Advancing Multimodal Earth Observation Foundation Models Wildfire Risk and Vegetation Recovery Assessments

Organization

National Aeronautics and Space Administration (NASA)

0346-NPP-NOV
26-JPL-Earth Sci

11/1/2026 6:00:59 PM Eastern Time Zone

The NASA Postdoctoral Program (NPP) offers unique research opportunities to highly-talented scientists to engage in ongoing NASA research projects at a NASA Center, NASA Headquarters, or at a NASA‑affiliated research institute. These one- to three-year fellowships are competitive and are designed to advance NASA’s missions in space science, Earth science, aeronautics, space operations, exploration systems, and astrobiology.

Under the guidance of a mentor, the participant will engage in a NASA‑aligned research project advancing data‑driven approaches to characterize wildfire risk and post‑fire vegetation recovery across the western United States. Initially focused on California’s ecosystems, this research investigates how structural vegetation interacts with climatic stressors, such as drought and prolonged dry conditions. The overarching objective of the project is to develop an extensible, scalable framework that integrates structural, radar, spectral, and temporal information to continually update estimates of vegetation vulnerability and fire response.

During the appointment, the participant will have the opportunity to collaborate on developing structural vegetation representations from LiDAR observations, including GEDI‑derived canopy height, vertical structure, and above ground biomass products. Through this research, the participant will learn to establish metrics for vegetation condition, pre‑fire fuel environments, wildfire‑related structural change, and post‑fire recovery.

The participant will also research multimodal data fusion, collecting and analyzing data to explore how LiDAR‑based observations can inform representations and encoder adaptations. This activity will provide hands‑on experience in making three‑dimensional vegetation structure accessible within data‑driven geospatial modeling processes.

Additionally, the participant will participate in evaluating L‑band SAR, particularly emerging observations from the NASA‑ISRO Synthetic Aperture Radar (NISAR) mission, as a scalable source of information for vegetation structure, biomass, disturbance, and recovery. By modeling the relationships between LiDAR‑derived metrics and NISAR observations, the participant will investigate whether SAR can accurately extrapolate LiDAR‑constrained structural data across broader spatial and temporal scales.

The participant will collaborate with researchers to evaluate encoder adaptations that incorporate both synthetic aperture radar and LiDAR‑derived vegetation characterizations into multimodal geospatial learning pipelines. These capabilities will be applied to Earth observation foundation models. The participant will analyze models such as NASA’s Prithvi, alongside other relevant foundation‑model architectures and fine‑tuning strategies.

For initial model development and fine‑tuning, the participant will learn to utilize existing multimodal biomass testbeds containing LiDAR‑derived above ground biomass reference measurements paired with Sentinel‑1 SAR and Sentinel‑2 multispectral time series. Subsequently, the participant will explore adapting these models for application in California using GEDI products, NISAR L‑band SAR, optical Earth observations, climatic indicators, and fire‑severity datasets.

Through these activities, the participant will gain valuable experience in illuminating the critical relationships among pre‑fire vegetation structure, climatic stress, and observed burn severity, while using post‑fire observations to quantify structural loss and recovery. By participating in the integration…