Project overview:
- Assess how climate change projections affect the viability of index-based insurance schemes for smallholder farmers.
- Quantify changes in indemnity frequency and basis risk under future climate scenarios (CMIP6), using observed and remote-sensing data.
Responsibilities / Tasks:
- Ingest and preprocess large-scale climate projections (CMIP6) and historical weather data (CHIRPS), and vegetation indices (MODIS NDVI).
- Build and evaluate index-based insurance triggers and simulate payouts; estimate crop-yield relationships using FAO crop yield statistics.
Technical environment & tools:
- Primary languages: Python or R.
- Libraries/tools: xarray / netCDF4 for climate data, Prophet or ARIMA for time series modeling, Matplotlib / Seaborn for visualization, Jupyter Notebooks + Git for documentation and reproducibility.
Methodology & analysis:
- Compute relevant climate indicators (e.g., rainfall deficits, cumulative anomalies) from CHIRPS and CMIP6 outputs; bias-correct or downscale as necessary.
- Model historical yield-weather/NDVI relationships (statistical models or machine learning) to link index triggers to expected losses and estimate basis risk.
Deliverables & expected outputs:
- Reproducible Jupyter notebooks and scripts to process data, run simulations, and produce figures.
- A technical report documenting methodology, results, sensitivity to model choices, and recommendations for index design under future climates.
Candidate profile & skills:
- Good programming skills in Python or R and experience with handling netCDF / xarray or equivalent.
- Knowledge of time series modelling (Prophet, ARIMA) and experience with geospatial / remote sensing datasets (MODIS NDVI, CHIRPS).
Application:
- To apply, send your CV, motivation letter and example code/notebooks to stages@hydatis.fr. You can also consult the company website: https://www.hydatis.com
- Use the email subject: "Application Clim-05: Assessing the Impact of Climate Change on Index-Based Insurance (PFE)" and indicate your preferred programming language (Python or R) and availability.