Postdoctoral Research Scholar at the University of Arizona applying machine learning to satellite and in-situ observations to build robust geospatial solutions across precipitation, snow & ice, land–atmosphere interactions, and climate data records.
I build end-to-end geospatial algorithms and workflows using satellite observations (IR/MW/optical), in-situ networks, and reanalysis to deliver operational Earth-observation products. Work spans precipitation, snow & ice mapping, rain-area detection, and uncertainty quantification, with emphasis on scalable Python/xarray pipelines and reproducible evaluation.
At UArizona I collaborate with multiple teams—including the NASA GPCP team—on AVHRR-based and blended retrievals while contributing to broader university research initiatives.
ML for retrievals and detection (classification, regression, ensembles, uncertainty), integrating physical constraints and multi-sensor data.
Blending satellite, CMLs, gauges, and reanalysis; evaluation for climate data records and long-term monitoring.
Scalable Python/xarray workflows, HPC, and cloud-ready processing for near-real-time and retrospective products.
Selected projects and collaborations across research and real-world applications.
Developing and evaluating an AVHRR-based precipitation mapping approach using double-ML strategies; contributing to long-term precipitation records and cross-sensor consistency.
Operationalizing my PhD algorithm by blending satellite, Commercial Microwave Links (CML), and station data for real-time rainfall maps and short-term nowcasts.
Validation and uncertainty evaluation for operational satellite precipitation retrievals.
Rainfall retrievals blending MSG SEVIRI with CMLs; rain-area detection and near-real-time mapping workflows.
Kumah, Zandi & Behrangi.
Read DOIKumah, Maathuis, Hoedjes & Su.
Read DOIKumah et al.
Read DOIZandi, Kumah & Behrangi.
Read abstractJW Harshbarger Building, Room 316D, University of Arizona