LeahN67/greenscope-asal-casuality-dataset

This repository documents the construction of a causal-ready climate dataset for Kenya’s Arid and Semi-Arid Lands (ASALs) # GreenScope ASAL Causal Data Cube (Module 0) **Infrastructure Test Project: Validating Causal-Ready Data Pipelines for Kenya's ASALs** ## 📌 Overview This repository serves as a **technical validation project** for the GreenScope Analytics platform. It demonstrates how to transform fragmented, multi-source environmental data into a **certified 3D Data Cube** specifically designed for Kenya's Arid and Semi-Arid Lands (ASALs). The project focuses on **Module 0: Data Capture & Readiness**, implementing the "Last Mile" of climate data engineering where raw datasets become analysis-ready, causally-structured data products. --- ## 🎯 Key Features - ✅ **Fully Automated Pipeline** - From raw data to analysis-ready cube - ✅ **Causal Structure** - Variables organized by causal hierarchy (Drivers → Stressors → Responses) - ✅ **Multi-Source Integration** - Harmonizes CHIRPS, ERA5, MODIS, ENSO, and IOD data - ✅ **Production Ready** - Includes error handling, logging, and resume capability - ✅ **Reproducible** - All paths configurable, no hardcoded values - ✅ **Open Source** - Apache 2.0 license, community contributions welcome --- ## 🏗️ GreenScope Infrastructure Principles As a test project for the GreenScope platform, this pipeline validates four core infrastructure pillars: ### 1. **Causal Admissibility** Variables are tiered (Global Drivers → Local Stressors → Ecosystem Responses) to prevent impossible causal loops and ensure scientifically valid inference. ### 2. **Physical Integrity** "Sanity Gates" enforce real-world physical bounds (e.g., non-negative rainfall, NDVI within [-0.2, 1.0]) to catch data corruption early. ### 3. **Auditability** SHA-256 checksums and immutable logging create a transparent data provenance record for every download and transformation. ### 4. **Interoperability** Harmonizes different spatial resolutions (CHIRPS 0.05°, MODIS 250m, ERA5 0.1°) into a unified, analysis-ready grid using bilinear interpolation. --- ## 🧱 Causal Registr …