carsontelford/RVF_Banana_PublicRepo

Repository for dataset analyzed in longitudinal analysis of RVF outbreaks and crop cultivation in Uganda from 2016-2024. # Banana Cultivation and Rift Valley Fever in Uganda This repository contains the public dataset and analysis code for Aim 1 of a dissertation project evaluating associations between agricultural land cover and Rift Valley fever (RVF) outbreaks in Uganda from 2016-2024. The analysis uses annual 5 x 5 km grid-cell observations in southwestern Uganda. Each grid cell-year includes RVF outbreak status, remotely sensed land-cover composition, and environmental or population covariates used in the main regression and sensitivity analyses. ## Repository Contents - `rvf.csv`: Main public analysis dataset with one row per grid cell-year. - `scripts/01_classification/`: Longleaf/HPC land-cover classification scripts for each annual Sentinel-2 raster stack. - `scripts/02_dataset/`: Local dataset assembly script that extracts land-cover and covariate summaries to grid cells. - `scripts/03_analysis/`: Local main model-fitting and figure-generation script. - `scripts/04_sensitivity/`: Longleaf/HPC stochastic classification and sensitivity-analysis scripts, plus the required C++ helper. - `scripts/05_diagnostics/`: Local/Longleaf diagnostic scripts for classification performance and calibration. Large intermediate rasters, raw satellite imagery, raw line lists, shapefiles, and private project folders are not included. ## Dataset The main dataset, `rvf.csv`, includes: - Spatial and temporal identifiers: `ID`, `x`, `y`, `year` - RVF outcome variables: `rvfbin`, `cumcases` - Land-cover variables: banana, coffee, ground crops, pooled crop land, non-crop land, and stochastic sensitivity-analysis versions - Environmental covariates: precipitation, temperature, elevation, slope, topographic wetness index, soil texture, and distance measures - Population/livestock covariates: human population density, livestock density, and related focal summaries Land-cover proportions are measured as proportions of non-water land area within each grid cell. Variables ending in `_stoch` correspo …