11 stations in Albania, 3 stations in Corsica and 1 in Malta) was carried out between 2009 and 2010 with the main purpose of updating our magnetic cartography. We describe the characteristics of magnetic first and second order networks, the magnetic measurements and the data reduction procedure. In agreement with the recommendations of MagNetE Committee, we report new repeat station data measured and reduced at 2010.0. An analytical expression, a second order polynomial, in latitude and longitude for the field elements, was determined, and coefficients for 2010.0 and average secular variation over the period 2005-2010, were obtained. The new maps for Italy, for D, F, H and Z at the epoch 2010.0, are shown. A selection of stations from the Italian Magnetic Network, based on their low values of anomaly with respect to a 'normal' field, is also proposed for future surveys.


Introduction
Geomagnetism is a key property of our planet.A background in geomagnetism can be found, for example, in Merrill et al. [1996], Campbell [2003], Lanza and Meloni [2006].The existence of a geomagnetic field allows animals, and human beings using a compass, to use it for orientation information.In this regard, the observation of the angular difference with respect to the geographic meridian North direction, the so called declination, has provided, for centuries, direction for navigation purposes.
In Italy, a long history of Earth's magnetic measurements and practice can be dated back in time at least to the 17th century but the stable engagement in regularly repeated magnetic measurements, started only after the national unification at the end of the 19 th century.In the 20th century, since the 1930's, a modern Italian Magnetic Network composed of repeat stations, regularly distributed over the Italian territory, and integrated by magnetic observatories, has allowed the systematic determination of the spatial structure and the time variation characteristics of the Earth's magnetic field over the country [Cafarella et al. 1992a[Cafarella et al. , 1992b]].
Table 1 shows a brief history of all magnetic surveys made in the Italian territory starting from 1640 to present, indicating the observed magnetic elements, together with the number of repeat stations (# Sites) and the corresponding observers (or Institutions involved).The magnetic survey of 1891-92 represents the first 3-component magnetic survey in the then unified kingdom of Italy.The latest modern magnetic surveys are reported in Molina et al. [1985], Meloni et al. [1988Meloni et al. [ , 1994]], Coticchia et al. [2001] and Dominici et al. [2007].The latest modern magnetic surveys made at the European level are reported in Korte and Fredow [2001] and Mandea [2004], for example.

The Italian Magnetic Network
Because of its institutional role, and with consideration to the peculiar geometric shape of our peninsula and islands, the Istituto Nazionale di Geofisica e Vulcanologia (INGV) undertakes a survey of measurements on the "first order network" every five years.In the latest survey of 2010, measurements were made at 131 repeat stations including 2 observatories, 11 stations in Albania, 3 stations in Corsica and 1 in Malta, as shown in Figure 1.An agreement with Albania allows to make common measurements on a network there established in previous surveys [Chiappini et al. 1997].The principal aims of this latest survey are the new magnetic maps compilation, the observation of secular variation of geomagnetic field elements and the computation of regional models to represent the main geomagnetic field elements pattern over Italy.
Magnetic repeat stations must satisfy the following requirements: i) absence of significant artificial and natural disturbance (for 'natural' we mean to avoid volcanic or other evident natural local anomaly source); ii) be representative of a quite large area (since we have avoided possible disturbances); iii) availability of bearings for azimuths with respect to the local

Subject classification:
Earth's magnetic field, Geomagnetic map, Magnetic surveys, Magnetic repeat stations, Secular variation, Global and regional models.geographic meridian.A magnetic repeat station consists of a mark on a concrete pillar, or on platform of a 1 m 2 in the area of the location selected for the magnetic measurement.

MagNetE workshop 2011
In Jankowsky and Suckdorff [1996] a full general description of magnetic repeat station characteristics is reported.
For each repeat station, a monograph (an example is shown in Figure 2) is prepared for the purpose of keeping note of magnetic measurements there taken, and to allow a quick determination of the correct location.
To meet all necessary requirements, the monograph consists of two parts: first, we have indications about the local area (e.g., the land owner), a brief description about 2010.0ITALIAN MAGNETIC NETWORK how to reach the location with the help of a detailed map, coordinates and altitude of the site, together with the date of the first series of measurements.In addition, the Italian institution that first occupied the repeat station is indicated by one of the following acronyms: For a detailed description of the magnetic characteristics over a certain geographical area, a suitable spatial scale is desirable.However, in order to achieve this goal, the resolution of the first order network is not adequate.For this reason, a second order magnetic data network, with a higher spatial density of measurements, was deployed.In Italy, the second order network consists of 1529 points of declination (D) measurements, determined in the 1930s by the Military Geographical Institute (IGMI); information on this network can be found in Talamo [1975] in which the last survey accomplished by the IGMI is reported.For the horizontal intensity (H), vertical intensity (Z) and total field (F), measurements undertaken in the period 1977-1981 on 2552 stations by the Finalized Project Geodinamica (PFG-CNR) are used.Information about this survey can be found in Molina et al. [1985].
In order to renew magnetic information on the second order network, data are updated on the basis of the resulting regional models obtained starting from the first order network new results.For the latest surveys (2000,2005) INGV, in cooperation with IGMI, prepared a booklet containing the national maps of the magnetic elements at the specific epoch (for the various procedures utilized in map editing see, for example, Dominici et al. [2007]).

Magnetic measurements
In the latest 2010 survey, traditional magnetic measuring systems for field measurements were used.A portable theodolite fitted with a fluxgate magnetometer, the so called Declination Inclination Magnetometer (DIM), was used for the angular measurements of the Earth's magnetic field, and a proton precession magnetometer was used for the measurement of the field intensity F. All other elements can be derived starting from these measurements.The same instruments are used for the measurements taken at observatories and have, therefore, high intrinsic accuracy.In case of field measurement operations, however, the accuracy of the final values has a larger error than what usually obtained in an observatory environment.Sources of error can occur as a consequence of temperature variations over the period of measurement, or because, in the field, the theodolite is positioned on a tripod, rather than on a more solid pillar (as in an observatory).Newitt et al. [1996] have suggested that, for the mid-geomagnetic latitudes, the typical errors for the elements observed on magnetic networks are: 20" for D, 1.5 nT for H, 1.0 nT for Z, 0.5 nT for F. We assume these errors also for the results of this 2010.0Italian survey.
For the azimuth determination, in the newly occupied stations, and in all cases in which the visibility of the old azimuth marks was limited or lacking, a gyroscopic theodolite was used.This instrument is preferred to Global Positioning System (GPS) instruments, since it can be used for measurements in tunnel or under roof-covered locations, or even where the presence of tall vegetation can cause GPS data problems, because of limited satellite visibility.Moreover, the two instruments show comparable precisions of about ten minutes of arc-second, on average.
As a rule, measurements are performed in series of 10, of which 5 in the morning, and then 5 in the afternoon.The measurements are then undertaken avoiding the time period of maximum magnetic field change due to diurnal variation.Moreover, in order to reduce the final error due to the magnetic field, diurnal variation 02 UT (Universal Time) was selected as the time to which all data were reduced.

Data analysis
After magnetic elements are observed at the repeat stations, the magnetic data need to be reduced to a magnetic quiet time condition.The following procedure was used.As above mentioned, magnetic elements observed at the repeat stations, are reduced firstly to 02 UT (i.e., to correct for the diurnal variation correction) assuming that variometer data can be used as a correction for rapid external field variations under the assumption that their spatial variations are negligible.The value of an element E (i.e., D, I or F) at station s, is reduced at 02 UT of day of measure d, by means of the following:  In order to evaluate the average main field contribution in a given area, magnetic measurements are processed for the determination of a so called normal field.This is an analytical expression in latitude and longitude that best interpolates, for the region of interest, the longest wavelength part of the measured magnetic field.Such a normal field doesn't reflect a well defined physical source, since it can contain long wave crustal origin contributions superimposed to the core origin field; it suffices, however, as an analytical level, from which, for example, the local magnetic anomalies can be derived.where E is the generic geomagnetic element, m the longitude, the latitude (both expressed in degrees or minutes of arc), a 0 is the base level of the models for each element, a n(n=1,5) are the coefficients (in nT/degree or nT/minutes of arc) in longitude and latitude that determine the spatial gradient for each element.
In the INGRF model, the coefficients a n(n=0,5) are determined by a least-squares error minimization technique from the repeat stations data.In order to identify the effect of stations located in anomalous areas, the polynomial coefficients were first computed using all the available stations and then the standard uncertainty associated with each single station (where is the sum of square differences between observed and computed values, n is the number of stations and 6 is the number of coefficients) was computed.The stations with residuals larger than 2v were eliminated, and computation was iterated until no residuals exceeded 2v remained [for details see Molina et al. 1985 andDominici et al. 2007].
The International Association of Geomagnetism and Aeronomy (IAGA) regularly releases the so called International Geomagnetic Reference Field (IGRF): the latest version of this standard mathematical description of the Earth's main magnetic field is IGRF 11 [see IAGA Working Group V-MOD 2010].The coefficients for this degree and order 13 main field model were finalized by a task force of IAGA in December 2009.The IGRF is the product of a collaborative effort between magnetic field modelers and the institutes involved in collecting and disseminating magnetic field data  A second order polynomial in latitude and longitude has been used for computing secular variation for the period 2005-2010; by means of this model all the second order network points were adjusted and reduced to 2010.0.Tables 2  and 3 report the coefficient values of the normal fields and of the secular variation models for the all periods, respectively.
The second order network data were then updated using the latest 2010 survey with the intent of producing new magnetic element maps at 2010.The updated second order network database was interpolated using an "Inverse Distance Weighted" algorithm [see for example Shepard 1968], in order to produce a regular grid (step: 0.05 degree, about 5 km) from which contour lines for magnetic elements were then drawn.In Figure 3  During the 2010 survey, a detailed investigation was performed in Tuscany (central Italy).In this region in fact, the visual inspection of the total field map had revealed a suspected magnetic anomaly, amplitude of about 400 nT, in the Monte Argentario area (Lat.42.5˚, Lon.11.3˚).In this area no natural probable geological source anomaly was found to support this feature.The anomaly was the result of a single data point value in the second order network.In order to verify this datum in the FHZ second order network database, new measurements on 19 stations around the anomalous area, were undertaken.The final result confirmed that the old datum was wrong and then the values in the database were recalculated and corrected.The result of this update is shown in detail in Figure 4.

Future development and Conclusions
Over the last thirty years, since the first complete survey of 1979.0 undertaken under the INGV coordination, the first order repeat station network has been updated every 5 years reoccupying all stations that was possible to repeat.In future, this practice may be subject to a revision for several reasons.One reason is the general reduction in funding to which all scientific institutions are subject, especially for financing future surveys.A second reason is the evaluation of the real scientific need to undertake the survey at such a large number of magnetic repeat stations.The difference between current repeat stations with respect to the INGRF79 (see Figure 5) shows that a more limited number of stations are already representative of the magnetic core field.Indeed 24 stations were found to be adequate to correctly represent the core field evolution.This selection of stations takes into account the geological characteristics, spatial distribution and low values of magnetic anomaly with respect to the core field (repeat stations with anomalies < ±20 nT in F).Using the selected 24 stations the average resulting spacing would increase considerably to around 100 km.This situation was checked also for a possible new reduced regional field; we called this INGRF_reduced, that was computed for all epochs from 1985.0 to 2010.0 at intervals of 5 years.Table 4 reports the coefficient values of the normal fields for the different periods.In view of the above mentioned situation for next mid-term survey (2012.5)we will then make repeat station survey measurements only on the 24 selected stations, represented with stars in the map of Figure 5.
Since all models were developed using independent polynomials for each element, the self-consistency of the resulting field data values must be checked.This consistency may be checked by means of the geometrical condition, that is the difference between the total intensity F and the square root of the sum of squares of the horizontal and vertical In these sites we tested the validity of the regional models by the comparison of the values predicted by models with the annual means of the observatories or the measurements in the stations.In Figure 5 the crosses mark the location of the two observatories of Castello Tesino (CTS) and L'Aquila (AQU).The stars mark the repeat stations of Masseria Lamia (MSL), Nuraghe Crichidoris (NRC) and Contrada Misteci (CNM).In Figures 7 and 8 we show the behaviour of the total field intensity (F), horizontal and vertical intensity (H and Z)

DF F H Z
2 2

= -+
easily be due to a bad elimination of the external field.For the observatory of Castello Tesino the INGRF data behaviour, in the period 1985-1990, is better fitted probably because in this case an intermediate model at 1987.5 was used, based on a specific reduced survey (see Table 1).
The comparison between CTS and CNM (Figure 9) shows we can estimate the behaviour of the various models in latitude, similarly to NRC (Figure 10) and MSL (Figure 11), in longitude.As for the observatories, for these stations the best estimation of horizontal intensity for the      The characteristics of magnetic first and second order networks in Italy and some surrounding areas, the magnetic measurements and reduction procedure, were described in this paper.Following the recommendations of MagNetE Committee, new repeat stations were measured and data finally reduced at 2010.0.The regional and/or normal field computation for the Italian region, an analytical expression of second order degree polynomial type, in latitude and longitude, was chosen for all elements, the coefficients of expression of normal fields at 2010.0 were computed and compared in order to make all possible standard modelling available to the scientific community.

Figure 2 .
Figure 2. Example of a repeat station monograph, prepared for the purpose of keeping note of the magnetic measurements, and to allow a quick determination of the correct location.
E s (2010.0)= E obs (2010.0)+ [E s (02 UT ) d − E obs (02 UT ) d ] (2) where E s (2010.0)= value of E at station s reduced at epoch 2010.0;E obs (2010.0)= average value of E at observatory at epoch 2010.0;E obs (02 UT ) d = value of E at observatory at time 02 UT of day of measure d.

For
the Italian region, in latest years we have computed a few reference fields: the normal fields INGRF (ING Reference Field) and ITGRF (ITalian Geomagnetic Reference Field); for this see for example De Santis et al. [2003].These reference fields are based on a second-degree polynomial expression of the form (3)

Figure 3 .
Figure 3. Top left: contour lines and coloured values of declination, D; top right: total intensity, F; bottom left: horizontal component, H; bottom right: vertical component, Z.All at 2010.0 over Italy.
we show maps of the various field elements, reporting the general behaviour of geomagnetic field over Italy for the epoch 2010.0.

Figure 4 .
Figure 4. Left: map of total field F at 2010.0, before the revision in the Albegna basin.Right: map of F at 2010.0, corrected after re-measured data in local survey (red point in the enlargement).

Figure 5 .
Figure 5. Anomalies of total field F at repeat stations with respect to INGRF at epoch 1979.0;units are nT; stars mark the location of repeat stations selected for a reduced network (see text).

Figure 6 .
Figure 6.Geometrical condition for the normal fields INGRF and ITGRF at epoch 2000.0 and for normal field INGRF_ reduced, at epochs 2000.0 and 2010.0.

Figure 7 .
Figure 7. Annual mean values of F, H, Z, D computed at Castello Tesino observatory (squares) together with the values predicted by various models (coloured lines).

Figure 8 .
Figure 8. Annual mean values of F, H, Z, D computed at L'Aquila observatory (squares) together with the values predicted by various models (coloured lines).

Figure 9 .
Figure 9. Measures of F, H, Z, D at CNM station (squares) together with the values predicted by various models (coloured lines).

Figure 10 .
Figure 10.Measures of F, H, Z, D at NRC station (squares) together with the values predicted by various models (coloured lines).

Figure 11 .
Figure 11.Measures of F, H, Z, D at MSL station (squares) together with the values predicted by various models (coloured lines).

Table 1 .
History of magnetic surveys in Italy from 1640 to present.Acronyms are explained within the text.Additional acronym of UCMG stands for the Italian Central Office of Meteorology.Brackets (…) indicate second order network measurements, i.e.only F, H, Z.Figure 1. Italian repeat station network.Red crosses indicate repeat stations, filled circles indicate observatories and diamonds indicate variometer stations; large circles indicate coverage areas for rapid time variations as recorded at observatories and variometer stations.
) where E s (02 UT ) d = value of E at station s reduced at time 02 UT of day of measure d; E var (02 UT ) d = value of E at variometer station s at time 02 UT of day of measure d; E s (t) d = value of element E observed at station s at time t at that day d; E var (t) d = value of element E at variometer station at time t at that day d.The element E computed at epoch 2010.0 ( January 1 st , 2010) is obtained following:

Table 2 .
Coefficients of normal fields for epochs from 1979 to 2010 for all magnetic elements.Latitude z and longitude m are referred to 42˚ N and 12˚ E respectively, with values expressed in minutes of arc.z = (Lat.− 42˚) in minutes; m = (Long.− 12˚) in minutes.

Table 3 .
Coefficients of normal fields secular variation for epochs from 1979 to 2010 for all magnetic elements.Latitude z and longitude m are referred to 42˚ N and 12˚ E respectively, with values expressed in minutes of arc.

Table 4 .
and declination (D) at the two observatories from 1979 to 2010 (annual mean values) and those predicted by IGRF, ITGRF, INGRF and INGRF_reduced.An offset for the horizontal intensity of the planetary models as IGRF and derived ITGRF appears.This could Coefficients of reduced normal field for epochs 1985 to 2010 for all magnetic elements.Latitude z and longitude m are referred to 42˚ N and 12˚ E respectively, with values expressed in minutes of arc.