Machine learning where water, energy, and climate meet.

T-Risk: a patented hybrid physics–ML climate transition-risk model

Architect and methodology owner of an energy-factor model (U.S. Patent No. 11,694,269) that embeds output from process-based climate and energy-system models — GCAM (PNNL/JGCRI), REMIND-MAgPIE (Potsdam Institute), and MESSAGEix-GLOBIOM (IIASA), calibrated to the IPCC Sixth Assessment and hindcast-validated against observed global emissions — inside machine learning pipelines that measure security-level sensitivity to energy-transition pathways spanning net-zero, current-policy, and intermediate futures.

Deployed in an investment strategy with $500M+ associated AUM · methodology deep-dive

SAFE: a four-module engine from climate models to firm-level prediction

First author of the SAFE technical whitepaper (v2.0, 2026): a framework built directly on three process-based integrated assessment models of the coupled energy–land–climate system — GCAM 6.0, REMIND-MAgPIE, and MESSAGEix-GLOBIOM — whose outputs are translated into scenario-conditional monthly financial paths for 20,000+ global firms. Four additive modules isolate distinct transmission channels: direct carbon cost, market-priced transition sensitivity via Bayesian hierarchical energy-factor betas (MCMC-estimated, re-fit quarterly), stranded assets through a merit-order carbon-budget allocation, and balance-sheet propagation feeding multi-horizon probability-of-default term structures. Cross-model output ranges bound structural uncertainty.

Every downstream credit-risk movement is auditable back to its physical-scenario origin — validated through a multi-layered QA pipeline with versioned, reproducible transformations.

Extreme weather and climate risk to energy supply

Doctoral research at the water–energy–climate nexus (advisor: Prof. Anu Ramaswami, NSF IGERT Fellow): quantifying how hydroclimatic variability and extremes — drought-constrained cooling water at thermoelectric plants, insufficient streamflow, and heat-driven demand — threaten the reliability of grid-scale electricity supply, in the U.S. and India. The framework was published in the Journal of Industrial Ecology, and a dissertation chapter with Prof. Balaji Rajagopalan statistically linked climate variability to grid-scale supply reliability.

Hydrology-aware modeling of coupled infrastructure systems — the founding thread of my career.

Monsoon variability and power-system reliability in India

As a U.S. Fulbright Scholar, studied how monsoon-driven hydroclimatic variability shapes electricity supply and demand — connecting seasonal hydrology, reservoir behavior, and grid operations in one of the world's most climate-exposed power systems.

Earth Institute Fellowship: climate risk to urban energy systems

As an Earth Institute Fellow at Columbia University — a postdoctoral appointment held jointly with a Lectureship in Statistics — extended the water–energy–climate research program to global scale: working with large observational, station-network, and scenario-model datasets to quantify how warming reshapes urban electricity demand for cooling and heating worldwide; published in Energy (Waite, Cohen, et al., 2017; full text), with an earlier working paper and analysis code in the open. Also taught data analytics to energy professionals in West Africa with UNIDO, including a published wind-resource analysis for the ECOWAS renewable-energy centre.

weatheR: an open-source weather-data pipeline for R

Co-developed weatheR, an R package for geo-referenced, quality-controlled batch retrieval of hourly weather-station records from NOAA's National Climatic Data Center — the world's largest active weather archive. The package handles station lookup by ID, city, or coordinates, data download, cleaning and interpolation, and visualization, making large meteorological datasets accessible to any analyst working in R.

Package on GitHub · tutorial & methodology · Python companion pyNOAA · more at github.com/Ecohen4

Lifecycle emissions of electricity generation, adopted by the IPCC

Contributed to NREL research on lifecycle greenhouse-gas emissions of electricity generation technologies that was adopted in the IPCC Special Report on Renewable Energy (Arvizu et al., 2011).