Thesis Topics
Non-Invasive Phosphate Detection in Soils Using Hyperspectral Imaging
1. Thema / Topic:
Non-Invasive Phosphate Detection in Soils Using Hyperspectral Imaging
2. Type of Thesis / Betreuung:
Supervisor: Prof. Dr. Joachim Müller
Contact Person: Khandoker Ahammad, Alice Reineke
3. Beschreibung der Arbeit / Description and objective:
Timely, non-destructive assessment of plant-available phosphorus (P) in soils is essential for sustainable fertilization and environmental protection. Conventional wet-chemistry extractions are accurate but slow, chemical-intensive, and not easily scalable. This thesis investigates whether the hyperspectral camera can enable non-invasive detection and screening of soil phosphate status using spectral analysis and classical statistics. The work concentrates on rigorous reflectance calibration and repeatable imaging of soils with known P levels; preprocessing (e.g., baseline/continuum correction, derivatives, and normalization) to reveal subtle spectral features related to soil constituents that co-vary with P availability within the camera’s spectral range; and the development of rule-based spectral indices and band-ratio diagnostics to classify soils into predefined P status classes and/or estimate P against laboratory references using simple calibration curves. The objectives of this study are:
· Capture hyperspectral images (HSI) of soils across predefined P levels.
· Standardize non-invasive imaging (illumination, dark/white refs, sample prep).
· Apply pre-processing (reflectance, smoothing, derivatives, continuum removal).
· Fit simple calibrations and rule thresholds; report R²/RMSE/bias/Bland–Altman.
· Test robustness across texture/moisture, define operational limits and quality control
4. Anforderungen / Requirements for the applicants
· Background in agricultural/environmental sciences, soil science or remote sensing.
· Interest in spectroscopy and hyperspectral imaging; careful experimental skills.
· Basic skills in data handling and classical statistics (e.g., regression, correlation, uncertainty analysis) in Python or similar.
· Good documentation habits and attention to details.
5. Kontakt / How to apply:
Please send your CV and your academic transcripts to:
Khandoker Ahammad: khandoker.ahammad@uni-hohenheim.de
Alice Reineke: a.reineke@uni-hohenheim.de
6. Bewerbungsfrist / Application deadline:
30th November 2025
7. Dauer der Arbeit / Duration:
Six months