This thesis describes a new method of deconvolving the bulk magnetic signal based on principal component analysis (PCA) of first-order reversal curve (FORC) diagrams. Magnetic records are crucial for the understanding of paleoclimate, and are a source of information on the geological history of Earth. The information extracted with established (conventional) methods may have limited fidelity due to alteration of iron-bearing minerals during diagenetic processes and the multiplicity of components contributing to the signal. FORC-PCA provides a reliable framework for extracting individual components, thus assessing the reliability of the record. We demonstrate the applicability of FORC-PCA in two case studies, supported by transmission electron microscopy (TEM) which links direct observations with characteristics derived from FORC-PCA.

In the first chapter an overview of conventionally used methods is given, introducing the problems related to unmixing bulk magnetic signals. The main challenge lies in the reliable and unambiguous identification of magnetic minerals through a bulk measurement.s. The major drawbacks of conventionally applied methods are their non-uniqueness and qualitative character.

The second chapter is dedicated to FORC diagrams and FORC-PCA, which has the potential to solve the outlined issues. FORCs are an advanced hysteresis measurement which provide the coercivity distribution of the whole population of magnetic grains in the sample; it is sensitive to mineralogy, domain state, grain size and magnetostatic interactions. This allows to characterise the bulk magnetic properties in a much more comprehensive way compared to established methods. Until recently, FORC diagrams were used in a qualitative way, providing fingerprints of known mineral phases. By applying PCA to suites of FORC diagrams, it is possible to deconvolve the signal and identify the constituent components, which may be impossible to distinguish in conventional methods.

The third chapter describes a case study of FORC-PCA applied to a set of cores sampled from the salt marshes of the Norfolk coast. The salt marshes offer diverse geochemical conditions in ponds located in close proximity to each other, representing vastly different diagenetic regimes, from ferruginous to sulphidic. The vertical distribution of magnetic minerals is characterised, linking their variation to geochemical gradients and ongoing diagenetic processes. The case study demonstrates the applicability of FORC-PCA on natural datasets, and highlights the importance of early diagenetic alteration of magnetic minerals in early stages of burial.

The fourth chapter describes a study of a marine core. % demonstrating the applications of FORC-PCA for environmental magnetic studies over geological timescales. A high-resolution record of climate change variation reflected in changes of magnetic mineral composition is obtained from the Iberian Margin, spanning 194,000 years over the past two glacial cycles. We identify three magnetic components, where the bulk susceptibility variations are controlled by a previously unreported magnetic mineral phase. Trabecular magnetite, named so for its characteristic skeletal internal structure, exhibits a stable low coercivity single-domain signature and is prevalent throughout the core. Formed through reductive dissolution of goethite, it preserves a record of climate change as a remarkably sensitive proxy to changes in diagenetic redox conditions in surface ocean sediments.

In the fifth chapter trabecular magnetite is characterised in an advanced TEM study covering imaging, electron tomography, holography, chemical and crystallographic characterisation. The results link the FORC-PCA-derived signature with direct observations, providing evidence that FORC-PCA can be used reliably as a tool for identifying and temporal tracing of magnetic minerals, even where conventional methods often fail.

The results and their implications are discussed in the sixth chapter, primarily highlighting the importance of early diagenetic processes in alteration of the magnetic record. FORC-PCA provides a way to reliably fingerprint diagenetic processes, emphasising that the preservation of the magnetic record is highly sensitive to changes in redox conditions triggered by climate change.