Saurabh Kaushik

Postdoctoral Research Associate, University of Wisconsin–Madison. Geospatial Data Science · Earth Observation · Foundation Models.

I am a Postdoctoral Research Associate at the University of Wisconsin–Madison, working with the Tellman Lab (PI: Beth Tellman) and the Schneider Lab (PI: Annemarie Schneider). My research sits at the intersection of GeoAI, Earth observation, and geospatial foundation models (GFMs) — with applications spanning flood inundation mapping, glacier and glacial-lake monitoring, sea-ice classification, calving-front delineation, and large-scale satellite data pipelines.

Previously, I was a Postdoctoral Researcher at The Ohio State University, where I developed Earth observation methods for global mapping of natural (gold) hydrogen sites using multisource remote sensing and deep learning. I earned my bi-nationally supervised PhD from the Academy of Scientific and Innovative Research and the German Aerospace Center (DLR), Germany, focusing on deep-learning frameworks for mapping Himalayan glaciers and glacial lakes, modeling glacial-lake outburst floods, and estimating long-term glacier velocities.

My broader research agenda is to build robust, scalable Earth observation systems that convert petabytes of multi-sensor satellite imagery — Sentinel-1/2, Landsat, NISAR, PlanetScope — into reliable long-term datasets for assessing risks in water resources, the cryosphere, and climate adaptation. I am actively pursuing faculty positions and senior research scientist roles.

I also contribute to teaching at UW–Madison, including course material for “AI for Sustainability” and “AI for Earth Observation.”

selected news

Feb 2026 Our latest study is now available on IEEE Xplore: official version here.
Jan 2026 Workshop paper at WACV (CV4EO) introducing Prithvi-CAFE for flood inundation mapping — arXiv:2601.02315.
Jan 2026 Workshop paper at WACV (GeoCV) presenting GLACIA — a multimodal LLM framework for glacial-lake segmentation with positional reasoning. Link.
Nov 2025 Preprint: Cryo-Bench — benchmarking geo-foundation models for the cryosphere across glacial lakes, glaciers, sea ice, and calving fronts. arXiv.
Feb 2025 Article in Water Resources Research on declining groundwater storage in the Indus Basin. DOI.
Sep 2024 New article in Remote Sensing of Environment — most updated glacial-lake inventory across High-Mountain Asia using deep learning. DOI.

selected publications

  1. arXiv 2026
    arXiv 2026

    Cryo-Bench: Benchmarking Geospatial Foundation Models for Cryosphere Applications

    Saurabh Kaushik, Lalit Maurya, and Beth Tellman

    arXiv preprint, 2026

  2. WACV 2026
    WACV 2026

    Prithvi-CAFE: A Convolutional Attention Fusion Encoder for Flood Inundation Mapping

    Saurabh Kaushik, et al.

    In WACV CV4EO Workshop, 2026

  3. WACV 2026
    WACV 2026

    GLACIA: Instance-Aware Positional Reasoning for Glacial-Lake Segmentation with Multimodal LLMs

    co-authored with Saurabh Kaushik, et al.

    In WACV GeoCV Workshop, 2026

  4. RSE 2024
    RSE

    Updated Glacial-Lake Inventory of High-Mountain Asia using Deep Learning

    Saurabh Kaushik, et al.

    Remote Sensing of Environment, 2024

  5. WRR 2025

See all publications →