The BoM (19∘30′ N, 99∘W; 9600 km2, 2240 m a.s.l.) is a hydrologically closed basin in the TransMexican Volcanic Belt (TMVB, Figs. 1–2). The emergence of the southern Chichinautzin volcanic field since 1.0 Ma (Arce et al., 2013) has been linked to closure of the basin and development of a lake system. Continued subsidence of this intermontane basin accommodated the accumulation of hundreds of meters of lacustrine sediments (Ortiz Zamora and Ortega Guerrero, 2010). Lake Chalco, the location of the MexiDrill site, was the southernmost of these lakes; its proximity to the volcanic peaks that contain the headwaters of streams feeding the lakes has made it the freshest of the lakes and likely the location of the most continuous sedimentation. Mexico City (then called Tenochtitlan) was established by the Aztecs in the 1300s on an island in the center of Lago de Texcoco (to the north of Chalco). In more recent times, the hydraulic regime of the BoM was heavily modified, lowering the water table to improve flood control, accommodate agricultural expansion, and provide water for urban development. The plain of Lake Chalco, located in the southeasternmost sub-basin of the BoM (Fig. 1), has an area of 120 km2 (Caballero and Ortega Guerrero, 1998), though the lake has been reduced to a shallow marshy remnant (Caballero Miranda, 1997).

An array of climate systems affects central and southern Mexico. Episodic northward surges of the ITCZ bring summer rains (Metcalfe et al., 2015). This circulation is distinct from that of the North American Monsoon (NAM). The BoM thus contrasts with northern Mexico and areas of the southwestern USA where the NAM delivers summer precipitation (greater than 60 % of the annual total) driven by land surface heating and convection on the high topography (Adams and Comrie, 1997). Modest winter season precipitation in central and southern Mexico is often associated with the southward penetration of mid-latitude frontal systems, locally known as “Nortes” (Reding, 1992; Pérez et al., 2014; Stahle et al., 2016).

Interannual variability of climate over Mexico is strongly influenced by ocean–atmosphere interactions, including the El Niño–Southern Oscillation (ENSO), the Atlantic Multidecadal Oscillation (AMO), and the Pacific Decadal Oscillation (PDO) (Magaña et al., 2003; Metcalfe et al., 2015; Stahle et al., 2016). Of all of these, ENSO has the strongest impact both in magnitude and stability of the teleconnection between sea surface temperatures and precipitation, but varies with both latitude and season. El Niño is associated with wet winters in the north but with drier summers over southern Mexico including the BoM (Magaña et al., 2003). The net result of these interactions is that modern precipitation in the BoM varies with northern Mexico and the southwestern USA on annual and interannual timescales, when ENSO and PDO have significant impacts. However, when a 20-year low-pass filter removes those influences, BoM precipitation is distinct from both regions (Fig. 2).

The potential value of the Chalco Basin sediment archive has been recognized for decades (e.g., Sears and Clisby, 1952; Bradbury, 1989), and a number of previous studies demonstrated the sensitivity of the system to climate changes, particularly to regional hydrological balance. These studies primarily focused on the past 40 kyr, utilizing rock magnetism and pollen and diatom distributions to reconstruct paleoenvironmental conditions (Bradbury, 1989; Lozano García et al., 1993; Lozano García and Ortega Guerrero, 1994; Urrutia Fucugauchi et al., 1994, 1995; Lozano García and Xelhuantzi López, 1997; Caballero Miranda, 1997; Caballero and Ortega Guerrero, 1998). Broadly, these studies indicate that lake level was highest prior to 35 ka, shallowed from 30 to 22.5 ka, and deepened from 22.5 to 10 ka, with maximum post-LGM freshening between 14 and 10 ka. During the early Holocene, the lake became shallow and marshy (consistent with the general drying trend of central Mexico; Metcalfe et al., 2000). More recently, Lozano-García et al. (2015) identified millennial-scale rainfall variability in the Chalco Basin during MIS 3, observing that regional runoff was lower during the LGM than during deglaciation, which contrasts with some other regional records such as Lago Petén Itzá (Hodell et al., 2008). Torres-Rodríguez et al. (2015) developed an 85 kyr record of drought and related fire history recorded in Chalco sediments showing increased fire frequency during MIS 3 in association with higher spring insolation. Finally, Correa-Metrio et al. (2013) demonstrate that Chalco and Petén Itzá both have similar post-LGM rapid warming histories and that the velocity of climate change at that time was one-fourth that of the last 50 years.

Earlier characterization of deep BoM sediments was motivated by a need to understand the basin's structure for evaluation of seismic hazards and of groundwater resource sustainability. After the 1985 Mexico City earthquake, five deep (∼2 km) boreholes were drilled to characterize sediments and underlying volcanic sequences in the BoM, and to evaluate the Cretaceous carbonate basement. Analyses of well cuttings showed that the uppermost section, a lacustrine sequence of volcanic ash and sand intercalated with clays, is present across the BoM with a variable thickness up to 300 m. These surveys also indicated the presence of a deeper lacustrine sequence, 100–200 m in thickness, composed of gravels, clayey sands, and marls as well as granular volcanic material below an intervening volcanic unit (composed of Pliocene and Quaternary tuffs, lavas, and conglomerates). This lower sequence forms the Chalco Aquifer (Krivochieva and Chouteau, 2003). The survey found the thickest sedimentary pile at the Tulyehualco-1 borehole, located at the western end of the Chalco Basin, leading subsequent researchers to target that locale for further study.

Seismic reflection surveys are a commonly used approach for selecting a drill site location with a long undisturbed record of continuous deposition. In lacustrine or marine conditions, acquisition of suitable seismic reflection data is a relatively straightforward exercise. Depending upon the depth of the target, suitable images of the sedimentary section can be easily achieved using either single-channel or multi-channel marine seismic surveying techniques. Collecting multi-channel seismic reflection data on land is far more challenging. Unlike the marine environment where the source and receivers are deployed in a relatively homogenous medium (the water column), on land the near-surface material is typically heterogeneous, often significantly negatively impacting the propagation of seismic energy.

In the Chalco sub-basin, the near-surface conditions have been severely impacted by the intensive agricultural development of the area. This resulted in significant reworking and amendment of unconsolidated near-surface soils and sediment. As a consequence it is difficult for surface-deployed seismic sources to couple sufficient energy into the underlying sediments to collect any meaningful seismic reflection data. Under such conditions, collection of high-quality seismic reflection data would require that the source (and possibly the geophones) be buried below the disturbed near-surface. Alternatively, Vibroseis vibrator trucks could inject energy into the underlying basin to produce meaningful reflections. Collecting the survey using either of these approaches would not have been feasible, as they would have entailed a field cost comparable to that of the actual drilling program.

We undertook a seismic reflection survey in 2011, with a strategy of using a Mini-Sosie source composed of an array of earth compactors to generate a randomly swept signal that could be processed in a similar fashion to Vibroseis surveys. However, once this survey was initiated it became apparent that the source was not capable of generating sufficient energy to pass through the unconsolidated and disturbed upper sections of Chalco sediments and produce meaningful reflections. Ultimately, the survey was acquired using a PEG-40 source, which generates seismic energy by striking a metal plate with an accelerated mass. The shot records produced by this source were significantly better than those generated by the Mini-Sosie, but still yielded few reflections and did not produce satisfactory survey results.

The inherent difficulty in utilizing standard seismic reflection techniques in this setting led us to employ a suite of non-traditional geophysical techniques to build a consistent picture of sediment thicknesses across the Chalco Basin. These techniques – subsidence mapping, gravity surveys, and passive seismic surveys – do not provide information on continuity of sedimentation or the presence of faults offsetting the sediments, but provided a consistent body of evidence that indicated a maximum sediment thickness of ∼300 m at the identified drilling location. In addition, transient electromagnetic (TEM) and magnetotelluric (MT) soundings undertaken at the time of the 2016 drilling campaign (Bücker et al., 2017) provide similar conclusions.

Subsidence mapping. Since the 1950s several areas of Mexico City, including the Chalco sub-basin, have experienced notable surface subsidence resulting from groundwater withdrawal for municipal needs that has caused the water table to drop. Subsidence rates are broadly correlative with the thickness of lacustrine sediment sequences and may thus be used to evaluate relative depth to basement across the basin. Combined interferometric synthetic aperture radar (InSAR) and GPS data reveal that parts of Mexico City have experienced subsidence rates that exceed 350 mm yr-1 (Cabral Cano et al., 2008, 2010). In the Chalco sub-basin these rates exceed 200 mm yr-1 and indicate maximal sediment thickness near the center of the basin (Fig. 3).

During the 2016 ICDP Mexidrill drilling campaign, three long drill cores (1A, 1B, 1C; see Table 1) were retrieved in the Chalco Basin, recovering the entire lacustrine sequence (upper 300 m) and continuing into the volcanic basement, down to 522 m total depth (Fig. 7). A suitable drilling site was identified in the agricultural fields in the Tulyehualco district, near the depocenter of the lake basin. Drilling was performed by Major Drilling de Mexico S.A., using a truck-mounted Boart Longyear LF90 rig with standard Boart Longyear HQ3 wireline diamond coring tools (Fig. 4); 1.5 m long core sections of 61 mm diameter were collected in polycarbonate liners. HWT casing was advanced as needed to stabilize the formation during continued drilling, and later retrieved, and drill fluid returns were processed in a centrifuge to separate solids and allow recirculation. All drilling used municipal water delivered to the site, with a tracer added to allow post-drilling assessment of fluid contamination in the core samples (Friese et al., 2017). Drilling operations began on 23 February 2016 and concluded on 30 March 2016, in two daily shifts of 12 h each. To improve recovery in the uppermost soft sediments, an additional shallow fourth hole (1D) was cored by hand on 1–2 April 2016 with a Usinger-type percussion corer (Mingram et al., 2007), with cores extruded into secondary liners.

On retrieval, cores were extracted from the drilling tool, capped, sealed with tape, and labeled according to standard LacCore/CSDCO protocols. Drilling metadata were captured throughout operations and transferred to ICDP and to the LacCore Facility for integration with all project datasets. Cores were passed through a Geotek MSCL-S with a magnetic susceptibility loop sensor after drilling, for high-speed, low-resolution assessment of stratigraphic completeness. At the conclusion of drilling, cores were crated and shipped via air freight to the LacCore Facility at the University of Minnesota for processing, description, scanning, subsampling, and permanent curation in the LacCore repository in refrigerated conditions (4 ∘C).

The Leibniz Institute for Applied Geophysics (LIAG) conducted geophysical downhole logging in the 1A borehole. Initially, the upper section (0–195 m) was logged, and after drilling of the second part down to 420 m, the logging was completed to total depth. The following tools (measurements) were run: spectral gamma ray (SGR: total gamma ray; potassium, uranium, thorium), magnetic susceptibility (Susz: susceptibility with standard and micro resolution), focused electric resistivity (Fel), dipmeter (caliper and resistivity), acoustic borehole televiewer (borehole image), sonic (velocity vp), as well as salinity and temperature. All the instruments – except the SGR – require an open hole. Therefore, after measuring the SGR through the pipe over the complete depth, the drill string was pulled successively to guarantee the stability of the borehole, and the remaining tools were run. The first data processing step is to splice the data to produce a continuous log over full depth. Figure 5 shows some of the geophysical properties over the complete length (0–420 m) of the 1A borehole. Gaps within the logs were caused mostly by challenging borehole conditions. Gamma ray data show quasi-cyclic variations in different period ranges, possibly representing glacial–interglacial to sub-orbital climate variations (cycles). This is also depicted in a wavelet analysis plot (Fig. 6), showing that variability in the range below ∼10 m is quite consistent throughout the upper 300 m. In contrast, longer cycles are dominantly present in the lower interval from ∼150 to 300 m depth. The magnetic susceptibility responds primarily to volcanic material in this location. As a result, it is generally low above ∼250 m, with few spikes reflecting volcanic events (tephra layers or aquatic influx of volcanic material). The increase in resistivity and velocity reflects the higher degree of compaction with increasing depth.