Undergraduate research: satellite data analysis of drought and natural water reservoirs in East Africa
# maple-scholars-2026
Undergraduate research: satellite data analysis of drought and natural water reservoirs in East Africa
---
## Overview
This project investigates the natural water "reservoirs" — groundwater, soil moisture, and seasonal wetlands — that sustain or end droughts in Kongwa District, central Tanzania. Kongwa is a largely rain-fed agricultural region with very little irrigation or built water storage, which makes it an ideal place to ask: when the rain stops, what holds water in the landscape, and how does the area recover?
The research is motivated by a public-health link. A previous study in the region found that a drought appeared to end earlier than rainfall alone could explain, coinciding with a measured drop in crop mycotoxins (toxins produced by fungi under crop stress). If a natural water source eased that drought sooner than expected, understanding it could have implications for food safety and water management across similar dryland farming regions.
**Researcher:** Timothy Masaba Okiror
**Faculty Mentor:** Prof. Paul Meyer Reimer (Physics)
**Institution:** Goshen College
---
## Research Question
What natural water reservoirs sustain or end droughts in Kongwa District, and can changes in satellite-measured water storage help explain the early drought recovery observed in prior research?
---
## Data Sets
**Raw GRACE / GRACE-FO (NASA JPL MASCON)**
Satellite measurements of total water storage anomaly (TWSA), derived from tiny changes in Earth's gravity field. Extracted via Google Earth Engine. Native resolution is coarse (~3 degrees). Units: centimeters of equivalent water height; baseline 2004–2010.
**GRACE-SeDA (Gou & Soja, 2024, Nature Water)**
A deep-learning-downscaled GRACE product that sharpens the blurry satellite signal to 0.5 degree resolution by fusing it with the WaterGAP global hydrology model. Primary dataset for the fine-scale analysis. Monthly, 2002–2022. Units: millimeters; baseline 2004–2009.
DOI:
doi.org …