About

Paulina Jaramillo is a Trustee Professor in the Department of Engineering and Public Policy at Carnegie Mellon University. Her research focuses on climate-resilient environmental systems and technologies, sustainable energy and transportation systems, and the societal implications of engineered systems. As a faculty member, she contributes to advancing understanding and solutions related to environmental sustainability and energy policy.

Research topics

• Computer Science
• Business
• Engineering
• Political Science
• Environmental science
• Economics
• Natural resource economics
• Computer Security
• Environmental protection
• Ecology
• Environmental resource management
• Finance
• Climatology
• Geography
• Geology
• Meteorology

Selected publications

• A Regression Framework for Spatial Downscaling of NASA GEOS-CF PM2.5 Predictions Using Satellite Observations, Spatial covariates, and Ground-Based Measurements

  Zenodo (CERN European Organization for Nuclear Research) · 2026-01-23

  peer-reviewOpen accessSenior author

  This repository contains the datasets and source code required to reproduce the 1-km resolution PM2.5 estimates for Africa as described in the accompanying manuscript: "A Regression Framework for Spatial Downscaling of NASA GEOS-CF PM2.5 Predictions Using Satellite Observations, Spatial Covariates, and Ground-Based Measurements." The project utilizes a Random Forest-based downscaling framework to bridge the gap between global chemical transport models (CTMs) and sparse ground-level monitoring. This framework produces high-resolution surface estimates that resolve intra-urban pollution gradients and align with measured patterns at U.S. embassy locations across the African continent. Contents: Code and Notebooks: 00_GEE_Data_Extraction.ipynb: This notebook executes the multi-source data extraction from the Google Earth Engine Python API. It performs the aggregation of GEOS-CF and environmental predictors into monthly and annual mean rasters at 0.25° (training) and 0.01° (downscaling) resolutions. 01_Baseline_Model.ipynb: This notebook executes the primary downscaling pipeline. It implements the stepwise, importance-driven forward predictor selection for each of the five African regions, followed immediately by the training and 10-fold cross-validation of the uncalibrated (CTM-only) Random Forest and OLS baseline models. 02_Calibrated_Model.ipynb: This notebook implements the final ground-calibration phase. It integrates the spatial scale indicator (Si) and U.S. Embassy ground monitoring data to bias-correct the baseline estimates and produce the final 1-km pollution surfaces. 03_Downscaled_Estimates_ByModel.ipynb: Generates final 1-km continental PM2.5 estimates. It applies the trained baseline and calibrated models to the high-resolution (1km) predictor grids to produce the final GeoTIFF outputs at annual and monthly timesteps. 04_Performance_Evaluation.ipynb: Performs statistical validation against independent ground measurements. It generates the continental and regional performance metrics (Tables 2 and 3) and produces the monthly concentration time series (Figure 3). 05_Spatial_Analysis_&_Mass_Conservation.ipynb: Evaluates the mass conservation of the downscaling framework. It compares aggregated 1-km estimates against the 25-km GEOS-CF truth (Table 4) and generates continental maps of annual mean PM2.5 and error distribution (Figure 4). 06_Predictor_Influence.ipynb: Analyzes variable importance across both model types and all regions. It generates the normalized importance heatmaps (Figure 5) used to interpret the primary environmental drivers of pollution in each region Data Files: Ground_Measurements.zip: Contains monthly aggregated PM2.5 concentrations (2019–2024) from U.S. Embassy monitors across the African continent used for model calibration and validation. Annual_Maps.zip: Final downscaled 0.01° (~1 km) annual mean PM2.5 mass (RH35) GeoTIFFs for all African regions. Monthly_Maps.zip: Final downscaled 0.01° (~1 km) monthly mean PM2.5 mass (RH35) GeoTIFFs for all African regions. All_Monthly_Datasets_25km.zip & All_Annual_Datasets_25km.zip: The regional 0.25° predictor and GEOS-CF target rasters required for model training and mass-balance evaluation Metadata: variable_selection_summary.json: The output generated by the baseline notebook containing the parsimonious list of predictors and performance progression (R², RMSE) for each region. Created by running 01_Baseline_Model.ipynb. performance_evaluation_results.csv: The compiled prediction-observation pairs used to generate the manuscript figures and tables. Created by running 04_Performance_Evaluation.ipynb.

  PublisherDOI

• A Regression Framework for Spatial Downscaling of NASA GEOS-CF PM2.5 Predictions Using Satellite Observations, Spatial covariates, and Ground-Based Measurements

  Zenodo (CERN European Organization for Nuclear Research) · 2026-01-23

  peer-reviewOpen accessSenior author

  This repository contains the datasets and source code required to reproduce the 1-km resolution PM2.5 estimates for Africa as described in the accompanying manuscript: "A Regression Framework for Spatial Downscaling of NASA GEOS-CF PM2.5 Predictions Using Satellite Observations, Spatial Covariates, and Ground-Based Measurements." The project utilizes a Random Forest-based downscaling framework to bridge the gap between global chemical transport models (CTMs) and sparse ground-level monitoring. This framework produces high-resolution surface estimates that resolve intra-urban pollution gradients and align with measured patterns at U.S. embassy locations across the African continent. Contents: Code and Notebooks: 00_GEE_Data_Extraction.ipynb: This notebook executes the multi-source data extraction from the Google Earth Engine Python API. It performs the aggregation of GEOS-CF and environmental predictors into monthly and annual mean rasters at 0.25° (training) and 0.01° (downscaling) resolutions. 01_Baseline_Model.ipynb: This notebook executes the primary downscaling pipeline. It implements the stepwise, importance-driven forward predictor selection for each of the five African regions, followed immediately by the training and 10-fold cross-validation of the uncalibrated (CTM-only) Random Forest and OLS baseline models. 02_Calibrated_Model.ipynb: This notebook implements the final ground-calibration phase. It integrates the spatial scale indicator (Si) and U.S. Embassy ground monitoring data to bias-correct the baseline estimates and produce the final 1-km pollution surfaces. 03_Downscaled_Estimates_ByModel.ipynb: Generates final 1-km continental PM2.5 estimates. It applies the trained baseline and calibrated models to the high-resolution (1km) predictor grids to produce the final GeoTIFF outputs at annual and monthly timesteps. 04_Performance_Evaluation.ipynb: Performs statistical validation against independent ground measurements. It generates the continental and regional performance metrics (Tables 2 and 3) and produces the monthly concentration time series (Figure 3). 05_Spatial_Analysis_&_Mass_Conservation.ipynb: Evaluates the mass conservation of the downscaling framework. It compares aggregated 1-km estimates against the 25-km GEOS-CF truth (Table 4) and generates continental maps of annual mean PM2.5 and error distribution (Figure 4). 06_Predictor_Influence.ipynb: Analyzes variable importance across both model types and all regions. It generates the normalized importance heatmaps (Figure 5) used to interpret the primary environmental drivers of pollution in each region Data Files: Ground_Measurements.zip: Contains monthly aggregated PM2.5 concentrations (2019–2024) from U.S. Embassy monitors across the African continent used for model calibration and validation. Annual_Maps.zip: Final downscaled 0.01° (~1 km) annual mean PM2.5 mass (RH35) GeoTIFFs for all African regions. Monthly_Maps.zip: Final downscaled 0.01° (~1 km) monthly mean PM2.5 mass (RH35) GeoTIFFs for all African regions. All_Monthly_Datasets_25km.zip & All_Annual_Datasets_25km.zip: The regional 0.25° predictor and GEOS-CF target rasters required for model training and mass-balance evaluation Metadata: variable_selection_summary.json: The output generated by the baseline notebook containing the parsimonious list of predictors and performance progression (R², RMSE) for each region. Created by running 01_Baseline_Model.ipynb. performance_evaluation_results.csv: The compiled prediction-observation pairs used to generate the manuscript figures and tables. Created by running 04_Performance_Evaluation.ipynb.

  PublisherDOI

• CLIMATE POLICY, AIR POLLUTION IMPACTS, AND THE DISTRIBUTION OF INCOME

  Climate Change Economics · 2026-03-20

  article

  This paper examines how policies intended to reduce carbon dioxide (CO 2 ) emissions affect air pollution exposure, mortality risk, and monetary benefits across the income distribution in the United States (U.S.). We use an energy system optimization model (ESOM) to translate several climate change mitigation policies into CO 2 -equivalent emission reductions. The ESOM also tracks emissions of three air pollutants: fine particulate matter, sulfur dioxide, and nitrogen oxides. The AP3 model links changes in emissions of local air pollutants to county-level ambient concentrations, exposure, mortality risk, and monetary damages. We present three central results. First, the monetary benefits from reduced air pollution exposure of the climate policies amount to less than 1% of real per capita income. Second, the monetary benefits are progressively distributed. Specifically, counties with a 10% higher real median income level tend to incur between 5% and 6% lower benefits from the carbon tax, the net zero scenario, and the clean electricity standard in 2030. These estimated elasticities are closer to zero in 2040 and 2050. Third, benefits are distributed progressively in the northeast and the western census regions, and regressively in the Midwest. In the southeast, benefits and income are uncorrelated.

  PublisherDOI

• Replication Data for "Power System Costs and Emissions Impacts of Data Center and Cryptocurrency Mining Expansion in the United States"

  Zenodo (CERN European Organization for Nuclear Research) · 2026-04-10

  datasetOpen access

  This dataset contains model inputs and outputs necessary to reproduce one scenario in the analysis presented in the manuscript “Power System Costs and Emissions Impacts of Data Center and Cryptocurrency Mining Expansion in the United States.” The scenario included reflects baseline digital demands, mid gas prices, and the inclusion of IRA subsidies as described in the manuscript. Datasets for other scenarios can be recreated using the references in the manuscript or requested from the corresponding author.

  PublisherDOI

• Electric vehicles and the future of personal transportation

  Edward Elgar Publishing eBooks · 2026-03-17

  book-chapter1st authorCorresponding
  PublisherDOI

• A Regression Framework for Spatial Downscaling of NASA GEOS-CF PM2.5 Predictions Using Satellite Observations, Spatial covariates, and Ground-Based Measurements

  2026-02-06

  articleOpen access

  High-resolution estimates of fine particulate matter (PM2.5) are crucial for exposure assessment and public health, yet in much of the Global South, monitoring is sparse, and global products (≥0.10°) fail to capture intra-urban variability. We develop a regression-based downscaling framework that combines NASA’s GEOS Composition Forecast (GEOS-CF; 0.25°) with dense spatial covariates and satellite observations to produce ~1 km surface PM2.5 estimates across Africa. Downscaling predictors include MODIS aerosol optical depth, Sentinel-5P NO2 columns, vegetation indices, land cover, population density, elevation, and meteorological data. We train separate linear Ordinary Least Squares (OLS) and non-linear Random Forest (RF) models for five African regions, calibrated with U.S. embassy PM2.5 measurements. Ten-fold cross-validation shows RF substantially outperforms OLS (RF R²≈ 0.93–0.96, RMSE ~4–6 µg/m3 vs. OLS R² ≈ 0.33–0.54, RMSE ~10–20 µg/m3) for reproducing GEOS-CF 0.25° PM2.5 Concentrations. Mass-balance evaluation indicates that the spatially downscaled 0.01° estimates conserve the mass of GEOS-CF 0.25° grid cells, demonstrating strong spatial invariance. Against the U.S. embassy reference monitors, the RF-downscaled product yields an RMSE of ~9 µg/m3, a normalized mean error of ~0.17, and an R² of 0.92. The resulting 0.01° PM2.5 estimates demonstrate that the Random Forest regression approach for downscaling can significantly sharpen CTM PM2.5 estimates and align with measured pollution patterns in urban, industrial, and dust-source regions in data-sparse settings. This approach provides granular exposure estimates to support air quality assessment, hotspot identification, and epidemiological and policy applications across Africa and similar regions.

  PublisherOA PDFDOI

• Spatiotemporal trends in cropland phenology in the Nigerian Sahel derived from MODIS NDVI

  Environmental Research Letters · 2026-04-21

  articleOpen accessSenior author

  Abstract Understanding how vegetation phenology is changing across dryland agricultural systems is crucial for predicting the effects of climate change and enhancing food system resilience. Here, we analyze two decades of seasonality changes in croplands across 10 states in Nigeria’s Sahel using MODIS MOD13A2 normalized difference vegetation index (NDVI) collection (2000–2022) and four key metrics: start of season (SOS), end of season (EOS), length of season (LOS), and NDVI amplitude (GSA). We fit mixed-effects models with sample (pixel)- and state-level random effects to quantify temporal trends and spatial variation. We used MANOVA and correlations to assess the multivariate structure and metric interrelationships. Our results reveal a significant delay in SOS (+0.26 d yr −1 ), no consistent temporal trend in EOS, and widespread shortening of growing seasons (LOS: −0.25 d yr −1 ). NDVI amplitude increased slightly (+0.000 36 yr −1 ), showing modest aggregate gains in peak vegetation vigor. Cumulative state-level changes show SOS delays of up to +15 d and LOS reductions exceeding 22 d in some states between 2000 and 2022. Sample-level analyses indicate high within-state heterogeneity, with random effects accounting for approximately 40%–55% of total variance. MANOVA results confirm a significant multivariate shift in seasonality structure, while cross-metric correlations show a strong coupling between delayed SOS and shortened LOS ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mi>r</mml:mi> <mml:mstyle scriptlevel="0"/> <mml:mo>≈</mml:mo> <mml:mstyle scriptlevel="0"/> <mml:mo>−</mml:mo> <mml:mn>0.82</mml:mn> </mml:mrow> </mml:math> ). Taken together, these findings highlight a vegetation system undergoing coherent yet spatially diverse transformation likely driven by climatic change and local land-surface processes. The observed trends have direct implications for rain-fed agriculture, food security, and climate adaptation strategies in northern Nigeria and the wider West African Sahel.

  PublisherDOI

• Power system costs and emissions from data center and cryptocurrency mining expansion in the United States

  Environmental Research Letters · 2026-05-12

  articleOpen accessSenior author

  Abstract Rapid growth in electricity demand from data centers and cryptocurrency mining could significantly alter the trajectory of the United States' power sector. We use an energy system optimization model to evaluate how projected demand through 2030 may influence electricity generation, power infrastructure investment, emissions, and costs under a range of scenarios. We model power sector capacity expansion and dispatch decisions across 26 interconnected power regions, incorporating policy constraints and spatial variation in renewable resources and transmission infrastructure. We find that data center and cryptocurrency demand could increase 2030 power sector CO2 emissions by up to 28% relative to a future with no data center growth, driven by increased generation from natural gas and coal plants. Regional effects vary substantially: coal-fired generation rebounds to meet demand in Northern Virginia, while Texas accommodates growth primarily through natural gas generation. Electricity costs, as measured by demand-weighted locational marginal prices, rise by up to 57% in some regions, with a national average increase of 6% to 29% across the modeled scenarios. Outcomes are highly sensitive to natural gas prices. Lower natural gas prices are associated with lower emissions before considering data center demand (with natural gas displacing coal), but result in large incremental emissions attributed to data centers as coal power utilization increases to meet the increased load. In contrast, higher natural gas prices shift a greater share of new demand toward cost-competitive renewable resources, thereby moderating emissions impacts. Reinstating federal incentives for renewable electricity would dampen both the cost and emissions attributable to data center demand. Distributing new data center and cryptocurrency demand more broadly across the grid reduces regional price spikes, but has little effect on national average electricity costs. Overall, these findings highlight the need for proactive planning and targeted policy to ensure that growing data center electricity demand does not strain efforts to achieve near-term climate and affordability goals.

  PublisherDOI

• State-led climate action can cut emissions at near-federal costs but favors different technologies

  Nature Communications · 2025-05-19 · 7 citations

  articleOpen access

  In the absence of comprehensive federal greenhouse gas mitigation policy, state-led strategies may play a pivotal role, particularly following the 2024 United States presidential election. Using a detailed energy system optimization model, we examine the outcomes of 23 climate-minded states pursuing net-zero emissions targets compared to a federal carbon cap achieving equivalent CO2-eq reductions. Here we show that state-led decarbonization results in distinct technology choices, a 0.7% increase in system costs, and nationwide emissions reduction of 46% — substantial, but insufficient for ambitious climate goals. This pathway relies more on electrification, with 952 terawatt-hours more generation in 2050, reallocating 17.2% of emissions to the power sector. Some regions favor solar, wind, and storage, while direct air capture emerges as critical, particularly in California and the Northeast. Inter-regional trading supports and complicates mitigation efforts, underscoring the need for careful policy design. Overall, our findings highlight how state-led and federal decarbonization approaches can yield differing energy portfolios to achieve similar emissions reductions. In the absence of federal decarbonization, states can drive significant greenhouse gas emissions reductions at comparable costs to federal action, with particular reliance on electrification of end uses and a decarbonized power grid.

  PublisherOA PDFDOI

• Overcoming built data-scarcity in developing cities: Hidden Markov Methods to construct reliable building footprint data across urban climate risk zones

  Environmental Research Infrastructure and Sustainability · 2025-09-25

  articleOpen access

  Abstract Prospective climate risk assessments for climate change adaptation and emergency management rely on reliable, accurate data about the built environment. Yet, urban areas in developing countries are growing rapidly, so data sources and methods that measure urban growth in a timely manner are critical. However, current methods that leverage satellite data and machine learning to produce building footprint datasets are prone to biases correlated with urban risk due to limited training data across different continents and types of urban areas, as well as challenges in interpreting satellite imagery across different urban forms. In this paper, we aim to improve the reliability of building footprint data across urban forms through the integration of limited local data using Hidden Markov Models. We present three key contributions: 1) An urban climate risk assessment framework to evaluate datasets derived from deep machine learning models and satellite imagery across urban forms; 2) A method for processing probabilistic outputs of aggregate building footprint data to account for uncertainty among risk classes; 3) A Hidden Markov Model method to calibrate CNN outputs in post-processing with small local datasets to overcome biases critical to climate risk assessments and downstream management decisions. In a case study of Kigali, Rwanda, we show that Hidden Markov Models calibrated on data from similar Local Climate Zones can improve the MSE of built area percent at a block scale from the current building footprint models at 6.8% down to 2.4%. Furthermore, these models reduce standard deviation in performance of estimation of percent built area across Local Climate Zones from 6.6% to 2.6%, reducing the variability in the reliability of built area estimates in high-risk Local Climate Zones.

  PublisherOA PDFDOI

Recent grants

• RIPS Type 2 Collaborative Research: Water and Electricity Infrastructure in the Southeast (WEIS) - Approaches to Resilient Interdependent Systems under Climate Change

  NSF · $1.7M · 2015–2019

• Consequential Life Cycle Assessment for Novel Energy Technologies: The Case of Energy Storage

  NSF · $300k · 2017–2021

Frequent coauthors

• Constantine Samaras

  University of Patras

  27 shared

• Nathaniel J. Williams

  Rochester Institute of Technology

  23 shared

• H. Scott Matthews

  ORCID

  23 shared

• Bart Nijssen

  17 shared

• Nicholas Z. Muller

  National Bureau of Economic Research

  16 shared

• W. Michael Griffin

  15 shared

• Joe Marriott

  University of Pittsburgh

  14 shared

• Michael Craig

  14 shared

Education

• B.S., Civil and Environmental Engineering

  Florida International University

  2003

• M.S., Civil and Environmental Engineering

  Carnegie Mellon University

  2004

• Ph.D., Civil and Environmental Engineering

  Carnegie Mellon University

  2007

Awards & honors

• Fellow of the Scott Institute for Energy Innovation
• Research affiliate of the Kigali Collaborative Research Cent…
• ELATES fellowship at Drexel University
• Accepted into AAAS fellowship