Pliocene–Pleistocene Paleomagnetic Secular Variation and Time-Averaged Field From the Northern Volcanic Zone of the Andes, Colombia

Abstract

Paleomagnetic results obtained from 38 Pliocene—Pleistocene volcanic flows from the Knot of the Pastos and surroundings of Puracé volcano and Popayán (southwestern Colombia) are presented. Using stringent quality criteria and excluding sites that classify as representatives of transitional states of Earth's field, a selected group of 27 sites (16 with normal polarity and 11 with reversed polarity) was obtained with a mean direction (Dec = 357.8°, Inc = 6.4°, α₉₅ = 7.5° and κ = 15) that coincides with the geocentric axial dipole field (GAD: Dec = 0^o, Inc = 3.2^o) and, unlike similar studies at similar latitudes, does not coincide with the GAD plus a 5% axial quadrupolar component (Dec = 0°, Inc = −1.08°). However, when serial correlation (SC) among several sites with high inclination anomalies is taken into consideration, the mean direction of two resulting groups of sites shows greater consistency with a field that includes a quadrupolar component. It is interpreted that the sites treated for SC record states of the field similar to today's field in the area of study, which is being affected by the South Atlantic Anomaly.

Introduction

Important efforts have been undertaken by the paleomagnetic scientific community to study the time-averaged field (TAF) of the Earth and its paleosecular variation (PSV). The past 5 Ma period of time, has been usually used for these kinds of studies because tectonic movements are often not big enough to be taken into consideration in paleomagnetic analysis. Phenomenological models of both TAF (e.g., Hatakeyama & Kono, 2002; Johnson & Constable, 1997) and PSV (e.g., Cromwell et al., 2018; McElhinny & McFadden, 1997; Tauxe & Kent, 2004) are based on available regional paleomagnetic studies. Among the several archives used in these kind of studies are volcanic (lava and pyroclastic) flows which are characterized by recording virtually instant moments of Earth's field as episodic eruptive activity in volcanic areas progresses.

This study presents paleomagnetic results from Pliocene—Pleistocene volcanic flows from the Andes of southwestern Colombia, that form part of the predominantly calc-alkaline volcanism of the Northern Volcanic Zone (NVZ) of the Andes (Clapperton, 1993) which extends from Colombia to Ecuador. The area of study can be divided in two subareas shown in Figure 1: (a) the area between the surroundings of the city of Popayán and the Puracé volcano, and (b) the northern part of the orographic complex called Nudo de los Pastos (Knot of the Pastos). These two subareas sit, respectively, in the central (that extends from Nevado del Huila to Doña Juana volcanoes) and southern (that extends from Galeras to Chiles-Cerro Negro volcanoes) segments of the NVZ (Monsalve, 2020; see also Hall & Wood, 1985; Marín-Cerón et al., 2019). In the two areas of study, the oldest eruptive centers have been often partially or completely eroded or destroyed during caldera forming explosive episodes of volcanic activity and newer volcanic structures have been formed. Most of the sampled volcanic products (lava and pyroclastic flows) emanated from volcanic centers that no longer exist or have been partially destroyed.

Figure 1. Location maps of the area of study. (a) General location of the two subareas of study. (b) Area between the surroundings of the city of Popayán and the Puracé volcano (Northwestern most volcano of the Coconucos Volcanic Chain). (c) The northern part of the orographic complex called Nudo de los Pastos (Knot of the Pastos). Yellow dots are sampling sites locations; red circles are the location of volcanoes.

Complementary information of this study is found in Sánchez-Duque (2012). A similar study by De Oliveira et al. (2022) partially overlaps this study's area at GVC and Morasurco volcano.

Paleomagnetic Sampling and Area of Study

Paleomagnetic sampling took place between 5 and 22 August 2008. Sampling was done using a gasoline powered drill, always intending to obtain fresh samples. Sampling sites were aimed at representing individual lava or pyroclastic flows. Most of the sampling sites were accessible by road (except site SW2 close to the crater of Puracé volcano), and located using GPS (configured according to the WGS84 spheroid). At least 8, but most often 10 oriented cores were drilled at each site. Cores were oriented using magnetic compass and sun compass when possible. A total of 40 sites were sampled totaling 406 drilled cores. The stratigraphic and age information about the sampling sites can be found in Text S1 and Figure S1 in Supporting Information S1.

According to the geologic studies upon which sampling was based, the sampled lava flows classify as andesites, dacites and rhyolites. These kinds of rocks are good paleomagnetic recorders when their magnetic minerals undergo deuteric oxidation (which occurs often) or exsolution (that occurs less frequently) during cooling of the lavas (Dunlop & Özdemir, 1997).

Laboratory Methods and Data Analysis

Paleomagnetic analysis was conducted at the Research Center for Paleomagnetism and Environmental Magnetism of the University of Florida (Gainesville, FL). A 2G discrete sample cryogenic magnetometer and a D-tech D-200 demagnetizer, were used in a shielded room. AF demagnetization was applied to all the samples using 12 demagnetization steps from 5 to 100 mT. All samples from three sites (SW1, SW10, and SW13) and several samples from three sites (SW15, SW28 and SW33) had high coercivity and were further demagnetized thermally from 300°C up to 670°C (when necessary) using at least 7 steps. The number of thermally demagnetized samples from sites SW15, SW28 and SW33 was 2, 1 and 4, respectively.

The software package PmagPy 2.73 for paleomagnetic data analysis was used in this study (Tauxe et al., 2016). Paleomagnetic directions from individual samples were obtained using principal component analysis (PCA) of vectors trending to the origin and great circle analysis (Kirschvink, 1980). PCA from individual samples was obtained from at least five demagnetization steps (without anchoring to the origin), and the cut-off value of MAD (maximum angular deviation) was <5.5°. Mean site directions were calculated using Fisher (1953) statistics or when it was applicable, the method for combined analysis of remagnetization circles and direct observations by McFadden and McElhinny (1988).

Quality criteria for selection of successful results from sites were: site-mean directions obtained from at least four samples; a cut-off value of α₉₅ (95% confidence angle around the mean direction) ≤ 5.0° (a value also applied by Mejia et al. (2005); Tauxe et al. (2000)) and a value of κ (precision parameter) > 50 (a value that was also applied by Cromwell et al. (2018)). The identification of transitional sites was made based on the iterative method developed by Vandamme (1994). Several groups of sites were chosen for special analysis, primarily based on the applied selection criteria and the transitional versus non-transitional nature of the sites.

The mean paleomagnetic direction among the analyzed groups of sites was calculated using Fisher (1953) statistics, and the values were compared with TAF and PSV models. The mean direction of the analyzed groups of sites was compared with simple zonal axial-component TAF models composed of a geocentric axial dipole (GAD) and a GAD plus a small percentage (5%, which is usually used) of axial quadrupolar component. In order to analyze PSV, the scatter of the VGP from all the sites relative to Earth's axis of rotation was calculated. Calculation of the VGP scatter (S_b) was made as described by Johnson and Constable (1996), based on the angular standard deviation of the VGP among the groups of sites analyzed and subtracting the effect of within site scatter. The 95% confidence limits of the resultant scatter were calculated using the method described by Cox (1969). Results from the analyzed groups of sites were also compared with results of similar studies from low latitude areas.

Results and Discussion Constraints on Magnetic Mineralogy From Demagnetization Data

The average natural remanent magnetization (NRM) of 37 sampled sites is 2.3 A/m and range from 0.1 to 7.6 A/m (Table S1 in Supporting Information S1). Off this range of NRM values, are two high and one low values corresponding to the remaining three sites. The lowest NRM value (0.005 A/m) was obtained from site SW35 and the highest NRM values were obtained from sites SW2 and SW6 (42.1 and 14.8 A/m, respectively). Medium destructive field (MDF) values of the samples that were AF (and not thermally) demagnetized range on average from 10 to 70 mT (Table S1 in Supporting Information S1). NRM values of most of the sites is consistent with the studied rocks (andesites, dacites and rhyolites). The highest NRM values are associated with high dispersion (Fisher statistics parameter κ) of the NRMs (also shown in Table S1 in Supporting Information S1) and suggests the influence of lightning strikes. Lightning strikes tends to decrease MDF values because a very strong IRM is induced to the low coercivity group of magnetic minerals being affected by this phenomenon. Several of the studied sites appear to having been affected by lightning strikes, for example, sites SW2 (Figure S2 in Supporting Information S1), SW6, SW9 and SW29 (Figure S2 in Supporting Information S1).

Most of the studied sites, that were AF demagnetized (and not thermally demagnetized) have coercivity ranges consistent with the presence of titanomagnetite and titanohematite as magnetic carriers. It is assumed that these minerals are more abundant and less abundant (or absent), respectively, as it commonly occurs in subaerial intermediate and felsic volcanic rocks.

For the samples that were thermally demagnetized, the unblocking temperature ranges at which ≤3% (UT3) of the NRM was demagnetized, was calculated (Table S1 in Supporting Information S1). UT3 range from 500 to 550°C (site SW1) suggests the presence of low Ti titanomagnetite, magnetite and/or low Ti titanohematite as magnetic carriers. UT3 ranges from 625 to 650°C (samples from sites SW28 and SW33) and 650–670°C (samples from sites SW13 and SW15) suggest the presence of low Ti titanohematite or hematite as magnetic carriers. UT3 ≥ 670°C (site SW10) suggests the presence of hematite as a magnetic carrier. As discussed in Section 4.2., of the six sites likely to contain titanohematite or hematite as magnetic carries, one yielded scattered results (site SW10), one was interpreted to be displaced, that is, not in situ (site SW13), one did not pass the selection criteria (site SW15), and only three sites (SW1, SW28 and SW31) form part of the selected group of 27 sites, considered more representative of this study.

Except for site SW10 that yielded scatter paleomagnetic results (Section 4.2), the direction of magnetization of components with unblocking temperatures above 580°C (held by hematite or titanohematite) are similar to those at lower unblocking temperatures. In addition, samples from the same site that were and were not subjected to thermal demagnetization show very similar paleomagnetic directions. These observations support the idea that both high and low unblocking temperatures components of magnetization were acquired virtually at the same time most likely during the emplacement of the lavas and thus carry a primary remanence magnetization.

Zijderveld plots from several of the above mentioned sites are shown in Figure S2 in Supporting Information S1 (SW1, SW15, SW28 and SW33).

Paleomagnetic Results: Paleosecular Variation and Time Average Field

The paleomagnetic results of the sampled volcanic flows are summarized in Table 1 and Figure 2. Site SW10 (not shown in Figure 2) yielded scatter results (α₉₅ = 87) and was not considered for analysis. Sites SW7 and SW36 were found to belong to the same lava flow because of their almost identical directions. In addition, both sites have very similar hand-sample petrographic appearance and are located in the same plateau, roughly 1 km apart. Samples from these two sites were therefore combined (SW7+36).

Table 1 Site-Mean Paleomagnetic Results^a

[Table omitted. See PDF]

^aSite is the sampling site nomenclature, Lat and Long are the latitude and longitude of the site location, N is the number of samples that were collected, n is the number of samples used to calculate the site-mean direction, nl and np indicate the number of lines/planes used in the calculation of site-mean directions which involved either Fisher statistics or the combined analysis using remagentization circles (McFadden & McElhinny, 1988); P is the paleomagnetic polarity: (N) normal, (R) reversed, (T) transitional, (D) site not in situ; Dec and Inc are paleomagnetic declination and inclination; α₉₅ is the 95% confidence cone, κ is the precision paramenter; Geo Unit is the geologic unit or volcano: POP, Popayán Formation; IGN are ignimbrites from POP; COC, Coconucos Formation; CHA, Chagartón member of the Coconucos Formation; CC, CN, GY, U, are the Cariaco, Coba Negra, Genoy and Urcunina stages of the Galeras Volcanic Complex (GVC); G is a basal flow from La Guaca cinder cone at GVC; MOR, undifferenciated lavas from Morasurco volcano; GRA, Guaitara river andesites; MA, Macas andesites; ACA, ancient Cumbal volcano andesites; PBA, Paja Blanca volcano andesites; CD, Colimba Domes and CH, Chiles volcano. Age is the age-range, or single available date, of the geologic unit, or the date obtained at the site location. In this table age-uncertainly range has only been taken into account for ages obtained from sampling-site locations. Plio–Holo is Pliocene–Holocene, Plio–Pleis is Pliocene–Pleistocene. Ref is the reference source of the age: (a) Calvache et al. (1997) based on ⁴⁰Ar/³⁹Ar and ¹⁴C dates; (b) Duque Trujillo et al. (2010) based on ⁴⁰Ar/³⁹Ar; (c) Gonzalez and Zapata (2003); (d) compilation by Gómez et al. (2015); (e) Marquínez et al. (2003); (f) Monsalve and Pulgarín (1999); (g) Orrego and Paris (1999); (h) Parra and Velasquez (2003); and (i) Torres (2010) based on ⁴⁰Ar/³⁹Ar.

Figure 2. Equal area projection of site-mean directions obtained in this study. Filled and empty symbols represent paleomagnetic vectors pointing down and up, respectively; black and brown crossed-out circles represent sites that did not pass the selection criteria and transitional directions, respectively; the brown square is the 14 July 2023 IGRF (Alken et al., 2021). Colors of circles represent sites from: Popayan Formation (light blue); Coconucos Formation (dark blue); Galeras Volcanic Complex (green) and the southern part of the knot of the Pastos (magenta). Serially correlated sites SC1 and SC2 are shown.

An initial group of 38 sites was established (hereafter GRS1, see Table 2). From this group of sites five sites (SW4, SW13, SW14, SW16 and SW31) were classified as transitional for having VGP cutoff latitude <57.6^o, a value obtained using the iterative method of Vandamme (1994). However, it is interpreted that three of these sites (SW13, SW14 and SW16), rather that recording true transitional states of the field, were most likely affected by displacements after their emplacement as lava flows. This conclusion is based on the fact that the three sites are located in a deep incision that “looks like a collapse scar that later has been modified by extensive erosion” (Calvache, 1995). It is also worth noticing that sites SW13 and SW16 have the highest inclination anomaly among all sites (Table 1). Sites GA4, GA5, GA7, GA8 and GA9 sampled by De Oliveira et al. (2022) in the same area, some of them apparently in locations overlapping this study's site-locations, were reported as transitional.

Table 2 Paleomagnetic Statistical Data Among Groups of Sites^a

GRS	Dec	Inc	N	α95	κ	VGP Long	VGPLat	A95VGP	K VGP	O.G.	O.G + 5% Q	St	Sb	Su	Sl	∆I
GRS1	356.6	2.2	38	8.9	8	169.0	86.5	6.90	12	Yes	Yes	23.9	23.8	28.3	20.6	−0.9
GRS2	358.8	6.4	33	7.8	11	215.9	87.9	4.70	29	Yes	Yes	15.2	15.1	18.1	12.9	3.3
GRS3	357.8	6.4	27	7.5	15	234.4	87.3	4.80	34	Yes	No	14.1	14.1	17.2	11.9	3.2
GRS4	357.0	2.9	24	7.4	17	192.4	87.0	4.90	37	Yes	Yes	13.6	13.5	16.8	11.3	−0.1
GRS5	357.3	1.5	22	7.9	17	178.3	87.1	5.30	35	Yes	Yes	14.0	13.9	17.5	11.5	−1.3

^aGRS is the group of sites; Dec and Inc are paleomagnetic declination and inclination; N is the number of sites, α₉₅ and κ are is the 95% confidence cone around the mean direction and its precision parameter, VGP Long and VGP Lat are the virtual geomagnetic pole longitude and latitude, respectively, A₉₅ and K are the 95% confidence cone around the mean VGP and its precision parameter, respectively; O.G./O.G + 5% Q, indicate whether the 95% confidence cone (α₉₅) of the mean direction overlap the directions of an Earth's magnetic field model consisting of a GAD/GAD plus a 5% axial quadrupole component, at the mean latitude among sites. Data of VGP scatter relative to the Earth's axis of rotation is given in columns: S_t (total scatter), S_b (scatter corrected for within-site scatter), S_u (upper 95% confidence limit of the scatter) and S_l (lower 95% confidence limit of the scatter); ∆I is the inclination anomaly.

Comparisons between the magnetic polarity of the studied sites and the expected polarity based on the magnetic polarity time scale (Cande & Kent, 1995) is shown in Figure 3. All sites show expected magnetic polarities except site SW07-36 that has reversed polarity and a K-Ar date of 590 ± 20 ka corresponding to the normal polarity chron c1n (Bruhnes). However, the age range of this site partly coincides with geomagnetic excursion Big Lost that extends 565 ± 10 Ka (Champion et al., 1988; Laj & Channell, 2015). More studies are needed in order to reach conclusions in this regard, considering that the Big Lost excursion has not been fully recognized (Laj & Channell, 2015) and that the radiometric date obtained from site SW07-36 could be somewhat inaccurate.

Figure 3. Magnetic polarity of the studied sites compared with the expected polarity according to the magnetic polarity time scale (Cande & Kent, 1995) and the age of the studied lavas. The age-intervals assigned to the sampling sites are indicated by a vertical line. Normal, reversed and transitional polarities are indicated by filled, empty and half-filled rhombuses. The age-range of site SW7-36 (with reversed polarity) partially coincides with magntic excursion Big Lost (see text). Acronyms are as in Table 1.

By discarding the transitional sites from GRS1, a second group of sites was obtained (hereafter GRS2, see Table 2). The mean direction among sites with VGP latitudes in the northern hemisphere and the antipodal mean direction among sites with VGP latitudes in the southern hemisphere, of groups of sites GRS1 and GRS2 overlap (at the 95% confidence level) making an angle of 8.9° and 2.3°, respectively, implying that these two data set pass the reversal test with a B and an A classification, respectively (McFadden & McElhinny, 1990).

It can be observed in Table 1 that: all sites fulfill the selection criterion of being obtained from at least four samples; six sites do not fulfill the selection criterion of having α₉₅ ≤ 5°; and three sites do not fulfill the selection criterion of having κ values > 50 (all of which do not pass the α₉₅ selection criterion either). By excluding the sites that did not pass the selection criteria (SW8, SW9, SW15, SW19, SW29, SW35) from GRS2, a selected group of 27 sites, that met the condition of being non-transitional and fulfilling the set-up quality criteria was obtained (hereafter GRS3, see Table 2) and it is considered the most representative of this study. The selected group of sites contains 16 normal-polarity and 11 reversed-polarity volcanic flows.

Even though sequences of volcanic flows were not very often observed during field work, an analysis of serial correlation (SC) was made for two groups of sites based on their stratigraphic and assumed stratigraphic identities, respectively. Such analysis has the propose of decreasing the weight in paleomagnetic calculations (among groups of sites) of sites from lava sequences, assuming that such flows oversample disproportionally the paleomagnetic field in time due to the episodic nature of volcanism. In this study the approach of obtaining mean directions and mean VGPs among the serially correlated sites and proceeding to recalculate site-mean directions was applied, as done by Cromwell et al. (2018). The first group of serially correlated sites (SC1) are the rhyolitic ignimbrites from the Popayan Formation conformed by sites SW01, SW37, SW39 and SW40, that form a cluster of normal polarity directions and have relative high inclination (Figure 2). The second group of serially correlated sites (SC2) is composed of sites SW5, SW7+36 and SW38, that also form a cluster of reversed polarity directions (Figure 2). This group of sites is rather speculative in the sense that it has been assumed that the three sites correspond to the Chagarton member of the Coconucos Formation. Site SW7+36 does belong to the Chagarton member (Marquínez et al., 2003; Monsalve, 2020), site SW38 is thought to belong to the Coconucos Formation (Torres, 2010) and, because the sequence of lavas and pyroclastic flows that crop out in site SW38 (samples were taken from an upper pyroclastic flows) sit on top of Paleozoic rocks (Arquia Complex) as described by Torres (2010), it is likely that site SW38 represents one of the oldest members of the Coconucos Formation, probably the Chagartón member. With regard of sites SW5 (a pyroclastic flow), while it is mapped by Marquínez et al. (2003) as Rio Negro member of the Coconucos Formation (stratigraphically above Chagartón member), it is mapped as “undifferentiated lavas and pyroclastics” in the geologic map by Torres et al. (2016) that is focused on the volcanology of the Coconucos Volcanic Chain. Therefore, it is a likely scenario that the pyroclastic flow sampled in site SW5 belongs to the Chagartón member of the Coconucos Formation, which implies that the source of this pyroclastic flow is the Chagartón volcano, with its remnant center (Monsalve, 2020) located roughly 9 km west of site SW5. The mean direction among the serially correlated sites SC1 and SC2 (Table 2) have α₉₅ = 4.9^o and 3.7^o, respectively, less than 5°, the applied selection criteria of site-mean directions in this study. Therefore, two additional groups of sites were analyzed that take into consideration SC1 (hereafter GRS4, see Table 2) and SC1 plus SC2 (hereafter GRS5, see Table 2).

Paleomagnetic statistics from the set-up groups of sites are shown in Table 2. There is coincidence with simple GAD and GAD plus 5% percent quadrupolar component (GADQ) models except for the group of sites GRS3 (Figure 4) which does not coincide with the GAD plus 5% percent quadrupolar component model. However, when SC is taken into consideration in the groups GRS4 and GRS5 coincidence with GADQ reappears even though α₉₅ values remain almost unchanged with respect to other groups. It is indeed expected that the groups of sites filtered for SC tend to approach closer to the field model that takes into consideration the quadrupolar component, due to the relative high inclination anomalies of the filtered sites which tend to mask the shallowing effect of the quadrupolar component (Figures 5a and 5b). The high inclination anomalies (relative to GAD) of sites SC1 and SC2 (Table 1 and Figure 2) are similar in magnitude (in absolute values) to those of the present IGRF (Alken et al., 2021) in the area of study which is affected by the South Atlantic Magnetic Anomaly (SAA), suggesting that at the time of their emplacement, the studied area was affected by this or a similar geomagnetic structure. This observation is consistent with the SAA being recurrently active in the geologic past as concluded in other studies (e.g., Engbers et al., 2020; Hare et al., 2018; Shah et al., 2016; Tarduno et al., 2015).

Figure 4. Equal area projection of the mean direction among the selected group of sites (GRS3) and other directions for comparison. (a) Mean direction of group of sites GRS3 and its 95% confidence cone (magenta circle); mean direction of group of sites GRS4 (light blue circle); mean direction of group of sites GRS5 (dark blue circle); GAD at the latitude of the area of study (dark brown square); GAD plus a 5% quadrupolar component (light green square); and IGRF (Alken et al., 2021) for 14 July 2023 (light brown square). Filled and empty symbols represent magnetic vectors pointing down and up, respectively. See text and Table 2 for an explanation about groups of sites. (b) Magnified view of area shown in (a).

Figure 5. Inclination anomaly relative to GAD of groups of sites with α95 confidence range and of an Earth's field composed of a GAD plus 5% axial quadrupolar component. (a) Inclination anomaly of all the groups of sites analyzed in this study and (b) Inclination anomalies of several low latitude studies summarized in Table 3.

Table 3 Summary of Recent TAF and Paleosecular Variation Studies Between −15^o and 15^o Latitude^a

Reference	Location	Lat.	Long.	Age	Dec	Inc	α95	N	Sb	Sl	Su
Kidane et al. (2003)	Afar, Ethiopia	11.0	42.0	0.6–3.3	359.7	16.4	5.7	26	11.6	9.8	14.3
Cromwell et al. (2013)	Costa Rica	10.1	−84.4	0.04–6	1.4	14.4	7.5	27	15.2	11.9	18.2
Sánchez-Duque et al. (2016)	Ruiz-Tolima Chain	4.9	−75.4	Pleistocene	2.6	4.8	4.4	45	12.6	14.7	11.0
Opdyke et al. (2010)	Loinyangalani, Kenya	3	37	Plio-Plestocene	1.1	−1.0	4.1	32	9.5	7.4	11.3
This Study	Southwestern Colombia	1.6	−77.1	Plio-Plestocene	357.8	6.4	7.5	27	14.05	11.9	17.2
De Oliveira et al. (2022)	Southern Colombia	1.27	−77.24	Plio-Plestocene	351.2	−3.4	6.2	30	15.2	12	17.6
Opdyke et al. (2010)	Mt. Kenya	0	37.5	Plio-Plestocene	1.4	−7.0	3.2	69	9.5	8.5	10.8
Opdyke et al. (2006)	Ecuador	−0.37	−78.3	Plio-Plestocene	359.9	−5.4	4.2	51	13.3	11.6	15.4
Kent et al. (2010)	Galapagos Islands	−0.9	270	0–3	357	1.2	3.0	61	11.4	10.2	13.0
Leonhardt et al. (2003)	Quixaba Formation, Brazil	−3.9	−32.5	1.8–3.3	0.6	−13.5	5.7	20	11.2	9.1	14.6
Elmaleh et al. (2004)	Java and Bawean Islands	−7.4	112	Bruhnes	0.1	−19.3	5.3	36	13.9	12.0	16.6

^aLat and Long are the mean latitude and longitude of the study location, Dec and Inc are the declination and inclination of the mean paleomagnetic direction, α₉₅ is the 95% confidence cone around the mean direction. N is the number of flows used for all summarized statistical calculations. S_b is the VGP scatter with 95% confidence range between S_l (lower limit) and S_u (upper limit). Numerical values under “Age” are in Ma units. Most referenced studies lack or have excluded transitional sites using the Vandamme’s criterion, except for the studies from Costa Rica and Galapagos Islands for which small recalculations were made to exclude trasitional sites using Vandamme's criterion. This table has been updated from Sánchez-Duque et al. (2016).

By comparing VGP scatter of this study and other studies in a latitude range of 15^o north and south (Figure 6), it can be observed that the selected group of sites from this study (GRS3) is among the ones that are more distant from Model G of PSV.

Figure 6. Model G of VGP scatter variation with latitude (line) and its 95% confidence limits (shaded area) plotted along VGP scatter results from published low-latitude studies and this study, summarized in Table 3. The 95% uncertainty range is calculated based on Cox (1969).

Conclusions

Paleomagnetic results from 38 Pliocene—Pleistocene volcanic flows from the central and northern parts of the NVZ of the Andes are presented and analyzed their significance in the context of PSV and TAF theory. The mean direction of a selected group of 27 sites that met stringent quality criteria coincides with the GAD direction at the area of study and does not coincide with a GAD plus a 5% quadrupolar component models of the field. However, there is coincidence with both the GAD and GAD plus 5% quadrupolar component of the field when SC among sites with high inclination anomalies is applied, a procedure made sequentially which decreased the number of data to 24 and 22. VGP scatter approaches Model G of PSV closer using the filtered data sets. This analysis suggests that the high inclination anomalies that characterizes the area of study today as a manifestation of the South Atlantic Anomaly has operated recurrently in the geologic past.

Acknowledgments

This research was funded through a Grant given to V. Mejia by Universidad Nacional de Colombia in 2006 (Hermes code 5178). Permission and companion to perform paleomagnetic sampling within the Puracé indigenous reserve was provided by community authorities. The authors thank valuable guidance about the geology of the area of study given by Maria Luisa Monsalve, Bernardo Pulgarín and Gloria Patricia Cortés, all of them from the Geological Survey of Colombia. Authors are grateful for having been able to do laboratory work at the Research Center for Paleomagnetism and Environmental Magnetism of the University of Florida and for the valuable assistance of Dr. Kainian Huang there. Important recommendations by three anonymous reviewers have improved this paper.

Data Availability Statement

Paleomagnetic data related to this study (Sánchez-Duque et al., 2024) is available at the MagIC (Magnetics Information Consortium) database, link: https://www2.earthref.org/MagIC/search.