FAdekilae@agcenter.lsu.edu Google Scholar ORCID GitHub
Frank Anyoka AdekilaeSoil Physics · Hydrology · Remote Sensing · Environmental Data Analysis
Portrait of Frank Anyoka Adekilae

PhD Student & Research Assistant
Founder, Tinga Lens

I am a PhD student and research assistant in the Soil Hydrology and Remote Sensing (SHaRE) Lab, School of Plant, Environmental, and Soil Sciences (SPESS) at Louisiana State University.

My research focuses on soil moisture dynamics, particularly the relationship between near-surface observations and rootzone soil water storage. I combine field measurements, satellite remote sensing, and computational methods to improve drought monitoring and support agricultural decision-making.

I am also the founder of Tinga Lens, an independent project that turns public satellite data into district-level maps of drought, fire, vegetation, flooding and more for Ghana.

Education

August 2024 – Present · Expected completion May 2030

PhD, Plant, Environmental, and Soil Sciences

Louisiana State University, Baton Rouge, Louisiana, USA · Concentration: Soil Physics and Remote Sensing

Completed July 2024 · Distinction

International MSc in Soils and Global Change (Erasmus Mundus)

Ghent University, Belgium, and Aarhus University, Denmark · Specialization: Physical Land Resources and Global Change

Completed September 2021 · First Class Honours

BSc, Natural Resources Management

Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana · Specialization: Silviculture and Forest Management

Research Interests

Surface–Rootzone Connectivity

How near-surface soil moisture observations relate to deeper soil water storage, and how long soils retain the signal of wet and dry periods.

Satellite Remote Sensing

Combining satellite soil moisture and thermal observations with ground measurements for hydrologic applications.

Drought & Precision Agriculture

Developing interpretable soil moisture indicators to support drought monitoring and agricultural decisions.

Machine Learning for Soil Properties

Evaluating how machine learning can predict soil hydraulic properties, and where data limitations, spatial scale, and transferability constrain it.

Evapotranspiration & Land–Atmosphere Interactions

Using surface and air temperature patterns to estimate daily evapotranspiration across different climate regimes.

Environmental Monitoring Networks

Designing and deploying soil, vegetation, and weather sensor networks that deliver quality-controlled, near-real-time data.

Publications

2026 · Peer-reviewed review article

Pedotransfer functions in the era of machine learning: Data limitations, spatial scale, transferability, and future directions

Vadose Zone Journal · doi:10.1002/vzj2.70154

2026 · Peer-reviewed article

Pseudo-diffusivity characteristic curves for surface–rootzone soil hydrologic connectivity

Vadose Zone Journal · doi:10.1002/vzj2.70084

Additional publications will be added as they become available.

Conference Presentations

December 2025 · New Orleans, Louisiana · Poster

Modeling rootzone moisture dynamics from surface observations using pseudo-diffusivity

Adekilae, F. A., & Sehgal, V. · American Geophysical Union (AGU) Fall Meeting

November 2025 · Salt Lake City, Utah · Short oral presentation and poster

Influence of soil and climate on surface–rootzone hydrologic connectivity

Adekilae, F. A., & Sehgal, V. · CANVAS 2025 (ASA, CSSA, SSSA Annual Meeting)

August 2025 · Fukushima, Japan · Flash oral presentation and poster

Enhancing rootzone soil moisture dynamics estimation with a state-based low-pass filter

Adekilae, F. A., & Sehgal, V. · 2025 Kirkham Conference

June 2025 · Fort Collins, Colorado · Oral presentation

Enhancing rootzone soil moisture dynamics estimation with a state-based model

Adekilae, F. A., & Sehgal, V. · 2025 National Soil Moisture Workshop, Colorado State University · Watch the recording

Awards & Honors

2026

Bayer Fellow

American Society of Agronomy (ASA), Crop Science Society of America (CSSA), and Soil Science Society of America (SSSA). Competitive fellowship selected from more than 175 graduate student and postdoctoral applicants, including Bayer mentorship, leadership development, CANVAS participation, and $1,000 in travel support.

2025

Kirkham Conference Travel Grant

Agronomic Science Foundation. Competitive $2,500 travel grant through the Kirkham Fund and Van der Ploeg Fund to attend and present research at the 2025 Kirkham Conference in Fukushima, Japan.

2022–2024

Erasmus Mundus Joint Master Scholarship

European Union. Fully funded €49,000 scholarship covering tuition, insurance, international travel, installation costs, and monthly subsistence for the International Master of Science in Soils and Global Change.

2017–2021

Mastercard Foundation Scholars Program Scholarship

Kwame Nkrumah University of Science and Technology (KNUST), Ghana. Fully funded four-year undergraduate scholarship covering tuition, accommodation, academic resources, leadership development, and living expenses.

Research & Field Experience

August 2024 – Present · Baton Rouge, Louisiana

Graduate Research Assistant, SHaRE Lab

Louisiana State University. Develop physics-based models of rootzone soil moisture, analyze large satellite and climate datasets, and help build and deploy LaCADIAN, a statewide climate and digital agriculture monitoring network. Explore LaCADIAN data (interactive map and downloads)

February – August 2022 · Ghana

Field Assistant, EU-funded ReDIAL Project

Kwame Nkrumah University of Science and Technology. Led farmer training on sustainable fertilization and conservation agriculture, coordinated soil sampling and field data collection, and supported climate-smart agriculture and sustainable land management with farming communities.

March – May 2022 · Remote

Remote Conservation Extern

National Geographic Society and The Nature Conservancy. Researched marine conservation challenges and community impacts, engaged community leaders and conservation practitioners, conducted a landscape and gap analysis, and developed a community-focused conservation proposal. View my StoryMap: The Story of the Hammerhead Shark Fishery in Ghana

Service & Leadership

2026 – Present

Founder, Tinga Lens

Started and lead an independent environmental monitoring project for Ghana. I choose the datasets, design the methods, build the processing, and review each map before it is published. tingalens.org

2025–2026

Scientific peer review

Reviewer for international journals and conferences in soil science, hydrology, and remote sensing.

December 2025 · New Orleans

AGU Fall Meeting workshop

Co-organizer and instructor, Large-Scale Geospatial Data Analysis and Visualization in R. Course materials

August 2025 · J-Village, Fukushima, Japan

2025 Kirkham Conference

Session co-organizer, Rapid Poster Presentations Session #3.

Louisiana State University

SPESS Graduate Student Association

Vice President (September 2026–present); Treasurer (October 2025–September 2026).

Community Engagement

September 2018 – Present · Natugnia, Ghana

Team Leader, Campaign Against Child Marriage

Lead annual community education and awareness initiatives on child marriage and its social and educational impacts; mentor youth on academic achievement and personal development; coordinate community stakeholders and volunteers; and lead fundraising for outreach, student recognition, and educational initiatives.

April – December 2020 · Natugnia, Ghana

Community COVID-19 Sensitization Team Leader

Led community COVID-19 awareness and public-health education in the local language and coordinated communication of safety practices.

Skills & Tools

Programming & Analysis

RPythonMATLABSPSSStatistical modelingTime-series analysisInverse modelingParallel computing

Soil & Hydrologic Modeling

HYDRUS-1DRETCAquaCropSoilViewSoil hydraulic characterizationRootzone soil moisture modeling

Geospatial & Remote Sensing

Google Earth EngineArcGIS ProQGISERDAS ImagineNASA SMAPNetCDF/HDF5

Environmental Data

USCRNSCANSoilGrids250mCGIAR aridityData quality control

Field & Laboratory

Sensor installationSoil samplingHYPROPWP4CKSATParticle-size analysisBulk density

Communication

Scientific writingConference presentationsPeer reviewWorkshop instructionStudent mentoring

Professional Memberships

  • American Geophysical Union (AGU), Student Member · 2025–Present
  • American Society of Agronomy (ASA), Graduate Student Member · 2024–Present
  • Crop Science Society of America (CSSA), Graduate Student Member · 2024–Present
  • Soil Science Society of America (SSSA), Graduate Student Member · 2024–Present

Portfolio

Tinga Lens

A reproducible, district-level environmental intelligence platform for Ghana

Tinga Lens is an independently developed environmental monitoring platform that turns public Earth observation, climate, soil, land-cover and biodiversity data into district-level information for Ghana. It covers 260 districts and presents the results as interactive maps, district profiles, downloadable outputs and plain-language explanations. Its outputs are not official warnings or field measurements. I am the project owner and developer.

Read the case study

The problem

Environmental data come from many independent providers, and using them together at district scale means working across different grids, time steps, formats, product definitions, access systems and licences. Users also need to know an indicator's reference period, source, uncertainty and meaning, not only its value.

What I built

A Python data-processing and update framework and a JavaScript web application. The repository separates the data-ingestion and indicator scripts (scripts/), the processed outputs (data/), the product configuration and interactive maps (js/) and the documentation (docs/). The script scripts/update.py runs each product according to data availability, required credentials, existing output files and rebuild rules. A GitHub Actions workflow runs the processing every Monday and publishes the updated datasets.

The inputs include CHIRPS rainfall, NASA SMAP Level-4 soil moisture, NASA VIIRS vegetation and active-fire data, Copernicus Sentinel-1 flood products, forest and human-modification products, agricultural and soil information, and GBIF occurrence records. Some products use Google Earth Engine, and others use files downloaded directly. The tools are Python, NumPy, pandas, GeoPandas, Rasterio, scikit-learn, NASA Earthdata access, the Google Earth Engine API, JavaScript and Leaflet.

Example: district rainfall anomalies

The drought pipeline (scripts/drought.py) turns monthly CHIRPS rainfall into a district-level three-month rainfall indicator:

  1. Retrieves compatible CHIRPS monthly grids, preferring CHIRPS v3.0 when it is available.
  2. Aggregates the monthly rasters to Ghana's districts, with special treatment for districts smaller than a grid cell.
  3. Computes rolling three-month totals for every district.
  4. Compares the latest total with the totals ending in the same month across 1991–2020.
  5. Calculates an empirical percentile, with ties counted as half, and assigns a relative dryness class.
  6. Leaves a district unrated when its baseline three-month rainfall is below 30 mm.
  7. Publishes the summary maps and each district's rainfall history as JSON files.

The percentile is P = 100 × (number of baseline totals below the current value + 0.5 × number tied) / number of baseline years

This indicator shows rainfall relative to a district's own history. It does not directly measure crop damage, streamflow, groundwater or hydrological drought.

Architecture and operations

Input data → product-specific Python processing → district-level JSON datasets → JavaScript and Leaflet maps and downloads. The scheduled workflow installs the dependencies, restores download caches, runs the processing script and commits the changed data files. Credentials for NASA Earthdata, FIRMS, GBIF and Google Earth Engine are read from repository secrets and are not in the published scripts. Datasets that update regularly are kept separate from those rebuilt only when a source or model changes. The workflow lets each layer finish even if another one fails, so successful layers are published and failures are reported afterwards.

Quality control and limits

Tinga Lens labels the type of evidence behind each product: observations, environmental conditions, products published by others, derived estimates and experimental models. This keeps model output from being presented as direct measurement.

The validation notes report a leave-one-fire-year-out test of the experimental fire-probability model, with ROC-AUC, Brier scores, calibration and a comparison with a seasonal reference model. They also include event-based checks of selected flood outputs against other mapped events and public reports. These are preliminary evaluations and not a systematic ground validation of every map.

Known limits are the use of district averages, reference periods that differ between products, bias in species occurrence records, cloud and revisit constraints, classification uncertainty, and the lack of systematic comparison with ground measurements in Ghana for most products.

Reproducibility and licensing

The code is public on GitHub and a versioned release is archived on Zenodo. The README explains how to install the dependencies and run the update process, and some products need external accounts. The code is under the MIT licence, original documentation and data are under CC BY-NC 4.0, and third-party terms apply to the source data.

What this shows

  • Coverage of 260 Ghanaian districts through one common reporting interface.
  • Integration of many independent environmental data sources and product types.
  • An automated, scheduled update workflow with product-specific rebuild logic.
  • A public site and repository, with a citable Zenodo release (v1.0.0).
  • Documented experimental model validation and known limitations.

I make no claims here about user numbers, prediction skill across all products or uptime, because those have not been measured. The case study is based on the public repository and website as reviewed on 9 November 2026.

Contact

School of Plant, Environmental, and Soil Sciences
Louisiana State University
Baton Rouge, Louisiana

Email: FAdekilae@agcenter.lsu.edu

GitHub · ORCID · Google Scholar