- Pre-registration nursing students
- No definition of master’s degree in nursing described in the publication
After the search, we collated and uploaded all the identified records into EndNote v.X8 (Clarivate Analytics, Philadelphia, Pennsylvania) and removed any duplicates. Two independent reviewers (MCS and SA) screened the titles and abstracts for assessment in line with the inclusion criteria. They retrieved and assessed the full texts of the selected studies while applying the inclusion criteria. Any disagreements about the eligibility of studies were resolved by discussion or, if no consensus could be reached, by involving experienced researchers (MZ-S and RP).
The first reviewer (MCS) extracted data from the selected publications. For this purpose, an extraction tool developed by the authors was used. This tool comprised the following criteria: author(s), year of publication, country, research question, design, case definition, data sources, and methodologic and data-analysis triangulation. First, we extracted and summarized information about the case study design. Second, we narratively summarized the way in which the data and methodological triangulation were described. Finally, we summarized the information on within-case or cross-case analysis. This process was performed using Microsoft Excel. One reviewer (MCS) extracted data, whereas another reviewer (SA) cross-checked the data extraction, making suggestions for additions or edits. Any disagreements between the reviewers were resolved through discussion.
A total of 149 records were identified in 2 databases. We removed 20 duplicates and screened 129 reports by title and abstract. A total of 46 reports were assessed for eligibility. Through hand searches, we identified 117 additional records. Of these, we excluded 98 reports after title and abstract screening. A total of 17 reports were assessed for eligibility. From the 2 databases and the hand search, 63 reports were assessed for eligibility. Ultimately, we included 8 articles for data extraction. No further articles were included after the reference list screening of the included studies. A PRISMA flow diagram of the study selection and inclusion process is presented in Figure 1 . As shown in Tables 2 and and3, 3 , the articles included in this scoping review were published between 2010 and 2022 in Canada (n = 3), the United States (n = 2), Australia (n = 2), and Scotland (n = 1).
PRISMA flow diagram.
Characteristics of Articles Included.
Author | Contandriopoulos et al | Flinter | Hogan et al | Hungerford et al | O’Rourke | Roots and MacDonald | Schadewaldt et al | Strachan et al |
---|---|---|---|---|---|---|---|---|
Country | Canada | The United States | The United States | Australia | Canada | Canada | Australia | Scotland |
How or why research question | No information on the research question | Several how or why research questions | What and how research question | No information on the research question | Several how or why research questions | No information on the research question | What research question | What and why research questions |
Design and referenced author of methodological guidance | Six qualitative case studies Robert K. Yin | Multiple-case studies design Robert K. Yin | Multiple-case studies design Robert E. Stake | Case study design Robert K. Yin | Qualitative single-case study Robert K. Yin Robert E. Stake Sharan Merriam | Single-case study design Robert K. Yin Sharan Merriam | Multiple-case studies design Robert K. Yin Robert E. Stake | Multiple-case studies design |
Case definition | Team of health professionals (Small group) | Nurse practitioners (Individuals) | Primary care practices (Organization) | Community-based NP model of practice (Organization) | NP-led practice (Organization) | Primary care practices (Organization) | No information on case definition | Health board (Organization) |
Overview of Within-Method, Between/Across-Method, and Data-Analysis Triangulation.
Author | Contandriopoulos et al | Flinter | Hogan et al | Hungerford et al | O’Rourke | Roots and MacDonald | Schadewaldt et al | Strachan et al |
---|---|---|---|---|---|---|---|---|
Within-method triangulation (using within-method triangulation use at least 2 data-collection procedures from the same design approach) | ||||||||
: | ||||||||
Interviews | X | x | x | x | x | |||
Observations | x | x | ||||||
Public documents | x | x | x | |||||
Electronic health records | x | |||||||
Between/across-method (using both qualitative and quantitative data-collection procedures in the same study) | ||||||||
: | ||||||||
: | ||||||||
Interviews | x | x | x | |||||
Observations | x | x | ||||||
Public documents | x | x | ||||||
Electronic health records | x | |||||||
: | ||||||||
Self-assessment | x | |||||||
Service records | x | |||||||
Questionnaires | x | |||||||
Data-analysis triangulation (combination of 2 or more methods of analyzing data) | ||||||||
: | ||||||||
: | ||||||||
Deductive | x | x | x | |||||
Inductive | x | x | ||||||
Thematic | x | x | ||||||
Content | ||||||||
: | ||||||||
Descriptive analysis | x | x | x | |||||
: | ||||||||
: | ||||||||
Deductive | x | x | x | x | ||||
Inductive | x | x | ||||||
Thematic | x | |||||||
Content | x |
The following sections describe the research question, case definition, and case study design. Case studies are most appropriate when asking “how” or “why” questions. 1 According to Yin, 1 how and why questions are explanatory and lead to the use of case studies, histories, and experiments as the preferred research methods. In 1 study from Canada, eg, the following research question was presented: “How and why did stakeholders participate in the system change process that led to the introduction of the first nurse practitioner-led Clinic in Ontario?” (p7) 19 Once the research question has been formulated, the case should be defined and, subsequently, the case study design chosen. 1 In typical case studies with mixed methods, the 2 types of data are gathered concurrently in a convergent design and the results merged to examine a case and/or compare multiple cases. 10
“How” or “why” questions were found in 4 studies. 16 , 17 , 19 , 22 Two studies additionally asked “what” questions. Three studies described an exploratory approach, and 1 study presented an explanatory approach. Of these 4 studies, 3 studies chose a qualitative approach 17 , 19 , 22 and 1 opted for mixed methods with a convergent design. 16
In the remaining studies, either the research questions were not clearly stated or no “how” or “why” questions were formulated. For example, “what” questions were found in 1 study. 21 No information was provided on exploratory, descriptive, and explanatory approaches. Schadewaldt et al 21 chose mixed methods with a convergent design.
A total of 5 studies defined the case as an organizational unit. 17 , 18 - 20 , 22 Of the 8 articles, 4 reported multiple-case studies. 16 , 17 , 22 , 23 Another 2 publications involved single-case studies. 19 , 20 Moreover, 2 publications did not state the case study design explicitly.
This section describes within-method triangulation, which involves employing at least 2 data-collection procedures within the same design approach. 6 , 7 This can also be called data source triangulation. 8 Next, we present the single data-collection procedures in detail. In 5 studies, information on within-method triangulation was found. 15 , 17 - 19 , 22 Studies describing a quantitative approach and the triangulation of 2 or more quantitative data-collection procedures could not be included in this scoping review.
Five studies used qualitative data-collection procedures. Two studies combined face-to-face interviews and documents. 15 , 19 One study mixed in-depth interviews with observations, 18 and 1 study combined face-to-face interviews and documentation. 22 One study contained face-to-face interviews, observations, and documentation. 17 The combination of different qualitative data-collection procedures was used to present the case context in an authentic and complex way, to elicit the perspectives of the participants, and to obtain a holistic description and explanation of the cases under study.
All 5 studies used qualitative interviews as the primary data-collection procedure. 15 , 17 - 19 , 22 Face-to-face, in-depth, and semi-structured interviews were conducted. The topics covered in the interviews included processes in the introduction of new care services and experiences of barriers and facilitators to collaborative work in general practices. Two studies did not specify the type of interviews conducted and did not report sample questions. 15 , 18
In 2 studies, qualitative observations were carried out. 17 , 18 During the observations, the physical design of the clinical patients’ rooms and office spaces was examined. 17 Hungerford et al 18 did not explain what information was collected during the observations. In both studies, the type of observation was not specified. Observations were generally recorded as field notes.
In 3 studies, various qualitative public documents were studied. 15 , 19 , 22 These documents included role description, education curriculum, governance frameworks, websites, and newspapers with information about the implementation of the role and general practice. Only 1 study failed to specify the type of document and the collected data. 15
In 1 study, qualitative documentation was investigated. 17 This included a review of dashboards (eg, provider productivity reports or provider quality dashboards in the electronic health record) and quality performance reports (eg, practice-wide or co-management team-wide performance reports).
This section describes the between/across methods, which involve employing both qualitative and quantitative data-collection procedures in the same study. 6 , 7 This procedure can also be denoted “methodologic triangulation.” 8 Subsequently, we present the individual data-collection procedures. In 3 studies, information on between/across triangulation was found. 16 , 20 , 21
Three studies used qualitative and quantitative data-collection procedures. One study combined face-to-face interviews, documentation, and self-assessments. 16 One study employed semi-structured interviews, direct observation, documents, and service records, 20 and another study combined face-to-face interviews, non-participant observation, documents, and questionnaires. 23
All 3 studies used qualitative interviews as the primary data-collection procedure. 16 , 20 , 23 Face-to-face and semi-structured interviews were conducted. In the interviews, data were collected on the introduction of new care services and experiences of barriers to and facilitators of collaborative work in general practices.
In 2 studies, direct and non-participant qualitative observations were conducted. 20 , 23 During the observations, the interaction between health professionals or the organization and the clinical context was observed. Observations were generally recorded as field notes.
In 2 studies, various qualitative public documents were examined. 20 , 23 These documents included role description, newspapers, websites, and practice documents (eg, flyers). In the documents, information on the role implementation and role description of NPs was collected.
In 1 study, qualitative individual journals were studied. 16 These included reflective journals from NPs, who performed the role in primary health care.
Only 1 study involved quantitative service records. 20 These service records were obtained from the primary care practices and the respective health authorities. They were collected before and after the implementation of an NP role to identify changes in patients’ access to health care, the volume of patients served, and patients’ use of acute care services.
In 2 studies, quantitative questionnaires were used to gather information about the teams’ satisfaction with collaboration. 16 , 21 In 1 study, 3 validated scales were used. The scales measured experience, satisfaction, and belief in the benefits of collaboration. 21 Psychometric performance indicators of these scales were provided. However, the time points of data collection were not specified; similarly, whether the questionnaires were completed online or by hand was not mentioned. A competency self-assessment tool was used in another study. 16 The assessment comprised 70 items and included topics such as health promotion, protection, disease prevention and treatment, the NP-patient relationship, the teaching-coaching function, the professional role, managing and negotiating health care delivery systems, monitoring and ensuring the quality of health care practice, and cultural competence. Psychometric performance indicators were provided. The assessment was completed online with 2 measurement time points (pre self-assessment and post self-assessment).
This section describes data-analysis triangulation, which involves the combination of 2 or more methods of analyzing data. 6 Subsequently, we present within-case analysis and cross-case analysis.
Three studies combined qualitative and quantitative methods of analysis. 16 , 20 , 21 Two studies involved deductive and inductive qualitative analysis, and qualitative data were analyzed thematically. 20 , 21 One used deductive qualitative analysis. 16 The method of analysis was not specified in the studies. Quantitative data were analyzed using descriptive statistics in 3 studies. 16 , 20 , 23 The descriptive statistics comprised the calculation of the mean, median, and frequencies.
Two studies combined deductive and inductive qualitative analysis, 19 , 22 and 2 studies only used deductive qualitative analysis. 15 , 18 Qualitative data were analyzed thematically in 1 study, 22 and data were treated with content analysis in the other. 19 The method of analysis was not specified in the 2 studies.
In 7 studies, a within-case analysis was performed. 15 - 20 , 22 Six studies used qualitative data for the within-case analysis, and 1 study employed qualitative and quantitative data. Data were analyzed separately, consecutively, or in parallel. The themes generated from qualitative data were compared and then summarized. The individual cases were presented mostly as a narrative description. Quantitative data were integrated into the qualitative description with tables and graphs. Qualitative and quantitative data were also presented as a narrative description.
Of the multiple-case studies, 5 carried out cross-case analyses. 15 - 17 , 20 , 22 Three studies described the cross-case analysis using qualitative data. Two studies reported a combination of qualitative and quantitative data for the cross-case analysis. In each multiple-case study, the individual cases were contrasted to identify the differences and similarities between the cases. One study did not specify whether a within-case or a cross-case analysis was conducted. 23
This section describes confirmation or contradiction through qualitative and quantitative data. 1 , 4 Qualitative and quantitative data were reported separately, with little connection between them. As a result, the conclusions on neither the comparisons nor the contradictions could be clearly determined.
In 3 studies, the consistency of the results of different types of qualitative data was highlighted. 16 , 19 , 21 In particular, documentation and interviews or interviews and observations were contrasted:
Both types of data showed that NPs and general practitioners wanted to have more time in common to discuss patient cases and engage in personal exchanges. 21 In addition, the qualitative and quantitative data confirmed the individual progression of NPs from less competent to more competent. 16 One study pointed out that qualitative and quantitative data obtained similar results for the cases. 20 For example, integrating NPs improved patient access by increasing appointment availability.
Although questionnaire results indicated that NPs and general practitioners experienced high levels of collaboration and satisfaction with the collaborative relationship, the qualitative results drew a more ambivalent picture of NPs’ and general practitioners’ experiences with collaboration. 21
The studies included in this scoping review evidenced various research questions. The recommended formats (ie, how or why questions) were not applied consistently. Therefore, no case study design should be applied because the research question is the major guide for determining the research design. 2 Furthermore, case definitions and designs were applied variably. The lack of standardization is reflected in differences in the reporting of these case studies. Generally, case study research is viewed as allowing much more freedom and flexibility. 5 , 24 However, this flexibility and the lack of uniform specifications lead to confusion.
Methodologic triangulation, as described in the literature, can be somewhat confusing as it can refer to either data-collection methods or research designs. 6 , 8 For example, methodologic triangulation can allude to qualitative and quantitative methods, indicating a paradigmatic connection. Methodologic triangulation can also point to qualitative and quantitative data-collection methods, analysis, and interpretation without specific philosophical stances. 6 , 8 Regarding “data-collection methods with no philosophical stances,” we would recommend using the wording “data source triangulation” instead. Thus, the demarcation between the method and the data-collection procedures will be clearer.
Yin 1 advocated the use of multiple sources of evidence so that a case or cases can be investigated more comprehensively and accurately. Most studies included multiple data-collection procedures. Five studies employed a variety of qualitative data-collection procedures, and 3 studies used qualitative and quantitative data-collection procedures (mixed methods). In contrast, no study contained 2 or more quantitative data-collection procedures. In particular, quantitative data-collection procedures—such as validated, reliable questionnaires, scales, or assessments—were not used exhaustively. The prerequisites for using multiple data-collection procedures are availability, the knowledge and skill of the researcher, and sufficient financial funds. 1 To meet these prerequisites, research teams consisting of members with different levels of training and experience are necessary. Multidisciplinary research teams need to be aware of the strengths and weaknesses of different data sources and collection procedures. 1
When using multiple data sources and analysis methods, it is necessary to present the results in a coherent manner. Although the importance of multiple data sources and analysis has been emphasized, 1 , 5 the description of triangulation has tended to be brief. Thus, traceability of the research process is not always ensured. The sparse description of the data-analysis triangulation procedure may be due to the limited number of words in publications or the complexity involved in merging the different data sources.
Only a few concrete recommendations regarding the operationalization of the data-analysis triangulation with the qualitative data process were found. 25 A total of 3 approaches have been proposed 25 : (1) the intuitive approach, in which researchers intuitively connect information from different data sources; (2) the procedural approach, in which each comparative or contrasting step in triangulation is documented to ensure transparency and replicability; and (3) the intersubjective approach, which necessitates a group of researchers agreeing on the steps in the triangulation process. For each case study, one of these 3 approaches needs to be selected, carefully carried out, and documented. Thus, in-depth examination of the data can take place. Farmer et al 25 concluded that most researchers take the intuitive approach; therefore, triangulation is not clearly articulated. This trend is also evident in our scoping review.
Few studies in this scoping review used a combination of qualitative and quantitative analysis. However, creating a comprehensive stand-alone picture of a case from both qualitative and quantitative methods is challenging. Findings derived from different data types may not automatically coalesce into a coherent whole. 4 O’Cathain et al 26 described 3 techniques for combining the results of qualitative and quantitative methods: (1) developing a triangulation protocol; (2) following a thread by selecting a theme from 1 component and following it across the other components; and (3) developing a mixed-methods matrix.
The most detailed description of the conducting of triangulation is the triangulation protocol. The triangulation protocol takes place at the interpretation stage of the research process. 26 This protocol was developed for multiple qualitative data but can also be applied to a combination of qualitative and quantitative data. 25 , 26 It is possible to determine agreement, partial agreement, “silence,” or dissonance between the results of qualitative and quantitative data. The protocol is intended to bring together the various themes from the qualitative and quantitative results and identify overarching meta-themes. 25 , 26
The “following a thread” technique is used in the analysis stage of the research process. To begin, each data source is analyzed to identify the most important themes that need further investigation. Subsequently, the research team selects 1 theme from 1 data source and follows it up in the other data source, thereby creating a thread. The individual steps of this technique are not specified. 26 , 27
A mixed-methods matrix is used at the end of the analysis. 26 All the data collected on a defined case are examined together in 1 large matrix, paying attention to cases rather than variables or themes. In a mixed-methods matrix (eg, a table), the rows represent the cases for which both qualitative and quantitative data exist. The columns show the findings for each case. This technique allows the research team to look for congruency, surprises, and paradoxes among the findings as well as patterns across multiple cases. In our review, we identified only one of these 3 approaches in the study by Roots and MacDonald. 20 These authors mentioned that a causal network analysis was performed using a matrix. However, no further details were given, and reference was made to a later publication. We could not find this publication.
Because it focused on the implementation of NPs in primary health care, the setting of this scoping review was narrow. However, triangulation is essential for research in this area. This type of research was found to provide a good basis for understanding methodologic and data-analysis triangulation. Despite the lack of traceability in the description of the data and methodological triangulation, we believe that case studies are an appropriate design for exploring new nursing roles in existing health care systems. This is evidenced by the fact that case study research is widely used in many social science disciplines as well as in professional practice. 1 To strengthen this research method and increase the traceability in the research process, we recommend using the reporting guideline and reporting checklist by Rodgers et al. 9 This reporting checklist needs to be complemented with methodologic and data-analysis triangulation. A procedural approach needs to be followed in which each comparative step of the triangulation is documented. 25 A triangulation protocol or a mixed-methods matrix can be used for this purpose. 26 If there is a word limit in a publication, the triangulation protocol or mixed-methods matrix needs to be identified. A schematic representation of methodologic and data-analysis triangulation in case studies can be found in Figure 2 .
Schematic representation of methodologic and data-analysis triangulation in case studies (own work).
This study suffered from several limitations that must be acknowledged. Given the nature of scoping reviews, we did not analyze the evidence reported in the studies. However, 2 reviewers independently reviewed all the full-text reports with respect to the inclusion criteria. The focus on the primary care setting with NPs (master’s degree) was very narrow, and only a few studies qualified. Thus, possible important methodological aspects that would have contributed to answering the questions were omitted. Studies describing the triangulation of 2 or more quantitative data-collection procedures could not be included in this scoping review due to the inclusion and exclusion criteria.
Given the various processes described for methodologic and data-analysis triangulation, we can conclude that triangulation in case studies is poorly standardized. Consequently, the traceability of the research process is not always given. Triangulation is complicated by the confusion of terminology. To advance case study research in nursing, we encourage authors to reflect critically on methodologic and data-analysis triangulation and use existing tools, such as the triangulation protocol or mixed-methods matrix and the reporting guideline checklist by Rodgers et al, 9 to ensure more transparent reporting.
Acknowledgments.
The authors thank Simona Aeschlimann for her support during the screening process.
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material: Supplemental material for this article is available online.
Saul McLeod, PhD
Editor-in-Chief for Simply Psychology
BSc (Hons) Psychology, MRes, PhD, University of Manchester
Saul McLeod, PhD., is a qualified psychology teacher with over 18 years of experience in further and higher education. He has been published in peer-reviewed journals, including the Journal of Clinical Psychology.
Learn about our Editorial Process
Olivia Guy-Evans, MSc
Associate Editor for Simply Psychology
BSc (Hons) Psychology, MSc Psychology of Education
Olivia Guy-Evans is a writer and associate editor for Simply Psychology. She has previously worked in healthcare and educational sectors.
On This Page:
Case studies are in-depth investigations of a person, group, event, or community. Typically, data is gathered from various sources using several methods (e.g., observations & interviews).
The case study research method originated in clinical medicine (the case history, i.e., the patient’s personal history). In psychology, case studies are often confined to the study of a particular individual.
The information is mainly biographical and relates to events in the individual’s past (i.e., retrospective), as well as to significant events that are currently occurring in his or her everyday life.
The case study is not a research method, but researchers select methods of data collection and analysis that will generate material suitable for case studies.
Freud (1909a, 1909b) conducted very detailed investigations into the private lives of his patients in an attempt to both understand and help them overcome their illnesses.
This makes it clear that the case study is a method that should only be used by a psychologist, therapist, or psychiatrist, i.e., someone with a professional qualification.
There is an ethical issue of competence. Only someone qualified to diagnose and treat a person can conduct a formal case study relating to atypical (i.e., abnormal) behavior or atypical development.
There are several places to find data for a case study. The key is to gather data from multiple sources to get a complete picture of the case and corroborate facts or findings through triangulation of evidence. Most of this information is likely qualitative (i.e., verbal description rather than measurement), but the psychologist might also collect numerical data.
Searching historical archives, museum collections and databases to find relevant documents, visual/audio records related to the case history and context.
Public archives like newspapers, organizational records, photographic collections could all include potentially relevant pieces of information to shed light on attitudes, cultural perspectives, common practices and historical contexts related to psychology.
Organizational records offer the advantage of often having large datasets collected over time that can reveal or confirm psychological insights.
Of course, privacy and ethical concerns regarding confidential data must be navigated carefully.
However, with proper protocols, organizational records can provide invaluable context and empirical depth to qualitative case studies exploring the intersection of psychology and organizations.
Follow specified case study guidelines provided by a journal or your psychology tutor. General components of clinical case studies include: background, symptoms, assessments, diagnosis, treatment, and outcomes. Interpreting the information means the researcher decides what to include or leave out. A good case study should always clarify which information is the factual description and which is an inference or the researcher’s opinion.
Case studies allow a researcher to investigate a topic in far more detail than might be possible if they were trying to deal with a large number of research participants (nomothetic approach) with the aim of ‘averaging’.
Because of their in-depth, multi-sided approach, case studies often shed light on aspects of human thinking and behavior that would be unethical or impractical to study in other ways.
Research that only looks into the measurable aspects of human behavior is not likely to give us insights into the subjective dimension of experience, which is important to psychoanalytic and humanistic psychologists.
Case studies are often used in exploratory research. They can help us generate new ideas (that might be tested by other methods). They are an important way of illustrating theories and can help show how different aspects of a person’s life are related to each other.
The method is, therefore, important for psychologists who adopt a holistic point of view (i.e., humanistic psychologists ).
Because a case study deals with only one person/event/group, we can never be sure if the case study investigated is representative of the wider body of “similar” instances. This means the conclusions drawn from a particular case may not be transferable to other settings.
Because case studies are based on the analysis of qualitative (i.e., descriptive) data , a lot depends on the psychologist’s interpretation of the information she has acquired.
This means that there is a lot of scope for Anna O , and it could be that the subjective opinions of the psychologist intrude in the assessment of what the data means.
For example, Freud has been criticized for producing case studies in which the information was sometimes distorted to fit particular behavioral theories (e.g., Little Hans ).
This is also true of Money’s interpretation of the Bruce/Brenda case study (Diamond, 1997) when he ignored evidence that went against his theory.
Breuer, J., & Freud, S. (1895). Studies on hysteria . Standard Edition 2: London.
Curtiss, S. (1981). Genie: The case of a modern wild child .
Diamond, M., & Sigmundson, K. (1997). Sex Reassignment at Birth: Long-term Review and Clinical Implications. Archives of Pediatrics & Adolescent Medicine , 151(3), 298-304
Freud, S. (1909a). Analysis of a phobia of a five year old boy. In The Pelican Freud Library (1977), Vol 8, Case Histories 1, pages 169-306
Freud, S. (1909b). Bemerkungen über einen Fall von Zwangsneurose (Der “Rattenmann”). Jb. psychoanal. psychopathol. Forsch ., I, p. 357-421; GW, VII, p. 379-463; Notes upon a case of obsessional neurosis, SE , 10: 151-318.
Harlow J. M. (1848). Passage of an iron rod through the head. Boston Medical and Surgical Journal, 39 , 389–393.
Harlow, J. M. (1868). Recovery from the Passage of an Iron Bar through the Head . Publications of the Massachusetts Medical Society. 2 (3), 327-347.
Money, J., & Ehrhardt, A. A. (1972). Man & Woman, Boy & Girl : The Differentiation and Dimorphism of Gender Identity from Conception to Maturity. Baltimore, Maryland: Johns Hopkins University Press.
Money, J., & Tucker, P. (1975). Sexual signatures: On being a man or a woman.
Political Science
05 July - 09 July 2021
Oslo Summer School for Social Sciences
Institution:
University of Oslo
Professor Andrew Bennett, Department of Government, Georgetown University, Washington DC., USA
PhD students, graduate or postgraduate student or professional from any social science field of study or related disciplines.
The central goal of the seminar is to enable students to create and critique methodologically sophisticated case study research designs in the social sciences. To do so, the seminar will explore the techniques, uses, strengths, and limitations of case study methods, while emphasizing the relationships among these methods, alternative methods, and contemporary debates in the philosophy of science. The research examples used to illustrate methodological issues will be drawn primarily from international relations and comparative politics. The methodological content of the course is also applicable, however, to the study of history, sociology, education, business, economics, and other social and behavioral sciences.
4000 NOK, The participation fee includes: Covering of expenses towards administration and honorarium for lecturers.
Pisa, Italy
20 May - 24 May 2025
Brussels, Belgium
05 February - 14 February 2025
Berlin, Germany
02 January - 24 January 2025
If you don’t want to miss out on new summer school courses, subscribe to our monthly newsletter.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Scientific Reports volume 14 , Article number: 20766 ( 2024 ) Cite this article
Metrics details
Rare earth elements (REEs) exhibit diagnostic absorption features in the visible-near infrared region, enabling their detection and identification via spectroscopic methods. Satellite-based remote sensing mapping of REEs, however, has not been attainable so far due to the necessity for high-quality hyperspectral data to resolve their narrow absorption features. This research leverages EnMAP hyperspectral satellite data to map REEs in Mountain Pass, California—a mining area known to host bastnaesite-Ce ore in sövite and beforsite carbonatites. By employing a polynomial fitting technique to characterize the diagnostic absorption features of Neodymium (Nd) at ∼ 740 and ∼ 800 nm, the surface occurrence of Nd was successfully mapped at a 30m pixel resolution. The relative abundance of Nd was represented using the continuum-removed area of the 800 nm feature. The resulting map, highlighting hundreds of anomalous pixels, was validated through laboratory spectroscopy, surface geology, and high-resolution satellite imagery. This study marks a major advancement in REE exploration, demonstrating for the first time, the possibility of directly detecting Nd in geologic environments using the EnMAP hyperspectral satellite data. This capability can offer a fast and cost-effective method for screening Earth’s surfaces for REE signature, complementing the existing exploration portfolio and facilitating the discovery of new resources.
Introduction.
Rare Earth Elements (REEs) are essential to many modern technologies, including electric vehicles, wind turbines, and smartphones 1 , 2 , 3 . This broad range of applications, combined with increasing demands and disruptions in the supply chain, has transformed REEs into strategic commodities and critical raw materials 4 . REEs are a group of metallic elements with similar chemical properties comprising the lanthanides (atomic number 57 to 71) plus Y (39), commonly divided into light (LREE) and heavy (HREE) subgroups, comprising La to Eu and Gd to Lu + Y, respectively 5 .
REEs naturally occur in a diverse array of minerals, including carbonates, phosphates, silicates, and oxides of which carbonates and phosphates constitute the most abundant and economically valuable minerals 2 . The REE carbonates include the fluorocarbonate minerals bastnaesite, synchysite, and parisite 6 . The RE 2 O 3 content of these minerals is exceptionally high, reaching up to 75 wt.% in bastnaesite 7 . LREEs typically concentrate in carbonates (i.e., bastnaesite; (Ce,La)(CO 3 )F) and phosphates (i.e., monazite; (Ce,La,Nd,Th)PO 4 ), whereas HREEs are commonly hosted by oxides and, partly, by phosphates, including xenotime ((HREE,Y)PO 4 ). Due to chemical similarity (ionic radii and oxidation states), REEs often substitute for one another and co-occur within the same mineral species 5 .
Contrary to their name, REEs are relatively abundant in the Earth's crust, though economically viable deposits are uncommon. Several deposit classes are recognized to host REEs 2 , with carbonatites being the predominant sources, accounting for more than 70% of global REO (Rare Earth Oxides) production. Two notable examples of such deposits are the Bayan Obo mine in China and the Mountain Pass in the US 8 . Carbonatites are defined as rocks with > 50% primary magmatic carbonates. Geologically, they occur in continental settings and based on their mineralogy and petrographic texture are divided into three distinct classes: calcitic (also referred to as sövite), dolomitic (beforsite), and ankeritic (ferrocarbonatite) 2 , 5 . Carbonatites predominantly host LREEs such as La, Ce, Pr, and Nd, with bastnaesite being the primary mineral exploited in many related deposits 5 , 7 .
The technique of reflectance spectroscopy has recently emerged as a fast and cost-effective analytic tool for detecting and quantifying REEs. Several REEs, including Nd 3+ , Pr 3+ , Sm 3+ , Dy 3+ , Er 3+ , Ho 3+ , and potentially Eu 3+ and Tm 3+ exhibit diagnostic absorption features in the visible-near-infrared (VNIR; 400–1000 nm) and partly in the shortwave-infrared (SWIR; 1000–2500 nm) wavelengths, allowing them to be detected via spectroscopic methods 6 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 . The narrow absorption bands of REEs observed in the VNIR, as exemplified in Fig. 1 , are attributed to 4f-4f intra-configurational electron transitions 6 , 11 , 15 .
Spectral signature of the rare-earth mineral bastnaesite in the VNIR–SWIR range. The spectrum of monazite is shown for comparison—data sourced from the USGS spectral library 21 . The key absorption features of Nd in the VNIR range are bolded. The inset graph provides a closer view of bastnaesite’s absorbing bands between 700 and 910 nm (marked by the solid bar), covering three diagnostic absorption features. The gray columns represent the spectral ranges used for polynomial fitting and remote sensing mapping of Nd in the study area.
Despite the effective shielding of the 4f orbitals by the 5s and 5p closed shells, the corresponding energy levels in the mineralogic phases are not fixed and rather undergo subtle changes, depending on the ligand type, coordination number, and polyhedron asymmetry. This variability leads to shifts in the position of absorbing bands within mineral phases, typically on the order of ∼ 10 nm in the VNIR range 6 , 15 , 17 . In other words, while the absorbing bands arise from REE ions, the host mineralogy plays an important role in determining the exact position of the absorption features and their intensities. The spectral behaviors of rare-earth minerals are already cataloged in several specialized spectral libraries 6 , 18 , 19 , 20 , 21 .
A growing number of studies have shown that Nd is the most spectrally active and readily detectable REE via spectroscopic methods 9 , 12 , 14 , 22 . The identification of Nd typically relies on the characterization of its most prominent and defining absorption features at ∼ 580, ∼ 740, ∼ 800, and ∼ 865 nm (see Fig. 1 ). By leveraging these distinctive features, the hyperspectral imaging technology has been able to detect Nd across various scales and conditions, spanning from close-range scanning of thin sections 23 and hand specimens in laboratory settings 6 , 24 to the mapping of vertical outcrops on the ground 25 , 26 and open-pit mines from airborne platforms 27 , 28 . More recently, Unmanned Aerial Vehicle (UAV)-based imaging systems have been employed to map REE-rich veins and outcrops at very high spatial resolutions 29 .
In contrast, direct detection of REEs by spaceborne satellite systems, such as ASTER and WorldView-3 multispectral instruments, has been unachievable, mainly due to the coarse spectral resolutions of multispectral datasets, rendering them unable to resolve the sharp yet narrow absorption features of REEs (Fig. 1 ), regardless of their spatial resolution 28 , 30 . While previous laboratory-based spectral simulations have demonstrated the potential of hyperspectral instruments, including the EnMAP satellite system, for direct REE detection 9 , the capability of the corresponding dataset has remained untested in real-world conditions.
This paper aims to bridge this gap and pave the way for further research by analyzing the EnMAP imaging spectroscopic data collected over the Mountain Pass REE mine in California, USA. Our study aims to prove the concept and recognize the potentials and limitations of spaceborne hyperspectral datasets for the direct detection and mapping of REEs. This is accomplished by studying the well-exposed, high-grade REE mine of the Mountain Pass area using the EnMAP imaging data at 30 m spatial resolution. Furthermore, the study seeks to evaluate the effectiveness of spectroscopic-based processing methods for REE detection aiming to provide a reliable site-independent mapping technique applicable to EnMAP and analogous hyperspectral remote sensing datasets.
Mountain Pass is located in southeastern California, approximately 65 km southwest of Las Vegas, in the Mojave Desert. Geologically, the area comprises a collection of Mesoproterozoic alkaline silicate intrusions (ca. 1.41 Ga) ranging in composition from mafic (shonkinite) through syenite to alkali granite. This suite is associated with a series of contemporaneous carbonatite dikes and intrusions 31 . The northern and eastern parts of the area consist of Proterozoic schists and gneisses, granitoids, and minor carbonatite intrusions. In contrast, the south and southeast are characterized by Paleozoic limestone, dolostone, and sandstone intruded by Jurassic granitic rocks and Cretaceous granodiorites 28 . The central and western parts are covered by folded, thrust-faulted Paleozoic carbonate and quartzose rocks 28 . A more detailed description of the area’s geology can be found in Castor 32 , Mars 30 , Mariano and Mariano 33 , and Watts, et al. 31 .
The carbonatites and the associated alkaline plutons constitute a suite of roughly tabular to lenticular, moderately west-dipping intrusions trending north-northwest within the ultrapotassic intrusive rocks 32 . The largest body, known as the Sulphide Queen carbonatite, is located in the center of the area, measuring 700 m in width and up to 150 m in thickness 10 and hosting the largest REE deposit in the US 7 , 31 . The Sulphide Queen carbonatite primarily consists of bastnaesite-barite sövite (calcitic) and bastnaesite-barite-dolomite (beforsite), or a mixture of both (dolomitic sövite), with the dolomitic carbonatite being more prevalent 32 .
Although the size of the Sulphide Queen carbonatite is modest, the orebody is highly enriched in LREE. The ore, which is recognized to be of igneous origin, typically contains 10–15% bastnaesite-Ce, 65% calcite/dolomite, and 20–25% barite. The bastnaesite mineral crystals are coarse-grained, typically measuring 300 µm in diameter with an average REE composition of 45.50% Ce, 15.82% Nd, 4.65% Pr, and 1.83% Sm, with lower quantities of Eu, Gd, Dy, Ho, and Er 32 . Other LREE-bearing accessory minerals are parisite, synchysite, monazite, and, less often, allanite 33 . Mining activities in the Sulphide Queen stock started in 1952 and ceased by 2002, leaving a reserve of > 20 million metric tons of ore at an average grade of 8.9% REO in place 34 . By 2007, the extraction of selected REE commodities from stockpiles resumed, and since 2018, the mine has been reactivated in response to the increased demand for REEs and geopolitical forces 31 .
Besides the main carbonatite body, there are numerous steeply inclined carbonatite dikes in the area, with the majority occurring in the vicinity of the Sulphide Queen orebody. Several of these dikes, particularly those adjacent to the mine, are known to contain bastnaesite, although most have low REO contents. A fenitized zone approximately 4 km southeast of the mine is also reported to host REE prospects containing allanite and bastnaesite minerals 32 .
The target area has been the focus of several remote sensing studies, primarily aimed at lithologic mapping using different multi- and hyperspectral datasets 28 , 30 , 35 . Based on field observations, vegetation covers between 10 to 30% of the surface, making it suitable for remote sensing studies.
The EnMAP data successfully resolved the REE- and carbonate-related features in the VNIR and SWIR ranges (Fig. 2 ). In the VNIR, it identified four diagnostic absorption features at ∼ 580, ∼ 740, ∼ 800, and ∼ 870 nm (Fig. 2 a). Within the SWIR range, it detected a deep carbonate feature at 2335 nm and two characteristics features related to bastnaesite at 2255 and 2316 nm (Fig. 2 b). The contained carbonate was identified as calcium carbonate, distinguished by a pronounced absorption feature at 2335 nm and the absence of a ferrous iron feature in the VNIR.
Continuum-removed reflectance spectra from EnMAP (in black) over the Sulphide Queen Mine compared to the laboratory-based spectrum of bastnaesite-rich ore from the mine site. The spectra are plotted in the VNIR ( a ) and SWIR ( b ) spectral ranges in native resolution. The laboratory spectrum, acquired using an ASD spectrometer, is sourced from the datasets published by Neave, et al. 9 . The vertical gray column in ( a ) highlights the EnMAP band affected by oxygen’s residual absorption feature at 760 nm. The minimum wavelengths were calculated by the polynomial fitting technique described in section " Processing methodology ". Note that the two graphs have different Y-axis scales.
The representative pixel spectrum over the open-pit mine shows good agreement with the laboratory spectroscopy of the orebody. Both datasets exhibit a comparable spectral pattern with the same number of absorption features and intensities across the VNIR and SWIR ranges, following continuum removal (Fig. 2 ). Notably, the positions of absorption minimums for the 580, 740, and 800 nm features are nearly identical in both datasets and for the characteristic features of bastnaesite, the difference is in the order of 1–2 nm (2255 vs. 2254 nm and 2216 vs. 2218 nm). However, the minimum wavelength differences for the 870 and 2330 nm features are significant. In the EnMAP data, the Nd feature occurs at a slightly longer wavelength (871 vs. 865 nm), and the carbonate feature appears at a shorter wavelength (2335 vs. 2342 nm) (Fig. 2 ). The latter is likely due to the spectral mixture of calcic carbonate with bastnaesite absorption features at the EnMAP ground sampling distance of 30 m. It is noteworthy that the bastnaesite spectrum depicted in Fig. 1 exhibits a different pattern compared to the laboratory plot in Fig. 2 b. The features at 2254, 2318, and 2342 nm appear respectively at 2249, 2314, and 2327 nm in Fig. 1 , possibly due to the complex/mixed mineralogy of the sample from the Mountain Pass. The 2255 nm feature observed in Fig. 2 b is speculated to arise from the hydroxyl bond in bastnaesite 6 .
Further distinctions include variations in the width of absorption features, which tend to be broader in the EnMAP data. Additionally, the right side of the 740 nm absorption feature in the EnMAP data is affected by a widespread residual O 2 absorption feature (Fig. 2 a). EnMAP also resolves an additional feature at 2200 nm, likely linked to clay minerals (Fig. 2 b). It is worth noting that the laboratory spectrum of this study closely resembles the spectral plot (published in Mars 30 i.e., Spectrum A in Fig. 7).
The distribution and relative abundance of Nd across the Mountain Pass area is illustrated in Fig. 3 a. Here, the spectral signature of Nd was mapped not only over the open-pit mine but also over the stockpiles, tailings storages, evaporation ponds, the crusher site, and the concentrator facilities (Fig. 3 b). The anomalies detected over the concentrator facility are probably the result of REE-bearing dust being transported westward from the mine crusher by the prevailing wind direction in the Mojave Desert. The Nd signature was also detected in several localities beyond the mining site, including at the edge of the Colosseum mine northward (Fig. 3 c) and over carbonate rocks in the west and southwest of the study area (Fig. 3 d–f). However, unlike the anomalies observed in the mining area, which form clusters of connected pixels, the peripheral anomalies are generally limited to a few pixels. In total, 740 pixels encompassing an area of 880,000 m 2 were identified to exhibit Nd features. The most prominent absorption feature was observed over the evaporation ponds, while the faintest was detected above the tailing storages (Fig. 3 b). The mapping method detected no anomalies over the fenitized zone southeast of the mining area (Fig. 3 a).
The spatial distribution and relative abundance of REEs in the Mountain Pass area, California. ( a ) Nd anomaly map (blue-red) yielded from spectral analysis of EnMAP hyperspectral data overlaid on enhanced albedo imagery. The area of the 800 nm absorption feature is used to indicate the relative abundance of Nd in the mapped pixels. The relative abundance of iron oxide and carbonate minerals are depicted in the background by orange and purple-red colors, respectively. ( b – f ) The same Nd anomalies from ( a ) overlaid on high-resolution satellite imagery of the area available on Google Earth. The data are from 29 th March 2021 at a ground sampling distance of ∼ 1 m. White rectangles in ( a ) define the outline of the images shown in ( b ) to ( f ). Major faults are shown by solid/dashed black lines.
In Fig. 3 a, the relative abundance of iron oxide minerals (i.e., hematite and goethite) and carbonates (i.e., calcite and dolomite) are depicted in orange and purple-red colors, respectively. Iron oxides are predominantly found in the NW to SE of the area, whereas carbonates are more abundant westward.
The statistical relationships between different spectral parameters within the mapped pixels are summarized in the scatterplots of Fig. 4 . During the spectral processing, it was noted that the minimum wavelengths of the 740 and 800 nm features vary within the ranges of 735–755 and 793–805 nm, respectively. These features exhibit a strong correlation in terms of absorption depth (R 2 = 0.88; Fig. 4 a), with the 800 nm feature appearing to be slightly deeper (see also Fig. 2 a). The interfering effect of residual O 2 absorption (Fig. 2 a) seems to be largely mitigated after excluding the corresponding band from the calculations.
Scatterplots of the spectral parameters of the Nd-bearing pixels derived from EnMAP data over the Mountain Pass area. ( a ) plot of the absorption depth at ∼ 740 nm (740D) against 800D. ( b ) plot of the absorption area at ∼ 740 nm (740A) against 580A. ( c ) plot of the absorption depth at ∼ 740 nm (740D) against 865D. ( d ) Plot of the minimum wavelength of the carbonate absorption feature against its depth for the pixels containing REE absorption features. The plotted data corresponds to the Nd anomalies mapped in Fig. 3 a. The solid red and dashed gray lines depict the best-fitted line to the data and the 1-to-1 line, respectively.
In contrast, the less prominent Nd feature at ∼ 580 nm (Fig. 2 a), while visually discernible in several Nd-bearing pixels, was found unsuitable for Nd mapping. This is primarily due to significant interferences from other scene components comprising green vegetation, causing noticeable shifts in the feature’s minimum wavelength making it difficult to track the via processing method. Nevertheless, the area of this feature correlates well with the area of the 740 nm feature (R 2 = 0.74; Fig. 4 b) and the 800 nm feature (R 2 = 0.66; not shown). The feature at ∼ 865 nm, although noticeable in some pixels over the orebody (Fig. 2 a), was not well-developed and therefore not resolvable in the EnMAP data. Statistically, it shows a weak correlation (R 2 = 0.36) with the depths of the absorption features at ∼ 740 and ∼ 800 nm (Fig. 4 c).
Figure 4 d depicts the plot of carbonate minimum wavelength against its depth for the pixels mapped in Fig. 3 a. In this plot, pixels from over the mining area and orebody exhibit wavelengths ranging from 2335 to 2350 nm and a relatively shallow carbonate absorption, typical of bastnaesite-rich calcic carbonatite. Pixels with similar absorption depths but shorter wavelength ranges (2310 to 2330 nm) were interpreted to arise from REE-bearing dolomitic carbonatite. The third cluster in Fig. 4 d represents isolated pixels mapped at the periphery of the mining area over carbonate rocks (highlighted in Fig. 3 d–f). These pixels are characterized by very shallow features at ∼ 740 and ∼ 800 nm but a deeper carbonate feature at wavelengths ranges between 2320 to 2340 nm. Verifying the presence of REEs/Nd in these pixels would indeed require ground truthing.
For reliable detection of REEs using spectral remote sensing data, it is essential to resolve multiple absorption features within the dataset. While some studies have successfully used three and occasionally four of the diagnostic absorption features of Nd 23 , 24 , 36 , many others have shown that not all the distinctive absorption features in the VNIR range, particularly those at ∼ 580 and ∼ 870 nm (refer to Figs. 1 and 2 a), are consistently present and resolvable in spectral data, even under optimal laboratory conditions 12 , 14 , 18 , 22 , 37 . Consequently, it is not surprising that the EnMAP data can only resolve the most prominent absorption features of Nd at ∼ 740 and ∼ 800 nm. This is consistent with the results of other remote sensing studies conducted to map REEs under open-air conditions using a UAV platform 29 . Conversely, relying solely on a single absorption feature can introduce large uncertainty in Nd detection 12 .
As demonstrated in this study, the minimum wavelength of the absorption features is as important and informative as the feature depth for REE detection. However, the minimum wavelengths of the absorption features are highly variable in spectral data. In laboratory studies, the minimum wavelengths of the 580, 740, and 800 nm features have been reported to vary from 575 to 590 nm, 740 to 747 nm, and 799 to 805 nm, respectively 9 . The variations retrieved from the EnMAP data, however, cover a wider range varying from 581 to 597 nm, 735 to 755 nm, and 793 to 805 nm, respectively. This wide range could be attributed to various factors, including intrinsic variations in the minimum wavelength of bastnaesite (typically on the order of ∼ 10 nm, as stated in the introduction), the co-occurrences of other REE-bearing minerals such as parisite, synchysite, and monazite inside the pixel footprint, the intimate/areal mixture of rare-earth minerals with other lithologic/background constituents (see below), the uncertainty of the retrieval method, and above all, limitations in the spectral sampling interval of EnMAP (i.e., 6.5 nm) compared to laboratory data.
It is important to note that each of these absorption features results from the superposition of several absorbing bands. For instance, the pronounced absorption feature at ∼ 740 nm is the result of at least six narrow absorbing bands centered at 733, 738, 741, 749, 755, and 762 nm (see the inset plot in Fig. 1 ), of which only four (i.e., at 734, 741, 747, and 757 nm) are discernible in the laboratory data of Fig. 2 a. A thorough analysis of these features can help characterize the mineralogical state of REEs and potentially unravel the presence of other REEs beyond Nd in spectral data.
In general, the ability to detect REEs spectrally could be affected by the following factors:
The overall albedo of the target and the contrast of the REE host with its background constituents . High proportions of opaque minerals such as magnetite (and allanite in non-carbonatite deposits) have been observed to dampen the spectral signal, contributing to low reflectance levels from the samples/surfaces and thus difficulty in REE detection 9 , 14 . In contrast, brighter backgrounds, exemplified here by the dominance of calcic carbonatite, can facilitate the detection of REEs.
The relative proportion of ferric (Fe 3 + ) iron minerals . The broad and intense absorption features of iron oxide minerals (i.e., hematite and goethite) in the VNIR region are reported to suppress the REE features significantly 10 , 12 , 22 , 25 . Simulated experiments have shown that even 1 wt.% of iron oxides can attenuate REE-related features, with the 580 and 870 nm features being particularly susceptible to suppression. In the range of 2 to 5 wt.%, iron minerals can readily dampen the features arising from 0.5 wt.% Nd, and at the 10 wt.% level, the REE features disappear entirely due to the dominance of ferric iron absorptions in the VNIR range 13 , 22 . As a general rule, the two weaker absorptions at ∼ 580 and ∼ 870 nm are more vulnerable and often go undetected in many spectral measurements (Todd Hoefen, personal communication). In the Mountain Pass area, although iron oxides are scarce over the open-pit mine, they are prevalent in the surrounding area, particularly over the alkaline intrusions eastward of the major fault lines (Fig. 3 a), contributing to the suppression of potential Nd features.
The fraction of vegetation cover . The presence of the green peak and chlorophyll absorption, respectively at ∼ 550 and ∼ 590 nm can undermine the REE feature at 580 nm. Presumably, the interference from vegetation in this area has impeded the mapping of the 580 nm feature in the EnMAP data, despite its existence and reasonable correlation with the 740 nm feature (Fig. 4 b). This is supported by the observation that pixels with the highest incidence of false-positives when using only the 580 nm feature for Nd mapping, are spatially associated with the highest Normalized Difference Vegetation Index (NDVI) values calculated from the same data. The interference from chlorophyll absorption may also explain the shift in the minimum wavelength of the 580 nm feature towards longer wavelengths (581 to 597 nm in EnMAP vs 575 to 590 nm in laboratory data). Further studies are required to understand the sensitivity of REE features to vegetation coverage/fraction.
The atmospheric correction effects . As illustrated in Fig. 2 a, the distinct O 2 -related absorption at 760 nm can interfere with the 740 nm feature of Nd. When the 740 nm feature surpasses the residual O 2 absorption, excluding the corresponding band from calculations, as demonstrated in this study, offers a simple yet effective solution to the problem. However, in situations where the feature is weakly developed and oxygen’s residual absorption predominates, excluding the band may not resolve the issue and could potentially lead to miscalculations of the spectral parameters, affecting the Nd mapping results. In contrast to O 2 , the residual water vapor effect appears as noise beyond 890 nm suppressing the 870 nm feature of Nd. While it is likely that the 870 nm feature may not be well-developed in the first place, the impact of water vapor residuals in weakening this feature within the EnMAP data needs to be considered. A more robust atmospheric correction procedure could certainly lead to better retrieval of REE signatures from EnMAP data.
The grain size effect. The size of REE-bearing grains is another factor affecting the intensity of Nd absorption features and, consequently, its detectability. Larger grain sizes absorb more light, leading to deeper absorption features 9 . In the Mountain Pass area, the relatively large bastnaesite grains, with an average diameter of 300 μm 32 , could be the reason behind the increased depth and width of absorption features in the EnMAP data (Fig. 2 a). However, variability in Nd grade and the scale effect (30 m image pixel vs point-scale ASD data) may have also played a role in this behavior.
The proportion of Nd (and total REEs) . Since the intensity of absorption features is proportional to the concentration of Nd in a sample/pixel, a higher concentration results in more pronounced absorption features, thereby facilitating spectral detection 9 , 12 , 14 , 17 . Based on this premise, while the exceptionally high concentration of Nd in the Mountain Pass area appears to have facilitated the remote sensing mapping, it is noteworthy that Nd was also detected over the tailings and waste storage sites (Fig. 3 b), indicating the detectability of lower grades of Nd via EnMAP data. In contrast, EnMAP was unsuccessful in mapping any Nd signatures over the fenitized zone and the adjacent areas (encircled in Fig. 3 a). This can be attributed to the small size of the carbonatite veins in this zone, the low content of REEs (Nd), as reported by Castor 32 , and the prevalence of iron oxides (see Fig. 3 a). Similarly, no carbonate signatures were detected over these veins using EnMAP's SWIR bands.
It's important to note that the depth of Nd's absorption features is reportedly influenced by the Nd to ΣREE (total REE) proportion, with higher ratios resulting in more pronounced absorption features 29 . The smallest REE-bearing target detectable at the 30 m pixel size of EnMAP, as well as the lowest level of Nd detectable spectrally (corresponding to the detection limit of EnMAP data), is currently unknown and should be addressed in future studies considering the noted factors. However, since reflectance spectroscopy has demonstrated a relatively low detection limit for Nd, ranging from 1000 to < 200 ppm 9 , 10 , 12 , 29 , it can be expected that under optimal environmental conditions, the EnMAP instrument will be sensitive to low grades of Nd/REEs in a pixel (see below).
The sensor effects. While EnMAP data exhibits excellent quality in both the VNIR and SWIR ranges, it is acknowledged that the bands at the longer wavelength end of the VNIR detector display erratic nonlinear behavior due to the fringing effect (EnMAP's unpublished internal report). The challenges faced by EnMAP in resolving the 870 nm feature may, in part, be attributed to this phenomenon, particularly beyond 900 nm, where the right shoulder of the feature is located.
Comparing the outcomes of this study with the analysis conducted by Mars 30 using WorldView-3 data underscores the significance of spectral resolution over spatial resolution in mapping REEs. Because despite WorldView-3's exceptional spatial resolution, it could not map Nd occurrences in the area. In contrast, EnMAP, with a spatial resolution of 30 m, succeeded due to its high spectral resolution and calibration accuracy. Certainly, high spatial resolution hyperspectral data can enable the detection and mapping of meter-scale veins in geologic outcrops. However, for spaceborne remote sensing data with restrictions in spatial resolution, enhancing the SNR and spectral resolution can increase their sensitivity and utility for REEs.
The spectral processing method. After testing various spectral processing methods, which included multiple target detection algorithms, similarity measures, feature fitting algorithms, and a support vector machine classifier 38 , it was observed that the choice of processing method has implications for successful Nd detection. Remarkably, none of the tested methods were able to generate results comparable to the map shown in Fig. 3 a (using the mapped pixels as endmembers/training data), highlighting the superiority of the absorption feature analysis and polynomial fitting technique for REE detection. This may explain why prior attempts to map REEs in the area using airborne data e.g., 28 , 35 were not very successful. The main advantage of the approach employed in this paper is that it does not require a priori knowledge about REE occurrences in a given area and rather it relies on the spectroscopic knowledge of rare-earth minerals for remote sensing mapping.
In summary, the ability to detect REEs using hyperspectral remote sensing data depends on geological and instrumental constraints. Geologically, it depends on the size of the target, its exposure level, the contained level of REEs, and the composition of accompanying minerals. Instrumentally, it primarily depends on the imaging system's SNR and spectral resolution, followed by spatial resolution, and the quality of atmospheric correction and processing methods.
While in this study, hundreds of pixels were identified to contain Nd, in similar remote sensing studies in the future, the detection of REE signatures, even in a single image pixel, should be considered promising for subsequent field studies. While identifying the rare-earth mineralogic host, as achieved here, may not be always practical or necessary for remote sensing studies, detecting the carbonate signature (via SWIR bands) in a carbonatite host 39 , 40 could further support the presence of REE in a target. It is important to note that as a remote sensing method, our methodology can only detect REE signatures at the surface without the ability to penetrate to depth.
This study demonstrated that EnMAP hyperspectral satellite data can directly and efficiently detect REEs in geological environments. EnMAP successfully resolved the distinctive absorption features of Nd at 740 and 800 nm arising from the Nd-rich bastnaesite ore in the Mountain Pass area. While EnMAP could resolve the feature at ∼ 580 nm, the feature was not suitable for REE mapping due to its low intensity and interference with iron oxides and the chlorophyll absorption feature occurring at ∼ 590 nm. EnMAP data was unable to confidently resolve the feature at ∼ 870 nm. The absorption feature analysis and polynomial fitting technique proved to be a superior and effective processing method for characterizing the prominent REE absorption features and mapping the occurrences and relative abundances of Nd in imaging spectroscopic data.
Detecting the spectral signature of REEs by spaceborne imaging spectroscopic data can take exploration activities for REEs to another level. Conventionally, carbonatite bodies, as the primary hosts of LREEs, have been explored through geophysical methods relying on airborne magnetic and radiometric surveys 41 . Introducing a remote sensing approach capable of detecting the contained REEs directly and mapping the underlying host mineralogy and alteration aureoles can complement the existing exploration portfolio, facilitating the discovery of new carbonatite bodies and REEs resources.
The EnMAP satellite data with its global coverage can be used to screen large areas for REE signatures. However, given its 30 m spatial resolution, it is expected to mainly detect well-exposed targets of sufficient Nd quantities/sizes in arid to semi-arid regions of the world. Advancements in atmospheric correction procedures and processing methods can aid in detecting lower grades and smaller Nd-bearing targets. Because REEs are often associated with each other, and because the host mineralogy does not highly modify the REE-related absorption features, remote sensing mapping of Nd should serve as an exploration pathfinder for light (and potentially heavy) REEs, irrespective of their deposit types.
Future work will involve establishing quantitative relationships between Nd grade and spectral signatures and testing the methodology across a diverse range of REE-rich deposits/prospects with varying levels of light/heavy REEs, outcrop exposures, geologic /conditions, and vegetation coverage. This could help to better understand the spectral behavior of REEs at EnMAP resolution and determine the instrument’s full capability in detecting and mapping REEs occurrences remotely.
Enmap hyperspectral data.
The EnMAP (Environmental Mapping and Analysis Program) hyperspectral satellite system was launched into orbit on April 1, 2022, and since November 2022 has been in routine operation 42 . EnMAP is a German satellite mission designed and operated by the German Aerospace Center (DLR) and funded by the Federal Ministry for Economic Affairs and Climate Action (BMWK) of Germany 43 , 44 .
The EnMAP data of the study area, collected on July 7th, 2022 at 18:47:54.75 UTC (11:47 local time) was obtained from the EOWEB® portal. The data was ordered using the following settings: Level 2A data with ozone and terrain corrections enabled, with no spectral interpolation, resampled by the nearest neighbor method. The data was processed using the March 2023 version of the EnMAP processor. The Level 2A orthorectified surface reflectance data of EnMAP comprises 224 spectral bands at 30-m spatial resolution. The VNIR bands used in this study cover the spectral range between 420 and 1000 nm at a spectral sampling interval of 6.5 nm and a spectral bandwidth of 8.1 nm. The VNIR bands maintain a signal-to-noise ratio (SNR) exceeding 400:1 and spectral stability better than 0.5 nm thanks to the instrument’s onboard calibration assembly 45 . These attributes render the EnMAP data an excellent choice for remote sensing mapping of REEs.
We applied a curve-fitting technique using a 4th-order polynomial 46 to detect and map REEs within the L2A data product. This technique enabled us to characterize the main absorption features of Nd at ∼ 580, ∼ 740, ∼ 800, and ∼ 865 nm (depicted in Fig. 1 ), as well as the carbonate feature between 2330 to 2340 nm. To achieve this, the local continuum was first removed between 520 to 900 nm for the VNIR and between 2230 to 2400 nm for the SWIR bands. Then, separate polynomials were fitted to the continuum-removed spectra within the ranges of 720–778, 770–825, 565–605, 825–895, and 2305–2365 nm (Fig. 1 ). Subsequently, the (real) root of the explicit first derivative was used to determine the wavelength of minimum reflectance (minimum wavelength). The coefficients of the fitted polynomial were also used to retrieve the depth, area, and width of the diagnostic absorption features. To eliminate the interfering effects of O 2 , the EnMAP band corresponding to oxygen’s residual absorption feature at 764 nm (band 62) was omitted from the calculations.
The retrieved spectral parameters then were subsequently employed in a stepwise decision-making process to identify Nd-bearing pixels. Initially, the pixels meeting the following criteria were isolated:
where λW and λD are the minimum wavelength and depth of the absorption feature centered at wavelength λ (nm). These results were further refined by retaining only the pixels that were linearly aligned in the scatterplot of 740D against 800D. Subsequently, the area of the 800 nm feature was used to represent the relative abundance of Nd in the mapped pixels. The relative abundance of carbonate rocks in the area was mapped based on the carbonate feature at ∼ 2340 nm (D > 0.13). The distribution of iron oxides was also mapped using \(\frac{{\lambda }_{690(nm)}}{{\lambda }_{450(nm)}}>\) 2.5. Finally, to better understand the spectral and statistical variability of the mapped pixels, 2D scatterplots were prepared from the retrieved spectral parameters. All these processes were applied to a spatial subset of the mosaicked EnMAP data covering the Sulphide Queen mine and the surrounding areas.
The obtained results were validated in three ways: (i) by comparing the EnMAP spectra to laboratory-based spectral measurements of a hand specimen collected from the Sulphide Queen mine, (ii) by superimposing the yielded anomalies over high-resolution satellite images of the area, available on Google Earth, and (iii) by matching the anomalies with local geologic data. The reflectance spectral data was collated from the datasets published by Neave, et al. 9 . The corresponding specimen (CR36), containing 30,848 ppm ( ∼ 3%) Nd, has been measured using an ASD Field-Spec Pro FR spectroradiometer, with sampling intervals of 1.4 and 2 nm between 350–1000 and 1000–2500 nm, respectively. The final spectrum has resulted from averaging tens of evenly spaced repeat measurements taken from across the sample surface so that the 1σ of the spectrum was < 0.5% relative 9 .
All EnMAP data are freely available through the EnMAP data access portal at the following link: https://www.enmap.org/data_access/ . The EnMAP data are licensed products of DLR [2022], all rights reserved.
Balaram, V. Rare earth elements: A review of applications, occurrence, exploration, analysis, recycling, and environmental impact. Geosci. Front. 10 , 1285. https://doi.org/10.1016/j.gsf.2018.12.005 (2019).
Article CAS Google Scholar
Wall, F. In Critical Metals Handbook Vol. 1 (ed. Gunn, G.) 439 (Wiley, 2014).
Google Scholar
Goodenough, K. M., Wall, F. & Merriman, D. The rare earth elements: Demand, global resources, and challenges for resourcing future generations. Nat. Resour. Res. 27 , 201–216. https://doi.org/10.1007/s11053-017-9336-5 (2018).
Commission, E. Critical Raw Materials Resilience: Charting a Path towards greater Security and Sustainability. 23 (Brussels, 2020).
Dill, H. G. The, “chessboard” classification scheme of mineral deposits: Mineralogy and geology from aluminum to zirconium. Earth Sci. Rev. 100 , 1–420. https://doi.org/10.1016/j.earscirev.2009.10.011 (2010).
Article ADS CAS Google Scholar
Turner, D. J., Rivard, B. & Groat, L. Visible and short-wave infrared reflectance spectroscopy of REE fluorocarbonates. Am. Miner. 99 , 1335–1346 (2014).
Article ADS Google Scholar
Verplanck, P. L., Mariano, A. N., Mariano, A. Jr., Verplanck, P. L. & Hitzman, M. W. Rare Earth and Critical Elements in Ore Deposits Vol. 18, 28 (Society of Economic Geologists, 2016).
Book Google Scholar
Balaram, V. Rare earth elements: A review of applications, occurrence, exploration, analysis, recycling, and environmental impact. Geosci. Front. 10 , 1285–1303. https://doi.org/10.1016/j.gsf.2018.12.005 (2019).
Neave, D. A. et al. On the feasibility of imaging carbonatite-hosted rare earth element deposits using remote sensing. Econ. Geol. 111 , 641–665. https://doi.org/10.2113/econgeo.111.3.641 (2016).
Article Google Scholar
Rowan, L. C., Kingston, M. J. & Crowley, J. K. Spectral reflectance of carbonatites and related alkalic igneous rocks; selected samples from four North American localities. Econ. Geol. 81 , 857–871. https://doi.org/10.2113/gsecongeo.81.4.857 (1986).
Clark, R. N. In Remote Sensing for the Earth Sciences Vol. 3 (ed. Rencz, A. N.) (Wiley, 1999).
Tan, W. et al. Feasibility of visible short-wave infrared reflectance spectroscopy to characterize regolith-hosted rare earth element mineralization. Econ. Geol. 117 , 495–508. https://doi.org/10.5382/econgeo.4877 (2022).
Tan, W. et al. Visible/near infrared reflectance (VNIR) spectral features of ion-exchangeable Rare earth elements hosted by clay minerals: Potential use for exploration of regolith-hosted REE deposits. Appl. Clay Sci. 215 , 106320. https://doi.org/10.1016/j.clay.2021.106320 (2021).
Möller, V. & Williams-Jones, A. E. A hyperspectral study (V-NIR-SWIR) of the Nechalacho REE-Nb-Zr deposit, Canada. J. Geochem. Explor. 188 , 194–215. https://doi.org/10.1016/j.gexplo.2018.01.011 (2018).
Dijkstra, A. H. et al. Identification of rare earth elements in synthetic and natural monazite and xenotime by visible-to-shortwave infrared reflectance spectroscopy. Phys. Chem. Miner. 51 , 16. https://doi.org/10.1007/s00269-024-01284-7 (2024).
Lorenz, S. et al. The potential of reflectance and laser induced luminescence spectroscopy for near-field rare earth element detection in mineral exploration. Remote Sens. 11 , 21 (2019).
Turner, D., Rivard, B., Groat, L. In: Simandl, G. J., Neetz, M. (eds) Symposium on Strategic and Critical Materials. 219–229 (British Columbia Ministry of Energy and Mines).
Morin Ka, S. Hyperspectral Characterisation of Rare Earth Minerals Vol. 50 (Geological Survey of Western Australia, 2012).
Percival, J. B. et al. Customized spectral libraries for effective mineral exploration: Mining national mineral collections. Clays Clay Miner. 66 , 297–314. https://doi.org/10.1346/CCMN.2018.064103 (2018).
Koerting, F. et al. A solar optical hyperspectral library of rare-earth-bearing minerals, rare-earth oxide powders, copper-bearing minerals and Apliki mine surface samples. Earth Syst. Sci. Data 13 , 923–942. https://doi.org/10.5194/essd-13-923-2021 (2021).
Kokaly, R. F. et al. USGS Spectral Library Version 7. Report No. 1035, 68 (Reston, VA, 2017).
Herrmann, S. Capacity of Imaging Spectroscopy for the Characterisation of REO, REE Bearing Minerals & Primary REE-Deposits (University of Potsdam, 2015).
Daempfling, H. L. C. et al. Automatic element and mineral detection in thin sections using hyperspectral transmittance imaging microscopy (HyperTIM). Eur. J. Mineral. 34 , 275–284. https://doi.org/10.5194/ejm-34-275-2022 (2022).
Gadea, O. C. A. & Khan, S. D. Detection of Bastnäsite-rich veins in rare earth element ores through hyperspectral imaging. IEEE Geosci. Remote Sens. Lett. 20 , 1–4. https://doi.org/10.1109/LGRS.2023.3249624 (2023).
Boesche, N. K. et al. Hyperspectral REE (rare earth element) mapping of outcrops—Applications for neodymium detection. Remote Sens. 7 , 5160–5186 (2015).
Miyatake, S. et al. In: 10th SEGJ International Symposium. 1–2.
Kilby, W. E., Riley, D. N. In: Simandl, G. J., Lefebure D. V. (eds) International Workshop, Geology of Rare Metals. 63–65 (BC Geological Survey).
Rowan, L. C. & Mars, J. C. Lithologic mapping in the Mountain Pass, California area using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data. Remote Sens. Environ. 84 , 350–366. https://doi.org/10.1016/S0034-4257(02)00127-X (2003).
Booysen, R. et al. Detection of REEs with lightweight UAV-based hyperspectral imaging. Sci. Rep. 10 , 17450. https://doi.org/10.1038/s41598-020-74422-0 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Mars, J. C. Mineral and lithologic mapping capability of WorldView 3 data at Mountain Pass, California, using true- and false-color composite images, band ratios, and logical operator algorithms. Econ. Geol. 113 , 1587–1601. https://doi.org/10.5382/econgeo.2018.4604 (2018).
Watts, K. E., Haxel, G. B. & Miller, D. M. Temporal and petrogenetic links between mesoproterozoic alkaline and Carbonatite Magmas at Mountain Pass, California. Econ. Geol. 117 , 1–23. https://doi.org/10.5382/econgeo.4848 (2022).
Castor, S. B. The Mountain Pass rare-earth carbonatite and associated ultrapotassic rocks, California. Can. Mineral. 46 , 779–806. https://doi.org/10.3749/canmin.46.4.779 (2008).
Mariano, A. N. & Mariano, A. Jr. Rare earth mining and exploration in North America. Elements 8 , 369–376. https://doi.org/10.2113/gselements.8.5.369 (2012).
Long, K. R., Van Gosen, B. S., Foley, N. K. & Cordier, D. In Non-Renewable Resource Issues: Geoscientific and Societal Challenges (eds Sinding-Larsen, R. & Wellmer, F.-W.) 131–155 (Springer, 2012).
Chapter Google Scholar
McDowell, M. L. & Kruse, F. A. In: Miguel Velez-Reyes & Fred A. Kruse (eds) Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XXI. 94721C.
Morin-Ka, S. Detection and Distinction of Rare Earth Elements Using Hyperspectral Technologies Vol. 16 (Geological Survey of Western Australia, 2018).
Qasim, M. & Khan, S. D. Detection and relative quantification of neodymium in Sillai Patti carbonatite using decision tree classification of the hyperspectral data. Sensors 22 , 7537 (2022).
Asadzadeh, S. & de Souza Filho, C. R. A review on spectral processing methods for geological remote sensing. Int. J. Appl. Earth Observ. Geoinform. 47 , 69–90. https://doi.org/10.1016/j.jag.2015.12.004 (2016).
Mars, J. C. & Rowan, L. C. ASTER spectral analysis and lithologic mapping of the Khanneshin carbonatite volcano, Afghanistan. Geosphere 7 , 276–289. https://doi.org/10.1130/ges00630.1 (2011).
Bedini, E. Mapping lithology of the Sarfartoq carbonatite complex, southern West Greenland, using HyMap imaging spectrometer data. Remote Sens. Environ. 113 , 1208–1219. https://doi.org/10.1016/j.rse.2009.02.007 (2009).
Simandl, G. J. & Paradis, S. Carbonatites: Related ore deposits, resources, footprint, and exploration methods. Appl. Earth Sci. 127 , 123–152. https://doi.org/10.1080/25726838.2018.1516935 (2018).
Storch, T. et al. The EnMAP imaging spectroscopy mission towards operations. Remote Sens. Environ. 294 , 113632. https://doi.org/10.1016/j.rse.2023.113632 (2023).
Guanter, L. et al. The EnMAP spaceborne imaging spectroscopy mission for earth observation. Remote Sens. 7 , 8830–8857 (2015).
Chabrillat, S. et al. in IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium. 5000–5003.
Carmona, E. et al. In: IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium. 1030–1033.
Haest, M., Cudahy, T., Laukamp, C. & Gregory, S. Quantitative mineralogy from infrared spectroscopic data. I. Validation of mineral abundance and composition scripts at the Rocklea Channel iron deposit in Western Australia. Econ. Geol. 107 , 209–228. https://doi.org/10.2113/econgeo.107.2.209 (2012).
Download references
This study was supported by the EnMAP science program (grant numbers 50EE1923 and 50EE2401) from the DLR Space Agency.
Open Access funding enabled and organized by Projekt DEAL.
Authors and affiliations.
Section of Remote Sensing and Geoinformatics, Helmholtz Centre Potsdam, GFZ German Research Centre for Geosciences, 14473, Potsdam, Germany
Saeid Asadzadeh, Nicole Koellner & Sabine Chabrillat
Institute of Soil Science, Leibniz University Hannover, 30419, Hannover, Germany
Sabine Chabrillat
You can also search for this author in PubMed Google Scholar
Conceptualization: SA Methodology: SA Investigation: SA, KN Data analysis and visualization: SA Supervision: CS Writing, review, and editing: SA, KN, CS.
Correspondence to Saeid Asadzadeh .
Competing interests.
The authors declare no competing interests.
Publisher's note.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .
Reprints and permissions
Cite this article.
Asadzadeh, S., Koellner, N. & Chabrillat, S. Detecting rare earth elements using EnMAP hyperspectral satellite data: a case study from Mountain Pass, California. Sci Rep 14 , 20766 (2024). https://doi.org/10.1038/s41598-024-71395-2
Download citation
Received : 06 February 2024
Accepted : 27 August 2024
Published : 05 September 2024
DOI : https://doi.org/10.1038/s41598-024-71395-2
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.
All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .
Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.
Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.
Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.
Original Submission Date Received: .
Find support for a specific problem in the support section of our website.
Please let us know what you think of our products and services.
Visit our dedicated information section to learn more about MDPI.
Gnss time series analysis with machine learning algorithms: a case study for anatolia.
2.1. data acquiring.
Click here to enlarge figure
3. methodology, 3.1. data segmentation and feature extraction, 3.2. accuracy and evaluation metrics, 4.1. preprocessing of the time series, 4.2. residual analysis, 5. discussion, 6. conclusions, supplementary materials, author contributions, data availability statement, acknowledgments, conflicts of interest, abbreviations.
GNSS | Global Navigational Satellite System |
TUSAGA | Turkish National Continuous GNSS Network |
RCMT | Regional Centroid Moment Tensor |
Mw | Moment Magnitude |
CME | Common Mode Error |
ML | Machine Learning |
LSTM | Long Short-Term Memory |
NAF | North Anatolian Fault |
EAF | East Anatolian Fault |
IGS | International GNSS Service |
AMLSTM NN | Attention Mechanism with Long Short-Time Memory Neural Network |
TP | True Positive |
TN | True Negative |
FP | False Positive |
FN | False Negative |
TP′ | True Positive Prime |
RNN | Recurrent Neural Network |
MSE | Mean Squared Error |
MAE | Mean Absolute Error |
RMSE | Root Mean Squared Error |
R | R-squared (Coefficient of Determination) |
ROC | Receiver Operating Characteristic |
AUC | Area Under the Curve |
GMT | Generic Mapping Tools |
Hyperparameter | Explanation |
---|---|
Number of trees | Number of boosting rounds |
Learning Rate | Step size shrinkage used to prevent overfitting |
Maximum depth | Maximum depth of a tree |
Minimum Child Weight | Minimum sum of instance weight (hessian) needed in a child |
Subsample | Fraction of observations to be randomly sampled for each tree |
Column sample | Fraction of features to be randomly sampled for each tree |
Hyperparameter | Explanation |
---|---|
Number of Layers | The depth of the network |
Number of Units per Layer | The number of memory cells in each layer |
Dropout Rate | The fraction of input units to drop during training to prevent overfitting |
Learning Rate | The step size for the optimizer |
Batch Size | The number of samples per gradient update |
Number of Epochs | The number of times the entire dataset is passed through the network during training |
Category | Features |
---|---|
Statistical Features | Mean, Standard Deviation, Skewness, Kurtosis |
Frequency Features | Fourier Transform Coefficients |
Trend Features | Linear Regression Coefficients, Polynomial Regression Coefficients |
Metric Features | Maximum Displacement, Minimum Displacement, Range |
Threshold Features | Threshold features for discontinuity detection |
Actual | Predicted | |
---|---|---|
TP and (TP′) | TN | |
FP | FN |
Model | Hyperparameter | Initial Value |
---|---|---|
XGBoost | Number of trees | 200 |
XGBoost, LSTM | Learning Rate | 0.1 |
XGBoost | Maximum depth | 7 |
XGBoost | Minimum Child Weight | 3 |
XGBoost | Subsample | 0.9 |
XGBoost | Column sample | 0.8 |
LSTM | Number of Layers | 2 |
LSTM | Number of Units per Layer | 100 |
LSTM | Dropout Rate | 0.3 |
LSTM | Batch Size | 64 |
Hyperparameter | Tested Values | Selected Value |
---|---|---|
Number of trees | [100, 200, 300, 500, 1000] | 200 |
Learning Rate | [0.1, 0.2, 0.3] | 0.2 |
Maximum depth | [3, 5, 7, 9] | 7 |
Minimum Child Weight | [1, 3, 5, 7, 10] | 5 |
Subsample | [0.5, 0.7, 0.9, 1.0] | 0.9 |
Column sample | [0.3, 0.5, 0.7, 0.8, 1.0] | 0.8 |
Number of Layers | [1, 2, 3] | 2 |
Number of Units per Layer | [50, 100, 150] | 100 |
Dropout Rate | [0.2, 0.3, 0.4] | 0.3 |
Batch Size | [16, 32, 64, 128] | 64 |
Metric | XGBoost | LSTM |
---|---|---|
MAE | 1.8 | 2.0 |
RMSE | 2.2 | 2.5 |
0.84 | 0.81 |
The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
Özbey, V.; Ergintav, S.; Tarı, E. GNSS Time Series Analysis with Machine Learning Algorithms: A Case Study for Anatolia. Remote Sens. 2024 , 16 , 3309. https://doi.org/10.3390/rs16173309
Özbey V, Ergintav S, Tarı E. GNSS Time Series Analysis with Machine Learning Algorithms: A Case Study for Anatolia. Remote Sensing . 2024; 16(17):3309. https://doi.org/10.3390/rs16173309
Özbey, Volkan, Semih Ergintav, and Ergin Tarı. 2024. "GNSS Time Series Analysis with Machine Learning Algorithms: A Case Study for Anatolia" Remote Sensing 16, no. 17: 3309. https://doi.org/10.3390/rs16173309
Article access statistics, supplementary material.
ZIP-Document (ZIP, 1173 KiB)
Mdpi initiatives, follow mdpi.
Subscribe to receive issue release notifications and newsletters from MDPI journals
IMAGES
VIDEO
COMMENTS
A case study is a research method that involves an in-depth examination and analysis of a particular phenomenon or case, such as an individual, organization, community, event, or situation. It is a qualitative research approach that aims to provide a detailed and comprehensive understanding of the case being studied. Case studies typically ...
The purpose of case study research is twofold: (1) to provide descriptive information and (2) to suggest theoretical relevance. Rich description enables an in-depth or sharpened understanding of the case. It is unique given one characteristic: case studies draw from more than one data source. Case studies are inherently multimodal or mixed ...
A case study is a detailed study of a specific subject, such as a person, group, place, event, organization, or phenomenon. Case studies are commonly used in social, educational, clinical, and business research. A case study research design usually involves qualitative methods, but quantitative methods are sometimes also used.
A case study is one of the most commonly used methodologies of social research. This article attempts to look into the various dimensions of a case study research strategy, the different epistemological strands which determine the particular case study type and approach adopted in the field, discusses the factors which can enhance the effectiveness of a case study research, and the debate ...
Definition, Methods, and Examples. Case study methodology offers researchers an exciting opportunity to explore intricate phenomena within specific contexts using a wide range of data sources and collection methods. It is highly pertinent in health and social sciences, environmental studies, social work, education, and business studies.
This book presents a disciplined, qualitative exploration of case study methods by drawing from naturalistic, holistic, ethnographic, phenomenological and biographic research methods.
A case study is a research approach that is used to generate an in-depth, multi-faceted understanding of a complex issue in its real-life context. It is an established research design that is used extensively in a wide variety of disciplines, particularly in the social sciences. A case study can be defined in a variety of ways (Table 5), the ...
The purpose of case study research is twofold: (1) to provide descriptive information and (2) to suggest theoretical relevance. Rich description enables an in-depth or sharpened understanding of the case. Robert Yin, methodologist most associated with case study research, differentiates between descriptive, exploratory and explanatory case studies:
A case study protocol outlines the procedures and general rules to be followed during the case study. This includes the data collection methods to be used, the sources of data, and the procedures for analysis. Having a detailed case study protocol ensures consistency and reliability in the study.
The term "case study" refers to both a specific research design or methodology, and a method of analysis for examining a problem. Mills et al. (2010) note that case study, both as a methodology and as a method—unlike many qualitative methodologies—is frequently used to generalize across populations.
Case study method is the most widely used method in academia for researchers interested in qualitative research (Baskarada, 2014). Research students select the case study as a method without understanding array of factors that can affect the outcome of their research.
Nevertheless, the case study researchers mentioned above emphasize different features. Stake points out that crucial to case study research are not the methods of investigation, but that the object of study is a case: "As a form of research, the case study is defined by the interest in individual cases, not by the methods of inquiry used."
Providing a complete portal to the world of case study research, the Fourth Edition of Robert K. Yin's bestselling text Case Study Research offers comprehensive coverage of the design and use of the case study method as a valid research tool. This thoroughly revised text now covers more than 50 case studies (approximately 25% new), gives fresh attention to quantitative analyses, discusses ...
A Case study is: An in-depth research design that primarily uses a qualitative methodology but sometimes includes quantitative methodology. Used to examine an identifiable problem confirmed through research. Used to investigate an individual, group of people, organization, or event. Used to mostly answer "how" and "why" questions.
Written in an accessible and jargon-free style, this book provides a comprehensive, student-friendly guide to the nature and use of case study research. Whether as part of a more substantial study or as the foundation for a self-contained smaller project, case studies provide viable and valuable alternatives to conducting large-scale research.
Abstract. This chapter explores case study as a major approach to research and evaluation. After first noting various contexts in which case studies are commonly used, the chapter focuses on case study research directly Strengths and potential problematic issues are outlined and then key phases of the process.
A case study is a research approach that is used to generate an in-depth, multi-faceted understanding of a complex issue in its real-life context. It is an established research design that is used extensively in a wide variety of disciplines, particularly in the social sciences. A case study can be defined in a variety of ways (Table 5), the ...
This book aims to provide case‐study researchers with a step‐by‐step practical guide to "help them conduct the study with the required degree of rigour" (p. xi). It seeks to "demonstrate that the case study is indeed a scientific method" (p. 104) and to show "the usefulness of the case method as one tool in the researcher's ...
Learn how to conduct a case study, a detailed study of a specific subject in its real-world context. Find out when to use a case study, how to select a case, build a theoretical framework, collect and analyse data, and write up your findings.
Definitions of qualitative case study research. Case study research is an investigation and analysis of a single or collective case, intended to capture the complexity of the object of study (Stake, 1995).Qualitative case study research, as described by Stake (), draws together "naturalistic, holistic, ethnographic, phenomenological, and biographic research methods" in a bricoleur design ...
Case study methodology can prove a confusing and fragmented topic. In bringing diverse notions of case study research together in one volume and sensitizing the reader to the many varying definitions and perceptions of 'case study', this book equips researchers at all levels with the knowledge to make an informed choice of research strategy.
Case studies are most appropriate when asking "how" or "why" questions. 1 According to Yin, 1 how and why questions are explanatory and lead to the use of case studies, histories, and experiments as the preferred research methods. In 1 study from Canada, eg, the following research question was presented: "How and why did stakeholders ...
The case study is not a research method, but researchers select methods of data collection and analysis that will generate material suitable for case studies. Freud (1909a, 1909b) conducted very detailed investigations into the private lives of his patients in an attempt to both understand and help them overcome their illnesses.
The course will begin with a focus on the philosophy of science, theory construction, theory testing, causality, and causal inference. With this epistemological grounding, the seminar will then explore the core issues in case study research design, including methods of structured and focused comparisons of cases, typological theory, case selection, process tracing, and the use of ...
Rare earth elements (REEs) exhibit diagnostic absorption features in the visible-near infrared region, enabling their detection and identification via spectroscopic methods. Satellite-based remote ...
This study addresses the potential of machine learning (ML) algorithms in geophysical and geodetic research, particularly for enhancing GNSS time series analysis. We employed XGBoost and Long Short-Term Memory (LSTM) networks to analyze GNSS time series data from the tectonically active Anatolian region. The primary objective was to detect discontinuities associated with seismic events.
Abstract. This chapter explores case study as a major approach to research and evaluation. After first noting various contexts in which case studies are commonly used, the chapter focuses on case study research directly.
During the pandemic, online-based learning faced numerous challenges, such as internet connectivity issues, difficulties in comprehension, an increase in assignments, one-sided interactions, and reduced concentration. Nevertheless, private tutoring institutions emerged as invaluable resources, enabling students to revisit lesson materials, enhance their understanding, and complete assignments ...