\makeatletter
\@ifundefined{HCode}
{\documentclass[screen,CRGEOS,Unicode,biblatex,published]{cedram}
\addbibresource{crgeos20241103.bib}
\newenvironment{noXML}{}{}
\let\citep\parencite
\let\citet\textcite
\def\xcitealp#1#2{\citeauthor{#1}, \citelink{#1}{#2}}
\def\citealt#1#2{\citeauthor{#1}, \citelink{#1}{#2}}
\newcommand*{\citelink}[2]{\hyperlink{cite.\therefsection @#1}{#2}}
\def\defcitealias#1#2{}
\let\citetalias\textcite
\let\citepalias\parencite
\def\thead{\noalign{\relax}\hline}
\def\endthead{\noalign{\relax}\hline}
\def\tbody{\noalign{\relax}}
\def\tabnote#1{\vskip4pt\parbox{.95\linewidth}{#1}}
\def\tsup#1{$^{{#1}}$}
\def\tsub#1{$_{{#1}}$}
\def\ndash{\text{--}}
\def\hyphen{\text{-}}
\def\og{\guillemotleft}
\def\fg{\guillemotright}
\def\0{\phantom{0}}
\makeatletter
\g@addto@macro{\UrlBreaks}{\UrlOrds}
\gappto{\UrlBreaks}{\UrlOrds}
\RequirePackage{etoolbox}
\usepackage{upgreek}
\usepackage{pifont}
\def\inlinefig#1{\includegraphics{#1}}
\def\jobid{crgeos20241103}
%\graphicspath{{/tmp/\jobid_figs/web/}}
\graphicspath{{./figures/}}
\def\xmorerows#1#2{#2}
\newcounter{runlevel}
\usepackage{multirow}
\def\nrow#1{\@tempcnta #1\relax%
\advance\@tempcnta by 1\relax%
\xdef\lenrow{\the\@tempcnta}}
\def\morerows#1#2{\nrow{#1}\multirow{\lenrow}{*}{#2}}
\let\MakeYrStrItalic\relax
\def\refinput#1{}
\let\ubreak\break
\csdef{Seqnsplit}{\\}
\def\botline{\\\hline}
\def\back#1{}
\newcommand*{\hyperlinkcite}[1]{\hyper@link{cite}{cite.#1}}
\DOI{10.5802/crgeos.331}
\datereceived{2024-12-31}
\daterevised{2025-04-07}
\datererevised{2026-03-05}
\dateaccepted{2026-03-09}
\ItHasTeXPublished
}
{
\PassOptionsToPackage{authoryear}{natbib}
\documentclass[crgeos]{article}
\usepackage[T1]{fontenc}
\usepackage{pifont}
\def\CDRdoi{10.5802/crgeos.331}
\def\citelink#1#2{\citeyear{#1}}
\def\xcitealp#1#2{\citealp{#1}}
\let\ubreak\relax
\makeatletter
}
\makeatother

\usepackage{upgreek}

\dateposted{2026-04-14}
\begin{document}

\begin{noXML}

\CDRsetmeta{articletype}{research-article}

\TopicFR{Zone critique et socio-\'ecosyst\`emes}
\TopicEN{Critical zone and socio-ecosystems}

\editornote{Article submitted by invitation}
\alteditornote{Article soumis sur invitation}

\title{Exploring the critical zone of a tropical megacity: insights
from environmental magnetism}

\alttitle{Exploration de la zone critique d'une m\'{e}gapole tropicale:
Enseignements tir\'{e}s du magn\'{e}tisme environnemental}

\author{\firstname{Let\'{i}cia} \lastname{Rangel Dantas}\CDRorcid{0000-0002-7774-8709}\IsCorresp}
\address{Instituto de Astronomia, Geof\'{i}sica e Ci\^{e}ncias
Atmosf\'{e}ricas, Universidade de S\~{a}o Paulo, S\~{a}o Paulo,
Brazil}
\email[L. Rangel Dantas]{lrdantas@usp.br}

\author{\firstname{Andr\'{e}a} \lastname{Teixeira Ustra}\CDRorcid{0000-0002-5230-1775}}
\addressSameAs{1}{Instituto de Astronomia, Geof\'{i}sica e Ci\^{e}ncias
Atmosf\'{e}ricas, Universidade de S\~{a}o Paulo, S\~{a}o Paulo,
Brazil}

\author{\firstname{Rosely}\nobreakauthor\lastname{Aparecida\nobreakauthor Liguori\nobreakauthor Imbernon}\CDRorcid{0000-0002-7886-1658}}
\address{Escola de Artes, Ci\^{e}ncias e Humanidades, Universidade de
S\~{a}o Paulo, S\~{a}o Paulo, Brazil}

\author{\firstname{Janine} \lastname{Ara\'{u}jo do Carmo}\CDRorcid{0000-0003-3683-3648}}
\addressSameAs{1}{Instituto de Astronomia, Geof\'{i}sica e Ci\^{e}ncias
Atmosf\'{e}ricas, Universidade de S\~{a}o Paulo, S\~{a}o Paulo,
Brazil}

\author{\firstname{Ricardo} \lastname{Hirata}\CDRorcid{0000-0001-9683-1244}}
\address{Instituto de Geoci\^{e}ncias, Universidade de S\~{a}o Paulo, Brazil}

\author{\firstname{Fabiana} \lastname{Pioker}\CDRorcid{0000-0001-8739-4599}}
\addressSameAs{2}{Escola de Artes, Ci\^{e}ncias e Humanidades, 
Universidade de S\~{a}o Paulo, S\~{a}o Paulo, Brazil}

\shortrunauthors

\keywords{\kwd{Critical zone science}
\kwd{Anthropogenic impact}
\kwd{Soil pollution}
\kwd{Iron biogeochemical cycle}}

\altkeywords{\kwd{Science des zones critiques}
\kwd{Impact anthropique}
\kwd{Pollution des sols}
\kwd{Cycle biog\'{e}ochimique du fer}}

\thanks{S\~{a}o Paulo Research Foundation (FAPESP Grant no.
2021-14808-6), Conselho Nacional de Desenvolvimento Cient\'ifico e
Tecnol\'ogico (CNPq grant no. 304695/2022-0), CAPES/FAPESP/CNPQ}

\begin{abstract}
Humanity is confronted with unprecedented challenges related to climate
change, water scarcity, and food security, as underscored by the United
Nations Sustainable Development Goals (SDGs). Overpopulated megacities,
such as S\~{a}o Paulo, are particularly susceptible to these issues,
which are anticipated to have significant environmental and social
consequences. Tackling these emerging challenges necessitates a
holistic approach to comprehending the increasing human influence on
physical, biological, and social environments. This paper emphasizes
the Critical Zone Observatory (CZO) initiative in a crucial area of
S\~{a}o Paulo, a megacity that has undergone various urban
transformations over the years. The S\~{a}o Paulo CZO seed site aims to
address vital questions concerning the anthropogenic impacts on
groundwater, soil, and vegetation. We present environmental magnetism
techniques and geochemical analyses carried out to enhance our
understanding of the structure and dynamics of the critical zone. The
findings reveal notable discrepancies between soil analyses obtained
through geochemical and environmental magnetic methods. Magnetic
parameters have pinpointed a significant interface in the saturated
zone, revealing transformations in mineral phases and grain sizes of
iron-bearing minerals. At greater depths, magnetic properties
distinguished samples closer to the Tiet\^{e} River, which is heavily
affected by sewage, potentially indicating geochemical changes caused
by interactions with polluted waters. Environmental magnetism offers
insights into the transformations occurring within both biotic and
abiotic processes. These dynamic processes are frequently influenced by
anthropogenic factors (e.g., soil contamination) and climatic events
(e.g., droughts and floods). Therefore, environmental magnetism serves
as a valuable tool for monitoring and understanding the resilience of
the critical zone. 
\end{abstract}

\begin{altabstract} 
L'humanit\'{e} est confront\'{e}e \`{a} des d\'{e}fis sans
pr\'{e}c\'{e}dent li\'{e}s au changement climatique, \`{a} la
rar\'{e}faction de l'eau et \`{a} la s\'{e}curit\'{e} alimentaire,
comme le soulignent les Objectifs de d\'{e}veloppement durable (ODD)
des Nations Unies. Les m\'{e}gapoles surpeupl\'{e}es, telles que
S\~{a}o Paulo, sont particuli\`{e}rement vuln\'{e}rables \`{a} ces
probl\`{e}mes, qui devraient avoir des cons\'{e}quences
environnementales et sociales importantes. Relever ces nouveaux
d\'{e}fis exige une approche holistique afin de mieux comprendre
l'influence humaine croissante sur les environnements physiques,
biologiques et sociaux. Cet article met en lumi\`{e}re l'initiative de
l'Observatoire de la Zone Critique (CZO) dans une r\'{e}gion cl\'{e} de
S\~{a}o Paulo, une m\'{e}gapole qui a connu diverses transformations
urbaines au fil des ann\'{e}es. Le site pilote du CZO de S\~{a}o Paulo
vise \`{a} r\'{e}pondre \`{a} des questions essentielles concernant les
impacts anthropiques sur les eaux souterraines, les sols et la
v\'{e}g\'{e}tation. Nous pr\'{e}sentons des techniques de
magn\'{e}tisme environnemental ainsi que des analyses g\'{e}ochimiques
r\'{e}alis\'{e}es afin d'am\'{e}liorer notre compr\'{e}hension de la
structure et de la dynamique de la zone critique. Les r\'{e}sultats
r\'{e}v\`{e}lent des divergences notables entre les analyses de sols
obtenues par des m\'{e}thodes g\'{e}ochimiques et celles issues du
magn\'{e}tisme environnemental. Les param\`{e}tres magn\'{e}tiques ont
permis de localiser une interface significative dans la zone
satur\'{e}e, r\'{e}v\'{e}lant des transformations des phases
min\'{e}rales ainsi que de la granulom\'{e}trie des min\'{e}raux
ferrif\`{e}res. \`{A} plus grande profondeur, les propri\'{e}t\'{e}s
magn\'{e}tiques ont permis de distinguer les \'{e}chantillons
situ\'{e}s \`{a} proximit\'{e} du fleuve Tiet\^{e}, fortement
impact\'{e} par les eaux us\'{e}es, ce qui pourrait indiquer des
modifications g\'{e}ochimiques li\'{e}es aux interactions avec ces eaux
pollu\'{e}es. Le magn\'{e}tisme environnemental apporte ainsi un
\'{e}clairage pr\'{e}cieux sur les transformations intervenant au sein
des processus biotiques et abiotiques. Ces processus dynamiques sont
fr\'{e}quemment influenc\'{e}s par des facteurs anthropiques (par
exemple la contamination des sols) et par des \'{e}v\'{e}nements
climatiques (tels que les s\'{e}cheresses et les inondations). Le
magn\'{e}tisme environnemental constitue donc un outil pr\'{e}cieux
pour surveiller et mieux comprendre la r\'{e}silience de la zone
critique.
\end{altabstract}

%\input{CR-pagedemetas}

\maketitle

\vspace*{1pc}

\twocolumngrid

\end{noXML}

\defcitealias{yu2024}{ibid.}

\section{Introduction}\label{sec1}

Megacities present numerous socio-environmental problems associated
with the patterns of \mbox{development} and transformation of space. One of
the main concerns from scientists has been identifying, monitoring, and
understanding the impacts of mitigation and/or environmental
restoration processes, and possible future changes in climate and land
in urban areas and their peripheries. In studies of climate and
anthropogenic impacts on the environment, the concept of a critical
zone and its monitoring are important instruments for understanding the
growing influence of humans. 

The ``Critical Zone'' (CZ), i.e., the zone between the inner atmosphere
and the rocks through which the groundwater circulates, supports all
the complex interactions between rocks, soil, water, air and organisms
that sustain life on Earth \citep{banwart2012}. The concept
encompasses its physical boundaries, the knowledge developed about its
functioning, and its recognition as a system vulnerable to human
activities \citep{lee2023}. Nevertheless, most studies still focus on
low-impact environments. Urbanization, although a major driver of
land-use and socioeconomic change, remains weakly integrated into
Critical Zone research \citep{yu2024}. This limitation restricts a
more comprehensive understanding of how urban transformations affect
the structure and functioning of this system.

Urbanization drives changes in hydrological and biogeochemical cycles
\citepalias{yu2024}. At the same time, fluvial corridors connect terrestrial
and aquatic environments and play a central role in mass and energy
exchanges. These environments include the active river channel, the
hyporheic zone, and the floodplain. Although geology and climate are
recognized as primary controls, the physical effects of human
interventions remain underrepresented in conceptual models of the
Critical Zone \citep{wymore2023}.

The challenge of crossing disciplines and scales of time and space to
understand the CZ has motivated scientists to establish networks of
Critical Zone Observatories (CZO) to investigate CZ processes in
different environments, covering gradients of various environmental
variables (e.g., climate, topography, lithology, land use, biology and
time). CZOs became an international initiative for developing
observatories along global gradients, through \mbox{collaborative} and
multidisciplinary research, aiming to understand the chemical,
physical, geological and biological processes that shape the Earth's
surface and sustain terrestrial life.


Multidisciplinary investigations explore specific topics that
contribute to important questions of the Critical Zone Science (CZS):
(1) What controls the properties and processes of the CZ? (2) What will
be CZ's response to climate change and land use? (3) How can advancing
knowledge of CZ be used to increase the resilience and sustainability
of ecosystems? These wide-ranging questions allow for diverse
approaches to quantify the role of environmental forces in the
biogeochemical processes that govern the long-term sustainability of
soil and water resources, and their responses to anthropogenic forces. 

Soil's biogeochemical processes have a wide influence on the function
of the CZ, structured by flows of energy and matter. Although
(bio)geochemical reactions occur on a molecular scale, they have
implications on the CZ's ability to sustain human life
\citep{perdrial2015}. In soil, water meets various exposed surfaces,
which affects the transfer of solutes in the heterogeneous environment,
driving different biogeochemical reactions which in turn control the
quality of the surface and groundwater. Geochemical gradients develop
due to spatial and temporal heterogeneity and are therefore affected by
groundwater and precipitation events.

The redox processes control the fate and toxicity of the elements in
the CZ, as well as of numerous organic pollutants. The dissolved oxygen
present in the upper parts of the soil quickly depletes in the
saturated zone. As a result, redox gradients develop. In this
environment, minerals that carry iron are of particular interest,
because they are abundant in soil and because they contain iron in
different redox states, thus functioning as donors and recipients of
electrons  \citep{liurev2012}. In addition, natural organic matter
plays a key role in electron transfer with mineral surfaces, as well as
the various microbial communities found in soils. More broadly, the
processes of transfer of electrons in soil can impact the redox state
and the bioavailability of nutrients such as N, S, Fe or Mn.

In the CZ, iron occurs in two main redox states: oxidized, being ferric
iron Fe (III), and reduced, ferrous iron Fe (II)  \citep{kappler2021}.
Under aerobic conditions, Fe (II) is rapidly oxidized to Fe~(III),
forming insoluble oxides and hydroxides, reducing the bioavailability
of iron in the aqueous phase and limiting biological production in
ecosystems \citep{kendall2012}. Under anaerobic \mbox{conditions}, Fe (II) is
used as an electron source by \mbox{phototrophic} \mbox{bacteria} to fix CO$_{2}$
\citep{weber2006}. Magnetic phases such as ferrihydrite also help to
stabilize organic matter (OM) against microbial degradation,
influencing soil carbon storage  \citep{eusterhues2011}. On the other
hand, alteration processes of magnetic minerals can also degrade OM,
releasing CO$_{2}$ and CH$_{4}$ into the atmosphere \citep{dong2023}.
Therefore, because they are coupled to the cycle of other elements,
such as carbon (C), transformations of iron (hydr)oxides can be an
indicator of biogeochemical changes in terrestrial ecosystems
\citep{raiswell2012}.

In contaminated soils, Fe- and Mn-bearing minerals, as well as organic
matter, are key compounds because they interact with each other and
constitute important electron shuttles. Their solubility and structure
control the mobility of many essential and toxic elements. For example,
Mn (IV) oxides, which are among the strongest oxidants in nature,
control the redox state of elements such as Cr, and are frequently
associated with the release of toxic Cr (VI) in natural systems.
Mixed-valence Fe minerals carrying both Fe (II) and Fe (III) (e.g.,
green rusts and magnetite) are of major interest, particularly in redox
transition zones within the CZ, and are reported in numerous studies to
act as strong reducing agents for nitrite, chlorinated and
nitroaromatic compounds, as well as elements such as Cr (VI), Se (VI)
or U (VI) \citep{davranche2020}.

The formation of iron oxides and hydroxides in soils results from
biogeochemical interactions between biotic (microorganisms and
vegetation) and abiotic (minerals, pH, O$_{2}$ and water) factors
\citep{jordanova2017}. In addition to environmental conditions, soil
magnetic mineralogy also strongly depends on the origin and,
consequently, on the initial mineralogical composition of the particles
\citep{maxbauer2016}.

Magnetic particles present in soils can be of lithogenic, pedogenic,
biogenic, or even anthropogenic origin. Lithogenic minerals are
inherited from the parent rock and usually have larger grain sizes. Due
to their smaller surface area and greater surface stability, primary
iron oxides are less \mbox{influenced} by chemical weathering. Pedogenic
magnetic minerals are formed by the transformations of mineral phases
during soil development \citep{jordanova2017}. Iron-reducing bacteria
are capable of precipitating magnetic minerals, degrading organic
matter \citep{lovley1987} or transferring electrons through the
process known as Direct Interspecies Electron Transfer (DIET)
\citep{xiao2018}. Like pedogenic particles, particles originated by
microbial processes are also nanometric (5--15~nm in diameter)
\citep{cornell2003}. Magnetic particles formed because of human
activities, such as combustion processes, can be dispersed as
atmospheric dust and deposited in the most superficial layers of the
soil. They are micrometric particles (2.5--10.0~$\upmu$m), reactive
and with significant potential for adsorption of heavy metals, which
can cause serious impacts on human health and significant contamination
of soil and water \citep{cornell2003,grobety2010}.

Environmental magnetism plays an important role in the study of
variations in soil magnetic properties, due to the sensitivity of iron
oxides and hydroxides to changes in environmental conditions
\citep{liurev2012}. The joint application of environmental magnetism
parameters aims to identify the mineralogical composition,
concentration, and granulometry of magnetic minerals in each
environment \citep{evans2003}. These parameters have been used in
studies of soil pollution \citep{barbosa2019}, biogeochemical processes
in contaminated areas \citep{moraes2022,ustra2021}, and correlation
between magnetic mineral content and environmental parameters, such as
precipitation \citep{balsam2011}. However, few studies have taken an
integrated approach in the context of critical zone science. A
multi-technique and transdisciplinary analysis allow a deeper
understanding, or new insights into the processes related to the iron
biogeochemical cycle.

In this paper, we show the results of the environmental magnetism study
of two land impacted sites of the S\~{a}o Paulo CZO seed site. Inspired
by existing CZO networks, this project aims to establish a stable
research infrastructure for long-term monitoring to study anthropogenic
impacts on CZ. We hope environmental magnetism techniques can help to
understand biotic/abiotic processes and redox changes in degraded urban
sediments. 

\section{S\~{a}o Paulo CZO seed site} \label{sec2}

\vspace*{-2pt}

The S\~{a}o Paulo CZO aims to understand how soil, water and vegetation
resources in a tropical megacity are responding to natural and
anthropogenic drivers. We explore the hypothesis of biotic and abiotic
redox controls the fate of pollutants and organic matter in CZ. Redox
reactions nourish iron-bearing minerals transformations, affecting
carbon, sulfur, and nitrogen cycles. Even though there is good
knowledge about soil reactions, there is great concern about how these
reactions will be impacted by climate and anthropogenic perturbations,
over long-time scales. In degraded land environments, Fe- and
Mn-bearing minerals and organic matter (OM) interact constitute
important electron shuttles. In fact, the diversity of microbial
metabolisms in soil originates from the energy source these organisms
need, as well as from the redox conditions.

CZOs can better identify and monitor biogeochemical processes in an
urban environment, where land is severely degraded by the presence of
different contaminants (organic and inorganic). A better understanding
of biogeochemical processes can be incorporated into quantitative
models and predict soil responses to land degradation and to climate
changes in urban areas in tropical climate. That is, to ``earthcast''
the CZ in the urban environment.

The CZO is structured in the disciplinary axes of geophysics,
biogeochemistry and hydrogeology. The database of the implementation
phase is composed of: geophysical imaging, soil geochemistry and
magnetic properties, groundwater geochemistry, atmospheric and
biological. We hope the observations enable the conception of soil
response models to anthropogenic impacts and may lead to earth cast the
CZ at a subtropical megacity.

\vspace*{-2pt}
\section{Methods}\label{sec3}
\vspace*{-2pt}

\subsection{Site introduction}\label{sec31}
\vspace*{-2pt}

This CZO is composed of three sites: Campus Capital USP Leste and
Parque Ecol\'{o}gico Tiet\^{e} (PET) as the impacted environment and
Parque Natural Municipal Fazenda do Carmo (PNMFC) as the natural
environment. 

USP Leste/PET location is under the Tiet\^{e} River influence area,
where historical data indicate intense anthropic activity. The studied
site is situated at a floodplain of the Tiet\^{e} River, in S\~{a}o
Paulo---SP, Brazil (Figure~\ref{fig1}). The area contains a series of
anthropogenic deposits (${\sim}$4~m) that overlies Quaternary fluvial
sediments (${\sim}$6 m) and Neogene sandstones. The entire site was
formerly used as a large drying pool for dredged sediments when a
nearby channel was opened. The sediments settled in the pool kept under
anoxic conditions the organic-rich sediments of the fluvial plain
inducing methanogenic processes and methane trapping within sandy
lenses at different depths of the section \citep{mendonca2015}. 
\mbox{PNMFC}
(not shown here) on the other hand, is in the same hydrographic
context, but the area is a Mata Atl\^{a}ntica biome preservation unity.

\begin{figure*}
\vspace*{-4pt}
\includegraphics{fig01}
\vspace*{-4pt}
\caption{\label{fig1}S\~{a}o Paulo CZO seed site localization. EACH
(USP Leste campus) is indicated by the white polygon. Red, yellow, and
green pins indicate the location of the drill cores.}
\vspace*{-6pt}
\end{figure*}

\vspace*{-2pt}
\subsection{Soil sampling}\label{sec32}
\vspace*{-1pt}

Soil samples were collected on 11/08 and 11/09/2022 during the drilling
of water monitoring drill cores, whose locations were defined by
previous geophysical acquisitions. Parameters were obtained for soil
samples from three drill cores, two in the Tiet\^{e} Ecological Park
(PET-1 and PET-2) and one in the School of Arts, Sciences and
Humanities (EACH-1). For the characterization of magnetic minerals,
samples were collected every 0.5 m. For the chemical analyses, sampling
was based on changes in color and texture observed during drilling,
with depth intervals ranging from 0.5 to 1.0~m for most samples.
Until the start of data acquisition for magnetic characterization, the
soil was stored in a refrigerator to minimize the alteration processes
of the magnetic minerals. Wet samples were prepared in cubic acrylic
boxes (7.0~cm$^{3}$) and capsules (only for measurement on a vibrating
sample magnetometer). Wet samples were used in order to minimize the
alteration processes of the magnetic minerals. The mass of the samples
was measured before acquisitions for subsequent data normalization.

\vspace*{-2pt}

\subsection{Soil geochemical}\label{sec33}
\vspace*{-1pt}

All samples were dried in an environment free of dust, gas, etc., at
room temperature, and then they were distorted, quartered, and milled
with ceramic balls up to 200 Mesh (74~$\upmu$m) granulometry. The
samples were analyzed by X-ray fluorescence spectrometry. The levels
presented were measured in a pressed sample, in the Standardless
calibration, relative to the analysis without standards of the chemical
elements comprised between fluorine and uranium, in an X-ray
fluorescence spectrometer, Malvern Panalytical brand, Zetium model. The
values were normalized to 100\%.

Geochemical analyses were used to determine the content (\%) of major
elements expressed in terms of oxides. In this work, only the contents
of SiO$_{2}$, Al$_{2}$O$_{3}$, Fe$_{2}$O$_{3}$, K$_{2}$O were used as
indicators of chemical weathering intensity, mineralogical composition
and granulometry, to corroborate the textural analyses. A loss of
ignition (LOI) was performed at 1020\textdegree~for 2~h. This
parameter represents the difference in the material weight, in
percentage. This difference is due to the loss of structural water from
clays or organic matter.

\subsection{Environmental magnetism}\label{sec34}

Magnetic properties of soil vary according to magnetic mineralogy,
magnetic minerals content and grain size. Magnetic susceptibility (MS)
variations can provide important information on minerals
transformations that result from (bio)geochemical changes. Laboratory
additional techniques are needed to provide specific information,
\mbox{unraveling} the magnetic minerals environment and fingerprints of
biogenic minerals.

Magnetic susceptibility ($\chi$)
is influenced by all magnetic
behaviors in the sample, such as diamagnetism (e.g., quartz),
paramagnetism (clay minerals), ferri- and antiferromagnetism (iron
oxides and hydroxides), associated with the total mineralogical
composition. Percent frequency-dependent magnetic susceptibility
($\chi_{\mathrm{FD}}\%$), is sensitive to ferromagnetic particles with
grain size in the range of superparamagnetic to stable single domain
(SSD)  \citep{dearing1996}.

Magnetic susceptibility (MS) is the capacity of a material to acquire
magnetization during the application of a uniform magnetic field.
Mass-normalized MS is given by ${\chi}$ (in m$^{3}$/kg).
Frequency-dependent magnetic susceptibility is a technique particularly
used to identify the presence of superparamagnetic (SP) particles 
\citep{dearing1996,evans2003}. SP particles can be identified by their
fast magnetization relaxation, observed by the loss of MS with the
increase of the external field oscillating frequency. We carried out
frequency-dependent MS measurements with MFK1 (AGICO Ltd.) at two
frequencies (F1 ${=}$ 976~Hz and F2 ${=}$ 15616~Hz), operating at room
temperature and in an AC field of 200 A/m. Each sample was measured 10
times, for each frequency, and then normalized by mass, with the data
expressed in m$^{3}$/kg. To quantify the SP response contribution to
magnetic susceptibility, $\chi_{\mathrm{FD}}\%$ was estimated as the
relative percentage of MS loss between the measurements at the
different\break frequencies.

Thermomagnetic curves are useful for identifying magnetic mineralogy by
identifying magnetic properties changes that occur at specific
temperatures, such as the Curie temperature
$T_{\mathrm{C}}=580$~\textdegree C  (from which ferri or ferromagnetic
minerals lose magnetization) or the N\'{e}el temperature
$T_{\mathrm{N}}=675$~\textdegree C (used for antiferromagnetic
minerals). 

Magnetic minerals can be altered during the heating steps, giving rise
to secondary minerals. Low-temperature acquisitions avoid this change,
allowing primary minerals to be identified by their low-temperature
transitions, such as the Morin (for hematite) and Verwey (for
magnetite) transitions, respectively \citep{dunlop1997}.
Thermomagnetic curves were acquired in two intervals: from room
temperature to 700~\textdegree C (high temperature curve, with heating
and cooling of the sample); and below room temperature (low temperature
curve). The measurements were carried out on the Kappabridge KLY-4
equipment (AGICO Ltd.), in an argon atmosphere to minimize sample
oxidation.

Magnetic carriers' size and mineral phase will influence the way
magnetic particles acquire remanent acquisition. Isothermal remanent
magnetization (IRM) can be measured after exposure to a field at
ambient temperature. Depending on the magnetic coercivity of the
mineral and the artificial field intensity, all magnetic moments will
be magnetized in alignment with the external field, and the induced
magnetization will not change when increasing the applied field,
reaching saturation isothermal remanence (SIRM). The anhysteretic
remanent magnetization (ARM) is a stable magnetization used as a proxy
for magnetite concentration. The ARM \mbox{measurement} involves magnetizing a
sample using a low field in the presence of an alternating magnetic
field that is smoothly reduced to zero.~

The ARM/SIRM ratio can be analyzed as an indicator of magnetite grain
size. Higher values of the ratio indicate a decrease in grain size,
while lower values indicate an increase in the size of the magnetic
population \citep{evans2003}. For ARM acquisition, the samples were
subjected to a 100~mT AF superimposed on a constant field of 
50~$\upmu$T. IRM acquisition was performed, inducing magnetization in the
following steps: 1000~mT and backfield 300, and 100~mT. Measurements
were performed at Laborat\'{o}rio de Paleomagnetismo of Universidade de
S\~{a}o Paulo (USPMag). Remanence was measured in a cryogenic
magnetometer type SQUID (Superconducting Quantum Interference Device),
model 755U (2G-Enterprises), and induction by a pulse magnetizer model
MMPM10 (Magnetic Measurements Ltd.), located in a shielded room with
ambient field of less than 200 nT. The values obtained were used to
calculate environmental magnetism parameters 
\mbox{(S-ratio,} HIRM and SIRM).

S-ratio provides an estimate of the relative number of grains with high
and low coercivity remanence, such as hematite (antiferromagnetic) and
magnetite (ferrimagnetic), respectively  \citep{bloemendal1992}. It is
defined by Equation~(\ref{eq1}):
{\begin{equation}\label{eq1}
\mbox{S-ratio} = 0.5\times (({\mathrm{IRM}}_{1000\,\mathrm{mT}}
-{\mathrm{IRM}}_{-300~\mathrm{mT}})/{\mathrm{IRM}}_{1000\,\mathrm{mT}})
\end{equation}}\unskip
where IRM$_{1000~\mathrm{mT}}$ is the remanence for an applied field of
1000~mT, also referred to as SIRM. IRM$_{-300~\mathrm{mT}}$ is the
remanence for a backfield of 300~mT. Values close to 0.5 indicate that
more than 50\% of the remanence is held by ferrimagnetic minerals. 

HIRM parameter represents the absolute concentration of
antiferromagnetic minerals in natural samples. It is given by
Equation~(\ref{eq2}):
{\begin{equation}\label{eq2}
\mathrm{HIRM} = 0.5\times (\mathrm{IRM}_{1000~\mathrm{mT}} + 
\mathrm{IRM}_{-300~\mathrm{mT}}).
\end{equation}}\unskip

The behavior of magnetic minerals can be identified by the shape of the
hysteresis curves, defined by the following parameters: saturation
magnetization (Ms), saturation remanent magnetization (Mrs), coercivity
or coercive field (Hc) and remanence \mbox{coercivity} (Hcr) 
\citep{dunlop1997,tauxe2008}. From these curves, it is possible to
differentiate the influence of the magnetic phenomena present in the
mineralogical assembly, namely: diamagnetism, as in quartz and
feldspar; paramagnetism, as in clay minerals, biotite and pyroxenes;
and \mbox{ferromagnetism}. The latter is observed in ilmenite
(antiferromagnetism), hematite (parasitic ferromagnetism), magnetite
and pyrrhotite (ferrimagnetism). 


Hysteresis curves also provide information about the domain structure
of the magnetic minerals present in the samples and, consequently,
about grain size. Ratios calculated from the hysteresis parameters
(Mrs/Ms and Hcr/Hc) were used as estimates of the domain states of the
magnetic carriers and, consequently, of the grain sizes present in the
samples. It is important to highlight that the Hcr parameter is
obtained during the acquisition of the IRM curves.

The hysteresis curves were obtained using a MicroMag Vibrating Sample
Magnetometer (Princeton Instruments corp.). A maximum field of 1000~mT
was used, and, after data acquisition, the paramagnetic effect was
corrected for the maximum magnetic field of 700~mT.


IRM curves were also obtained using MicroMag Vibrating Sample
Magnetometer. Magnetization acquisition was performed at 200 points,
with nonlinear distribution, and the maximum applied field was 1000~mT.
Data were analyzed using the online software MAX UnMix 
\citep{maxbauer2016}. The program was built using the shiny package for
R studio and allows the quantification of the contribution of the
magnetic phases that make up the IRM acquisition curve of each sample.
This is done by fitting the curve using log-Gaussian functions,
separating components with different coercivity spectra 
\citep{kruiver2001,robertson1994}.

Parameters based on anhysteretic remanent magnetization (ARM) are
influenced by SSD ferromagnetic particles \citep{egli2002},
functioning as a qualitative indicator of magnetic grain size
contribution. Saturation isothermal remanent magnetization (SIRM) is
associated with the remanence-bearing minerals contribution,
encompassing grains of low and high coercitivities. HIRM refers to
minerals of high coercivity (e.g., hematite and goethite). The S-ratio
is a qualitative indicator of the relative proportion of low and high
coercivity minerals.

\section{Results}\label{sec4}

\subsection{Textural and geochemical soil characterization}\label{sec41}

Profiles of geochemical and textural variation at depth were generated
from soil samples collected during drilling of monitoring wells in
PET-1, PET-2 and EACH-1 (Figure~\ref{fig2}). 

\begin{figure*}
\vspace*{-4pt}
\includegraphics{fig02}
\vspace*{-3pt}
\caption{\label{fig2}Distribution of geochemical and textural
parameters in PET-1 (close to Tiet\^{e} River), PET-2 (close to SP-070
highway) and EACH-1.}
\vspace*{-5pt}
\end{figure*}

In PET-1, the sandy-silty to fine sandy texture occurs up to
approximately 2.5~m in depth. In the most superficial samples (up to
1.0~m), the presence of processed wood and debris was observed. This is
consistent with previous environmental assessment reports in the PET
region, which identified a fill thickness of less than 2.0~m in depth
\citep{conam2017}. At 4.0~m, the transition from coarse sandy sediment
to gravelly to clayey occurs. The latter presents a greenish-gray
coloration, indicating reduced sediment. After some time exposed to
atmospheric air, the formation of an outer layer with a yellowish color
change was noted. However, during sampling, reddish portions were
observed in the reduced sediment at 5.5, 6.5 and 7.0~m. This suggests
that the material was already in these conditions in the subsurface.

Geochemical data also show a well-marked contrast at 4.0~m depth.
SiO$_{2}$ content ranges from 45.80\% to 93.90\%, increasing with
depth. Al$_{2}$O$_{3}$, Fe$_{2}$O$_{3}$ and the loss of ignition
parameter (LOI) show a decreasing trend. Their values range from
27.40\% to 2.97\% (Al$_{2}$O$_{3}$), 7.65\% to 0.95\%
(Fe$_{2}$O$_{3}$), and 18.90\% to 0.25\% (LOI). On the other hand, the
K$_{2}$O content shows a more expressive decrease in the vadose zone
and a more subtle decrease in the saturated zone, with values between
2.70\% and 0.95\%. An increase in the SiO$_{2}$ content associated with
the decrease in the others is compatible with the leaching process and
predominance of silicates that are more stable to chemical weathering,
such as quartz, muscovite and heavy minerals. This is corroborated by
the subtle decrease in K$_{2}$O values and the identification of
muscovite in the finer-grained samples, in addition to the observation
of partially prismatic minerals in the coarser-grained samples. Loss of
ignition values are higher below 3.0~m depth, where clayey material
predominates, and may be associated with dehydration of clay minerals
or organic matter content. The same does not occur at the
sandy-gravelly level, where there is a sharp decrease in LOI, which
increases again at the reduced clayey level.

In PET-2, the soil has a sandy-silty to fine sandy texture up to
approximately 2.0~m in depth. The presence of processed wood and debris
was also observed in samples up to 1.0~m. At 2.0 to 2.5 m, a
predominantly black clayey layer occurs, possibly associated with
variations in the water table. Clayey-silty sediments occur up to 4.4 m
in depth, followed by a sandy-gravelly level of up to 5.5 m. Above this
depth, clayey levels are observed interspersed with levels of
predominantly fine sand texture, both with color indicative of reduced
sediment.

SiO$_{2}$ content ranges from 23.60\% to 93.80\%. Al$_{2}$O$_{3}$,
Fe$_{2}$O$_{3}$ and LOI range from 29.70\% to 3.09\%, 4.84\% to 1.18\%,
59.10\% to 0.21\%, respectively. K$_{2}$O content presents higher
values in the more clayey levels, ranging from 3.59\% to 0.25\%. All
oxides of major elements show a decrease, while there is an increase in
the LOI value, in the water table \mbox{variation} zone, suggesting a layer
rich in organic {matter}. Increase in the SiO$_{2}$ content together with
the decrease in the other oxides is compatible with the predominance of
silicates that are more stable for chemical weathering, with their peak
value in the sandy-gravelly level.

In EACH-1, the soil has a sandy-clay texture (up to 2.3 m) and
sandy-clay to silty-clay in the water table variation zone (up to 3.0
m). During sampling, the presence of debris, processed wood, and
plastic waste was also frequently observed. Above 4.0~m in depth, dark
clayey to silty-clay layers occur, with a sandy level close to 5.0~m.
Previous studies identified a fill thickness of around 4.0~m 
\citep{conam2017}. Dark clayey to silty-clay layer has been associated
with sediments rich in organic matter (TOC ${>}$ 40\%),  removed from the
Tiet\^{e} River channel during its rectification process 
\citep{mendonca2015}.

Major element contents range from 49.10\% to 88.10\% (SiO$_{2}$),
29.40\% to 6.52\% (Al$_{2}$O$_{3}$), 7.90\% to 1.26\%
(Fe$_{2}$O$_{3}$), while the variation of the loss of ignition (LOI)
parameter is from 15.00\% to 0.88\%. The variation of these values is
relatively homogeneous along the profile, except for LOI (at 5.5 m),
due to the presence of organic matter. The subtle increase in
Al$_{2}$O$_{3}$ and decrease in SiO$_{2}$ at this depth is also
associated with the increase in clay content.
\vspace*{-1pt}

\subsection{Magnetic properties distribution}\label{sec42}
\vspace*{-1pt}

Figure~\ref{fig3} shows profiles of magnetic parameters variation at
depth that were generated from soil samples collected during drilling
of monitoring wells in \mbox{PET-1,} PET-2 and EACH-1. 


\begin{figure*}
\includegraphics{fig03}
\vspace*{-3pt}
\caption{\label{fig3}Distribution of magnetic, and textural parameters
in PET-1 (close to Tiet\^{e} River), PET-2 (close to SP-070 highway)
and EACH-1. ${\chi}_{\mathrm{F}1} =$ low frequency susceptibility (976~Hz);
${\chi}_{\mathrm{F}2} =$ high frequency susceptibility (15616~Hz);
${\chi}_{\mathrm{FD}}\% =$ percent frequency-dependent magnetic
susceptibility; ARM ${=}$ anhysteretic remanence acquired at  100~mT;
SIRM ${=}$ isothermal remanence acquired at 1000~mT; S-ratio ${=}$ low
coercivity minerals contribution; HIRM ${=}$ isothermal remanent
magnetization of high coercivity minerals; w.t.\ ${=}$ water table.}
\vspace*{-1pt}
\end{figure*}
     
In PET-1, low-frequency magnetic susceptibility profile ($\chi_{\mathrm{F}1}$)
shows three sets of well-pronounced values are observed, higher
($3.58\times  10^{-7}~\mathrm{m}^{3}$/kg) from surface to 1.0~m, lower
($2.02\times 10^{-8}$ to $5.14\times  10^{-9}~\mathrm{m}^{3}$/kg)
between 1.0 and 3.7~m, and intermediate ($6.37\times  10^{-8}$ to $6.29
\times 10^{-8}~\mathrm{m}^{3}$/kg) from 4.0~m depth. Higher values
close to the surface may be associated with the magnetic enhancement of
the soil profile, mainly due to the presence of pedogenic magnetite
and/or maghemite \citep{jordanova_environ2017}. The contrast at 4.0~m depth may
be directly associated with the transition from sandier to clay
material. This can also be observed in the ARM/SIRM profile, which is
sensitive to stable particles that retain remanence 
(SSD grain).\looseness=-1

In $\chi_{\mathrm{FD}}\%$, the distribution of the proportion of
superparamagnetic (SP) nanometric particles is more heterogeneous (6.8
to 14.1\%) up to 4.0~m, and more homogeneous (4.0 to 5.1\%) above this
depth. $\chi_{\mathrm{FD}}\%>6\%$ suggests a significant proportion of
SP grains, while $\chi_{\mathrm{FD}}\%<5\%$ indicates a predominance of
SSD grains or extremely fine grains (${<}$5~nm) of the SP fraction 
\citep{dearing1996}. $\chi_{\mathrm{FD}}\%>10\%$ indicates a
predominance of SP grains. The main $\chi_{\mathrm{FD}}\%$ peak is in
the transition zone from the vadose zone to the saturated zone, which
may indicate the influence of possible fluctuations in the water table,
and consequent redox variations, on the formation of SP particles.
Lower ARM/SIRM values, above 4.0~m, also indicate the presence of
ultrafine particles.


Higher S-ratio values (above 0.8) are relatively constant, indicating
that more than 50\% of the remanence is retained by low-coercivity
minerals, which are corroborated by the lower HIRM values. In contrast,
the HIRM peaks indicate that high-coercivity particles are predominant
in such deep samples. This is also evidenced by ratio values lower than
0.5, corresponding to more than 50\% of the remanence retained by
high-coercivity particles. The ARM/SIRM values are also consistent with
the S-ratio values, indicating that more than 50\% of the remanence is
retained by low-coercivity SSD grains.

SIRM/${\chi}$ is used to indicate the presence of iron sulfides 
\citep{peters1998}. However, the increase in ${\chi}$ values,
associated with the presence of paramagnetic and superparamagnetic
minerals, reduces the ratio, which may mask the presence of sulfides,
when they occur in smaller quantities. SIRM/${\chi}< 4$~kA/m indicates
low-coercivity magnetite/titanomagnetite, while  SIRM/${\chi} >
500$~kA/m indicates pyrrhotite. In the latter case, greigite presents
slightly lower values, but like the more coercive magnetite. In PET-1,
the values range from 2.02 to 105.66~kA/m, indicating that there is no
presence of sulfides in the samples.

In PET-2, $\chi_{\mathrm{F}1}$ values show a more heterogeneous distribution.
However, higher values (from $2.49\times 10^{-8}$ to  $4.53\times
10^{-8}~\mathrm{m}^{3}$/kg) are observed above the water table, and
lower values (from  $1.10\times 10^{-9}$ to $1.31\times
10^{-8}~\mathrm{m}^{3}$/kg) are observed in the saturated zone. At a
depth of 2.0~m, $\chi_{\mathrm{F}1}=-2.73\times 10^{-9}~\mathrm{m}^{3}$/kg and
is associated with a clayey material rich in organic matter
(diamagnetic behavior). The contrast between the more sandy and clayey
sediments is not as expressive in the magnetic parameters of this well.
This may be due to the greater number of intercalations, making the
difference in susceptibility more subtle.

$\chi_{\mathrm{FD}}\%$ shows a larger (6.0 to 15.4\%) and more
heterogeneous distribution of SP particles from 3.0 to 7.0~m depth,
compared to the PET-1 samples. $\chi_{\mathrm{FD}}\% < 5\%$ are
observed in the vadose zone (1.5~m) and in the saturated zone (7.5~m),
which may indicate a greater influence on SSD grains or finer grains of
the SP fraction, based on the higher and lower ARM/SIRM values,
respectively. $\chi_{\mathrm{FD}}\% <2\%$ appears only in the
organic-rich interval. This suggests the predominance of particles that
do not exhibit superparamagnetic behavior, such as coarser grains
multidomain (MD)  \citep{dearing1996}.

S-ratio values (0.77 to 0.95) present a homogeneous distribution, with
lower values more concentrated above the water table. In other words,
the influence of low coercivity minerals predominates, mainly in the
saturated zone. The HIRM profile shows a bimodal distribution, with
higher \mbox{values} \mbox{predominantly} in the vadose zone, indicating a higher
concentration of high coercivity minerals in this environment. ARM/SIRM
values are partially compatible with the S-ratio values, indicating
that more than 50\% of the remanence is retained by low coercivity SSD
grains, mainly in the saturated zone. It is also noted that the peaks
of the S-ratio occur mainly in the water table variation zone and in
more clayey levels in the saturated zone.

SIRM/${\chi}$ ranges from 7.93 to 149.41 kA/m, indicating the absence
of iron sulfides in the PET-2 samples.

In EACH-1, $\chi_{\mathrm{F}1}$, in addition to a heterogeneous distribution,
also shows a concentration of higher values (from $3.31\times 10^{-8}$
to $7.17\times 10^{-8}~\mathrm{m}^{3}$/kg), above the water table, and
lower values (from $1.03\times 10^{-8}$ to $3.23\times
10^{-8}~\mathrm{m}^{3}$/kg), below. No significant contrasts are
observed that could be associated with the granulometric variation of
the material, although the lower $\chi_{\mathrm{F}1}$ values at depth may be
related to the clearer, reduced intervals.

In $\chi_{\mathrm{FD}}\%$, the distribution of the proportion of
superparamagnetic (SP) particles is more homogeneous in the vadose zone
(5.0 to 5.9\%). These values, compared with the variations in the
ARM/SIRM profile, may indicate a lower influence of SP particles in the
intervals with a higher ratio. In the saturated zone, lower (3.4 and
4.6\%) and higher (5.7 to 11.7\%) values of $\chi_{\mathrm{FD}}\%$,
associated with low ARM values, may indicate variation in the
granulometry of SP grains, with significant proportions above 5.0~m
depth.

S-ratio values can be divided into three intervals. The first, above
the water table, with values from 0.67 to 0.80. The second and third,
in the saturated zone, with values from 0.72 to 0.83 and from 0.86 to
0.90, respectively. Lower S-ratio values are associated with higher
HIRM values, indicating an influence of high-coercivity minerals,
mainly in the vadose zone. Below the water table, the opposite
relationship suggests that more than 50\% of the remanence is retained
by low-coercivity grains, with their predominance above 6.5 m depth. In
the ARM/SIRM profile, a well-marked contrast can be observed by the
vadose zone/saturated zone transition zone, below which the highest
values are associated with higher S-ratios, indicating the presence of
low-coercivity SSD particles. On the other hand, in the vadose zone,
there is less influence of these particles on the ARM/SIRM \mbox{ratio}.


SIRM/${\chi}$ parameter presents values from 8.27 to 134.34 kA/m,
suggesting the absence of iron sulfides.\looseness=-1

\subsection{Magnetic grain size evaluation} \label{sec43}

Hysteresis loops were obtained for soil samples collected during
drilling of the monitoring wells PET-1, PET-2 and EACH-1. Subsequently,
mass normalization and correction of the paramagnetic effect were
performed, with a maximum magnetic field of 700~mT to better observe
the influence of ferromagnetic particles. The curves exhibit para- and
ferromagnetic behaviors, with a significant influence on paramagnetic
minerals, such as micas and clay minerals, a result consistent with the
description of the soil samples.

Figure~\ref{fig4} shows examples of hysteresis curves for samples from
the three drill cores, at different depths.


\begin{figure*}
\vspace*{4pt}
\includegraphics{fig04}
\vspace*{4pt}
\caption{\label{fig4}Hysteresis curves of soil samples from drill cores
in the Tiet\^{e} Ecological Park (PET-1 and PET-2) and the School of
Arts, Sciences and Humanities (EACH-1), close to surface (vadose zone)
and at depth (saturated zone). Ms ${=}$ saturation magnetization; Mrs ${=}$
saturation remanent magnetization; Hc ${=}$ coercivity.}
\end{figure*}

In PET-1 and PET-2, the Mrs/Ms and Hcr/Hc values indicate mixtures of
low coercivity particles (such as magnetite) in the SP-SSD interval. In
contrast, in EACH-1, the ratios suggest PSD magnetite particles. In the
saturated zone, the soil samples from drill cores PET-1 and PET-2
present distinct behavior both in terms of composition and grain size.
In the first case, the curve and ratios are compatible with MD hematite
grains \citep{ozdemir2014}, while in the second case, they
indicate the presence of MD magnetite particles.

\subsection{Magnetic mineralogy}\label{sec44}

Isothermal remanent magnetization (IRM) acquisition and demagnetization
curves were acquired for soil samples from drill cores PET-1, PET-2,
and EACH-1. Representative samples were selected from each well, two in
the vadose zone, one in the transition zone (${\sim}$2.0~m)
and the other in the saturated zone, which exhibits two main patterns
(Figure~\ref{fig5}). The sample at 0.5 m was not recovered in EACH-1.
IRM curves indicate contributions from low (magnetite) and high
coercivity minerals (such as hematite and goethite), in which the
\mbox{magnetization} does not saturate below 1000~mT. The exception is PET-2
and EACH-1 (7.0~m), in which the curve saturates below 300~mT,
suggesting only a contribution from low coercivity minerals.


\begin{figure*}
\vspace*{-4pt}
\includegraphics{fig05}
\vspace*{-6pt}
\caption{\label{fig5}Isothermal remanent magnetization (IRM)
acquisition curves of samples from drill cores in the Tiet\^{e}
Ecological Park (PET-1 and PET-2) and the School of Arts, Sciences and
Humanities (EACH-1), close to the surface (vadose zone) and at depth
(saturated zone). Hcr ${=}$ remanence coercivity.}
\end{figure*}

In the PET-1 well samples, the influence of high coercivity minerals is
more evident at depth. It is also important to highlight that the
contribution of the magnetization of low-coercivity magnetic carriers,
for a field of 300~mT, represents more than 90\% of the total
magnetization for most samples up to 4.0~m deep. On the other hand,
above 4.0~m, the magnetization of these magnetic carriers does not
exceed 25\% of the total magnetization of the samples. Well-marked
contrast at this depth was also observed in other parameters, such as
${\upchi}_{\mathrm{F}1}$, ${\chi}_{\mathrm{FD}}\%$, and, only for some depths,
in the saturated zone, in the S-ratio and HIRM.\looseness=-1

Coercivity spectra were obtained for the IRM curves
(Figure~\ref{fig6}). Their components are obtained by fitting
log-Gaussian curves, each of which is characterized by three main
parameters: mean coercivity (Bh); dispersion (DP), which quantifies the
scattering of coercivities; and saturation remanent magnetization
\citep{robertson1994}. Skewness (S) and observed contribution (OC)
parameters also provide additional information about coercivity
distribution for the same grain population (component) and relative
contribution, respectively  \citep{maxbauer2016}. \looseness=-1


\begin{figure*}
\includegraphics{fig06}
\caption{\label{fig6}Coercivity components (C) obtained from the
isothermal remanent magnetization acquisition curves of soil samples
from drill cores PET-1, PET-2 and EACH-1, at different depths.}
\end{figure*}

Modeling performed for PET-1, PET-2 and EACH-1 shows two low and high
coercivity components in the vadose zone samples (Table~\ref{tab1}). Samples
with the lowest coercivity correspond to more than 58\% of the
spectrum, with values of 48~mT (PET-1), 67~mT (PET-2) and 54~mT
(EACH-1). The remainder of the spectrum is influenced by the high
coercivity samples, with values of 580~mT (PET-1) and 1021~mT (PET-2),
and 1044~mT (EACH-1). Soil samples in the saturated zone show a common
behavior for the PET-2 and EACH-1 drill cores, consisting of only one
low-coercivity component (24~mT and 65~mT). On the other hand, in
PET-1, one high-coercivity component (575~mT) was necessary to fit the
curve. 

\begin{table*}
\caption{\label{tab1}Coercivity components for soil samples from drill
cores PET-1, PET-2 and EACH-1, at different depths}
\tabcolsep3pt
\begin{tabular}{ccccccccccc}
\thead
\xmorerows{1}{Specimen}  & \multicolumn{5}{c}{Component 1} & \multicolumn{5}{c}{Component 2}\\\cline{2-6} \cline{7-11}
& \parbox[t]{1cm}{\centering Bh (mT)} & \parbox[t]{2cm}{\centering Bh (log~10~units)} 
& DP & S & OC${\cdot}$mean & \parbox[t]{1cm}{\centering Bh (mT)} & 
\parbox[t]{2cm}{\centering Bh (log~10~units)} & DP & S & OC${\cdot}$mean\\  
\endthead
PET1--0.5 m & \048 & 1.68 & 0.43 & 0.77 & 0.86  & \0580 & 2.76 & 0.35 & 0.90 & 0.14\\
PET1--7.0 m & 575 & 2.76 & 0.26 & 0.85 & 1.00 & --   & -- &-- & -- & --\\
PET2--0.5 m & \067 & 1.83 & 0.46 & 0.91 & 0.78 & 1021 & 3.01 & 0.36 & 0.73 & 0.22\\
PET2--7.0 m & \024 & 1.39 & 0.41 & 0.99 & 1.00 & -- & -- & -- & -- & --\\
EACH1--1.0 m & \054 & 1.73 & 0.43 & 1.01 & 0.58 & 1044 & 3.02 & 0.42 & 0.79 & 0.42\\
EACH1--7.0 m & \065 & 1.81 & 0.45 & 0.93 & 1.00 & -- & -- & -- & -- & --
\botline
\end{tabular}
\tabnote{Bh ${=}$ mean coercivity; DP ${=}$ dispersion parameter; S ${=}$
skewness; OC ${=}$ observed contribution.\vspace*{-2pt}}
\end{table*}

Low coercivity values (${<}$25~mT and $\mathrm{DP}=39$--40) near the
surface can be associated with pedogenic particles, while at greater
depth (saturated zone), with particles resulting from microbial
activity \citep{egli2004}. Values close to 38~mT ($\mathrm{DP}=47$) 
can indicate {detrital} magnetite particles. Components with Bh equal to
57~mT ($\mathrm{DP}=45$), 69~mT ($\mathrm{DP}=26$) and 80~mT 
($\mathrm{DP}=32$), despite the high dispersion, can be associated with
high-coercivity magnetite. High DP values, in this case, may suggest
the presence of other components  \citep{maxbauer2016}. For high
coercivity components, Bh above to  200~mT (DP ${=}$ 27) may be
associated with hematite, while Bh well above 1000~mT (DP ${=}$ 25) may
correspond to\break goethite.

Due to the noisy coercivity spectra, it was not possible to model the
sample components for all depths. However, IRM data show a contribution
of high coercivity minerals also below the water table.

\begin{figure*}
\vspace*{6pt}
\includegraphics{fig07}
\vspace*{6pt}
\caption{\label{fig7}High-temperature curves for selected soil samples
from the drill cores of the Tiet\^{e} Ecological Park (PET-1 and PET-2)
and the School of Arts, Sciences and Humanities (EACH-1), close to the
surface (vadose zone) and at depth (saturated zone).}
\end{figure*}

Thermomagnetic curves show the thermal transformations that occur
during heating and subsequent cooling of the sample. Magnetic
susceptibility values (${\upkappa}$) are plotted as a function of
\mbox{temperature}~(T). The ${\upkappa} -T$ curves of fourteen soil samples
from the three drill cores were selected to provide evidence of the
mineralogical composition at different depths and water saturation
levels \mbox{(Figure~\ref{fig7}).} It is noted that the magnetic
susceptibility values of the cooling curves are higher than those of
the heating curves, indicating the formation of a new, strongly
magnetic phase. This behavior can be attributed to the neoformation of
SP and SSD magnetite during heating of the samples in reducing
conditions. This is observed in samples containing iron hydroxides, in
the presence of organic matter, or Fe-bearing clay minerals, or with
adsorbed iron \citep{hanesch2006,hirt1993}.\looseness=-1


Heating curves of the soil samples at 0.5, 1.0 and 2.3 m depth (vadose
zone), in the three drill cores, show the presence of two peaks during
heating, although with different intensities. The increase in magnetic
susceptibility occurs at temperatures of 250~\textdegree C and 
500~\textdegree C. The first peak may be related to the presence of
ferrihydrite being transformed into maghemite. Of the samples from the
saturated zone, this peak is evident only at 5.5~m (PET-2), but it also
appears at 7.0~m (PET-1). Furthermore, at 0.5~m (PET-2), the sharp loss
of susceptibility between 300 and 400~\textdegree C may be caused by
the inversion of fine-grained (pedogenic) maghemite to hematite 
\citep{liu2005}.

The second peak, more evident in the vadose zone soil samples (in all
three drill cores) and more subtle in the saturated zone sample
(EACH-1), corresponds to the reductive formation of magnetite from
hematite \citep{dunlop1997}. The behavior at  500~\textdegree C is also
seen in the transformation of lepidocrocite into maghemite 
\citep{hanesch2006}. Most of the newly formed SP and SSD particles with
a dominant grain size around 35~nm, mainly magnetite, are produced
above 500~\textdegree C  \citep{liu2005}. This is evidenced by the
decrease in magnetic susceptibility at 580~\textdegree C (Curie
temperature of magnetite).

A residual magnetic susceptibility above 600~\textdegree C can also be
observed at 1.0 and 5.5~m (PET-2) and 2.3 and 7.0~m (EACH-1),
suggesting the presence of hematite that is more stable upon heating.
Lepidocrocite can be dehydrated during heating (250~\textdegree C),
becoming maghemite (450~\textdegree C) and, later, hematite. 
\citep{dunlop1997}. At 5.5~m (PET-2), the loss of susceptibility from
300~\textdegree C onwards, without a subsequent increase of
500~\textdegree C, may indicate the inversion of maghemite to hematite
added to the presence of fine-grained hematite particles that are not
reduced to magnetite during heating \citep{liu2005}. The transformation
of maghemite to hematite can also occur from 550~\textdegree C onwards.
This is because maghemite with coarser grains, \mbox{depending} on the degree
of crystallographic perfection, has a higher inversion temperature than
finer grains  \citep{gallagher1968}. Despite the results presented,
\mbox{additional} data from chemical analyses would be necessary to
effectively determine the magnetic mineralogy. This is because the
behavior of the curves may be due to different mineral interactions.

\section{Discussion}\label{sec5}

In all drill cores, higher magnetic susceptibility values occur in
samples closer to the surface, although in PET-1 there is a decrease
from depths above 0.5~m. In this case, the magnetic parameters
influenced by grain composition and size do not show the same
discrepancy between the values at 0.5~m and those at subsequent depths.
The higher $\chi$ value, although reflecting all magnetic contributions
in the sample (para-, dia-, and ferromagnetic), is indicative of the
concentration of these minerals. Furthermore, it is important to
highlight that this is a backfill area  \citep{conam2017}, whose
mineralogical and chemical composition can vary spatially, as evidenced
by the geochemical data, thereby impacting the magnetic signature. High
${\chi}$ values at shallow depths in the other wells may be influenced
by the presence of superparamagnetic (SP) ferrimagnetic minerals
(magnetite and maghemite), related to magnetic enhancement during
pedogenesis  \citep{maxbauer2016}. This phenomenon is related to
well-drained and oxidizing environments, where the pedogenic formation
of magnetite is induced by variations in soil moisture and local redox
conditions \citep{ahmed2018}. However, it usually occurs in the first
meters of the soil \citep{jordanova_environ2017}, a region outside the scope of
the sampling carried out in this work. Therefore, these higher magnetic
susceptibility values in the vadose zone may be due to other factors.
Perhaps differences in anthropogenic deposit interfaces result in the
contrast of this physical property.

Despite the anthropogenic impacts on the sites, all regions are
vegetated, which provides greater nutrient cycling and mobilization of
chemical elements. Both organic matter and water level fluctuations can
affect the redox dynamics and, \mbox{consequently}, the 
\mbox{biogeochemical}
processes that occur in the three sites. This affects the dynamics of
dissolution and precipitation of magnetic phases, since these processes
are impacted by both biotic and abiotic changes.

$\chi_{\mathrm{FD}}\%$ peaks, indicative of the abundance of
superparamagnetic (SP) particles, are more associated with the
influence of water saturation. Peaks in the saturated zone may be
associated with the activity of iron-reducing bacteria that increase
the amount of ultrafine magnetite particles resulting from their
metabolic activity  \citep{lovley1987}. Higher values of ARM/SIRM, a
parameter indicative of single domain (SD) magnetic particle size, also
appear more frequently below the water table. In previous studies in
the EACH area, higher content of SP particles in the saturated zone was
related to methanogenic zones in the deposits of dredged sediments from
the Tiet\^{e} River \citep{ustra2021}. Ultrafine particles are
generated because of methane production through the coupling between
iron-reducing bacteria and methanogenic bacteria, which reduces the
CO$_{2}$ associated with organic matter. In contrast, there is
production of SP magnetite because of the reduction of phases
containing Fe (III)  \citep{lovley1987}.\looseness=-1

This signature was also observed in the PET-2 and EACH-1 profiles and
may be associated with the accumulation of methane in the sandy lenses
intercalated with the clayey layers  \citep{mendonca2015,ustra2021}.
Since the clayey sediments in the saturated zone of PET-2 do not
indicate significant organic matter content, the higher concentration
of ultrafine particles may have another origin. In fact, the iron
oxidation and reduction reactions that produce the ultrafine magnetic
particles are believed to occur cyclically or simultaneously in many
environments. The distinct biotic and abiotic processes are actually
thought to overlap and even compete with each other 
\citep{kappler2021}. However, additional data is needed to corroborate
this interpretation. Regarding PET-1, the same pattern was not
observed, with a low concentration of SP particles predominating below
the water table.

S-ratio values close to 1.0, associated with low HIRM values, suggest
the predominance of magnetite and maghemite along the profiles of the
three drill cores. In PET-2 and EACH-1, higher HIRM values indicated
the significant presence of high-coercivity minerals (such as hematite)
above the water table. The same does not occur in PET-1, whose S-ratio
(lower) and HIRM (higher) values showed a predominance of more coercive
particles at depth (7.0~m).

This suggests that proximity to the Tiet\^{e} River may influence the
dynamics of magnetic mineral formation. A greater occurrence of
low-coercivity \mbox{minerals} (e.g., magnetite) is expected in saturated,
more reducing environments. On the other hand, high-coercivity minerals
(e.g., hematite) tend to occur in more oxidizing environments.
Transformations resulting from the interaction of biogeochemical
processes could alter the redox dynamics of the environment. Another
possibility would be changes during sampling, however, only samples
from the saturated zone of PET-1 present a different coercivity
\mbox{distribution}.

Isothermal remanent magnetization curves and hysteresis data also
corroborated the occurrence of a mixture of minerals with different
coercivities in the vadose zone (PET-2 and EACH-1) and the presence of
hematite at depth (PET-1). In the latter, the behavior in an apparently
reducing environment may indicate the occurrence of some alteration or
punctual biogeochemical process. Mn (IV), nitrate and nitrite can act
as chemical oxidants of Fe (II) under anoxic conditions
\citep{kappler2005}. The thermomagnetic curves also indicated the
occurrence of ferrihydrite in the vadose zone and in the saturated zone
of PET-2. In PET-1 and EACH-1, ferrihydrite was identified only in the
samples from the vadose zone. Ferrihydrite can alter to more stable
minerals such as hematite or goethite.

This study allowed a better understanding of the architecture of the CZ
in a floodplain impacted by urbanization. The information obtained by
the macroscopic description of the samples, contents of major elements,
loss of ignition and mineralogy also indicated three main interfaces:
(a) vadose zone-saturated zone boundary; (b) fine-grained sediments
(${<}$fine sand)-coarse-grained sediments (${>}$coarse sand) boundary; and
(c) coarse-grained sediments (${>}$coarse sand)-fine-grained sediments
(${<}$fine sand) boundary. These boundaries were created by antropogenic
action of laying all the impacted material from the Tiet\^{e} River.

Interface (a) is mainly observed in the magnetic parameters. An
exception is PET-2, whose layer rich in organic matter obliterates the
data dependent on magnetic susceptibility but stands out in the loss of
ignition. Interfaces (b) and (c) have similar characteristics for PET-2
and EACH-1. They are mainly observed in the contrasts of the magnetic
parameters, but also in geochemical data and in the textural analysis,
although there is no clear correspondence with these data for the
EACH-1 samples. Therefore, magnetic data are sensitive to abrupt
changes in texture and mineralogical composition, which can occur in
places where there is backfill deposition.

In fluvial corridors, activities such as sediment mining and
channelization alter morphology, hydrological dynamics, and
connectivity within the Critical Zone. Even historical interventions
continue to influence stratigraphy, biota, and system resilience
\citep{wymore2023}. In urban areas, these effects combine with ongoing
pressures, reinforcing the need to incorporate the human dimension into
Critical Zone analysis. Magnetic properties along soil profiles help
detect these disturbances and separate depositional events from
post-depositional changes, making magnetic mineralogy a useful tool for
interpreting recent floodplain evolution and Critical Zone responses to
human impacts.

From a biogeochemical perspective, urbanization introduces pollutants
and alters nutrient dynamics. Soil compaction changes pore structure,
reducing permeability and gas exchange. In addition, lower inputs of
plant-derived organic matter modify soil carbon storage \citep{yu2024}.

This work also allowed the evaluation of anthropogenic impacts on the
geochemical environment. Magnetic susceptibility, ARM/SIRM profiles,
and the thermomagnetic curves reveal a great distinction between the
PET-1 region, located very close to the Tiet\^{e} River and the park's
lagoon, and PET-2, further away from the park. One hypothesis raised
from this work is that the change in the magnetic properties of PET-2
reflects geochemical alterations caused by the interaction of the
environment with the waters of the Tiet\^{e} River, at depths greater
than 6.0~m. This feature is interpreted as the hyporheic zone, a
critical interface between surface water and groundwater, occurring
beneath and adjacent to stream beds where dynamic mixing of these
waters takes place. It is situated within saturated streambed
sediments, extending laterally or vertically depending on geology (from
centimeters to tens of meters).


The hyporheic zone is far more than a physical transition layer; it is
a dynamic nexus where \mbox{hydrology}, ecology, and geochemistry converge.
The dynamic interface results in distinct redox gradients, impacting
the transport of redox-sensitive \mbox{metals} from the hyporheic zone to the
local watershed, thereby influencing the overall nutrient cycling and
water quality. The interaction of iron-bearing particles within
hyporheic zones involves complex biogeochemical processes influenced by
redox gradients, mineral solubility, organic matter interactions, and
hydrological connectivity. These zones have been recognized as hotspots
for biological activity and cycling of metals and nutrients 
\citep[][and references therein]{dongwater2020}. 

Redox reactions of iron widely influence the biogeochemical cycles of
nutrients (e.g., carbon) as well as the transport of contaminants such
as chlorinated compounds in aquifers  \citep[][and references
therein]{kappler2021}. The production and release of iron are
critically important for geochemical processes in hyporheic zones. For
example, release rates of soluble Fe (II) species resulting from
microbial dissimilatory Fe (III) reduction in these zones are not fully
understood, and could be controlled by site-specific geochemical
factors such natural organic matter  \citep{dongwater2020}.

Redox transformations of iron, such as dissolution and precipitation
and, consequently, the availability of this element, are caused by
chemical processes, generally mediated by microorganisms 
\citep{kappler2021}. The oxidation of Fe (II) can occur under oxic or
anoxic conditions, while the reduction of Fe (III) occurs in anoxic
environments, influencing the dynamics of organic and inorganic
pollutants \citep{kappler2005}.

Based on the physical and chemical parameters measured in the water
from the monitoring wells, seasonally, the environment was
characterized as reducing, with circumneutral pH. In these
environments, despite the low solubility of Fe (III) minerals,
complexation by organic compounds can lead to high concentrations of
dissolved Fe (III) \citep{cornell2003}, as observed in the chemical
analyses of the water. The highest levels of total Fe and dissolved Fe
occurred precisely in well EACH-1, where the sediments rich in organic
matter are deposited.

Sampling with the Dutch auger was not very efficient for recovering the
samples, especially below the water table. In addition, it did not
allow for greater detailing of the profile, due to the auger's zone of
influence being approximately 20 cm high. Therefore, it was difficult
to accurately define the depths at which certain behaviors occur in the
magnetic signatures, beyond the limits of the critical interfaces.


To minimize the effects of alterations resulting from sample handling
and exposure to atmospheric air, measurements were performed in a
manner that preserved collection conditions as much as possible.
However, even during sampling, the brief exposure of samples,
previously in a saturated and reduced medium, may have caused
alterations in less stable and more reactive magnetic phases. The real
extent of this oxidation process and how much it could affect the data
could not be estimated.

Another important point concerns the significant influence of the
paramagnetic signal of clay minerals, especially in the most reduced
soil samples. Parameters dependent on magnetic susceptibility, such as
SIRM/${\chi}$, influenced by the paramagnetic contribution, may have
been underestimated, making a more reliable characterization of
magnetic mineralogy difficult. This contribution also made it difficult
to process and obtain saturation magnetization and coercivity
parameters from hysteresis curves. Therefore, parameters
calculated/obtained from magnetic susceptibility and hysteresis curves,
although in agreement with other independent data (IRM curves, S-ratio,
HIRM, ARM/SIRM), must be analyzed with caution.

Based on the information and data presented above, it would be
interesting to also verify the seasonal influence on the magnetic
parameters of the study areas. This could be used to investigate
whether variations in physical--chemical parameters, such as
oxidation--reduction potential and pH, particularly in water samples
from the Tiet\^{e} Ecological Park, imply changes in magnetic
mineralogy. On the other hand, it could also be verified whether these
variations indicate an influence on surface water bodies, particularly
in the PET-1 well, located between the Tiet\^{e} River and a lake in
the park. This could be done using tracers inserted at specific points
in the lake and river  \citep{hissler2014}, considering the local
topography and hydrodynamics.

It would also be important to perform sampling with higher resolution,
allowing for a more detailed characterization of the successions of
sedimentary deposits and, consequently, of magnetic \mbox{mineralogy}. Due to
the problems reported above regarding sample recovery, a more detailed
collection could be focused on the vadose zone. This is because, in
addition to the greater ease of sampling, it would also align with
other geophysical methods whose signal is significantly influenced by
soil saturation. This would allow for increased resolution in data
\mbox{acquisition} by these other methods and would also be within the zone of
influence of the vegetation root system. This would allow for more
detailed characterizations of the geophysical properties and,
consequently, more complete interpretations regarding the interaction
of processes in the vadose zone.

In summary, despite the challenges, environmental magnetism suggests
transformations that provide insights into biotic and abiotic
processes. These dynamic processes are often affected by anthropogenic
(e.g., soil pollution) and climatic (e.g., droughts and floods)
impacts. Thus, environmental magnetism is a powerful tool in monitoring
and understanding the resilience of CZ. In future studies of the
S\~{a}o Paulo CZO, magnetic properties can help build a conceptual
model of the site, placing the hyporheic zone with respect to the
groundwater flow, placing biogeochemical hotspots that transform iron
and therefore other minerals. With this conceptual model, it is
possible to pursue the main goal of a CZO, which is to predict the CZ's
evolution. 

\section{Concluding remaks}\label{sec6}

PET and EACH revealed striking differences between soil analyses by
geochemical and environmental magnetic techniques. One important
interface can be identified at around 4.0--5.0~m deep, close to the
Tiet\^{e} River. This interface is most clearly visualized in the
environmental magnetic parameters, which show mineral phases and grain
size transformations of Fe-bearing mineral below this depth. This study
also presents evidence of the influence of interfaces with different
granulometric contributions and water saturation on variations in
magnetic grain sizes. This is particularly important when it comes to
urbanized regions, where there is anthropogenic deposition of
unconsolidated material. 

Sampling sites show differences in sediment layers successions,
resulting from anthropogenic processes, such as rectification of the
Tiet\^{e} River, deposition of sediments taken from the river channel
and backfill. Variations in magnetic mineralogy may indicate
biogeochemical hotspots associated with the activity of methanogenic
and iron-reducing bacteria (EACH-1) or the influence of polluted
surface waters (PET-1). 

Magnetic mineralogy applied to floodplain soils provides evidence of
fluvial dynamics and human \mbox{influence} within the Critical Zone. It
supports the \mbox{assessment} of system evolution and resilience under
increasing anthropogenic pressure. Magnetic properties along soil
profiles help distinguish depositional events from post-depositional
changes, offering a useful basis for interpreting recent floodplain
development and the response of the Critical Zone to human impacts.

Organic rich urban tropical soils present unique challenges with
respect to the role of the climate of dissolved organic carbon and its
controls over ecological ecosystems and biogeochemical cycles. The
integrated interpretation of magnetic properties, physicochemical and
geochemical parameters can reveal others dynamic interface, including
the unsaturated/saturated and hyporheic zone, a known biogeochemical
hotspot that drives Fe-bearing mineral transformations. 

This work presents a CZO effort, in a key site within S\~{a}o Paulo
Megacity that has witnessed different urban settings through time. The
S\~{a}o Paulo CZO seed site is dedicated to exploring driving questions
related to anthropogenic impacts on groundwater, soil and vegetation.
In this context, environmental magnetism techniques, geochemical and
biological analysis provided insights into CZ's dynamics. Our results
show differences in sediments that are related to anthropogenic
processes. Variations in magnetic mineralogy indicate biogeochemical
hotspots associated with land degradation. Characterizing the
architecture and dynamics processes in the subsurface of an urbanized
region allows an integrated understanding of the role of anthropogenic
changes in the natural environment and the implications for ecosystem
services.

S\~{a}o Paulo CZO seed site is the first investigation site self-called
a CZO in Brazil. Today, there are a few CZOs dedicated to studying
anthropogenic impacts on soils. However, the S\~{a}o Paulo CZO seed
site is the first observatory installed in a highly urbanized context,
in a subtropical climate. This puts this research in a position to
generate unique data for soil resources anthropogenic impacts. The seed
site has a long road ahead, still it is already enabling a better
understanding of the feedback \mbox{between}\unskip\break biotic/abiotic processes and
redox changes in organic-rich sediments. Nevertheless, other CZOs
dedicated to studying anthropogenic impacts on soils in different
climates can contribute to a better \mbox{understanding} of anthropogenic
impacts on this essential natural resource.

\section*{Acknowledgements}

The authors thank the S\~{a}o Paulo Research Foundation (FAPESP Grant
no.~2021-14808-6) for funding the project and Conselho Nacional de
Desenvolvimento Cient\'{i}fico e Tecnol\'{o}gico (CNPq 
grant no.~304695/2022-0) for productivity scholarship. This work has been
(partially) performed at USPMag lab at Instituto de Astronomia,
Geof\'{i}sica e Ci\^{e}ncias Atmosf\'{e}ricas (IAG) at Universidade de
S\~{a}o Paulo (USP) funded by CAPES/FAPESP/CNPQ.

\CDRGrant[FAPESP]{2021-14808-6}
\CDRGrant[CNPq]{304695/2022-0}

\section*{Declaration of interests}
The authors do not work for, advise, own shares in, or receive funds
from any organization that could \mbox{benefit} from this article, and
have declared no affiliations other than their research organizations.

\back{}

\printbibliography
\refinput{crgeos20241103-reference.tex}

\end{document}