Postdoctoral Research Associate I

We seek a post-doctoral researcher to support machine learning for flood detection with satellite imagery. This person will join the Social Pixel lab and support Dr. Tellman and the lab to improve on existing models with new methods and develop models on new satellite sensors for water detection. This position is funded through two existing NASA grants but with the goal of developing new projects.

The first NASA funded project Operationalizing VIIRS for water mapping, and river width evaluation for SWOT reaches with Planetscope and Sentinel-1 uses Planetscope, VIIRS, MODIS, Sentinel-1/2, and deep learning models to detect water for automated detection of river reaches. We have already developed labeling and sampling datasets for this project, and the postdoc will work on building machine learning models of high accuracy. The second NASA funded project Understanding Flood Risk in Human Altered Landscapes from cities to farms: inferences from satellites and machine learning focuses on flood mapping in Bangladesh. We have already developed a 20-year time series fusing MODIS and Sentinel-1 (Giezendanner et al 2023) and are working on a VIIRS extension. The postdoc will support the lab in leveraging this data to simulate per-pixel estimates of flood inundation return periods from observations and elevation data and compare results to flood models produced by the Bangladesh government.

The post doc will work with existing research assistants and post docs to deploy and improve deep learning models of inundation. We anticipate that as these two projects conclude, the postdoc will either work on newly funded projects, which may include building a multi-sensor database of historical floods for the USA or innovating on our existing CNN (Convolutional Neural Network) models across 6 sensors using new machine learning techniques such as transformers. Given that we have built up several large hand labeled datasets across sensors for flooding over the last several years, the post doc will have an opportunity to quickly run experiments and write papers with our datasets. The Social Pixel lab has a strong emphasis on applications for social and environmental justice, and you will have the opportunity to work with actors outside academia in government, business, and non-profits we partner with.

There is some flexibility in the FTE for this position, which will be discussed at time of interview.

The University of Arizona has been recognized for our innovative work-life programs. For more information about working at the University of Arizona and relocations services, please click here .

Job Duties:

• Collaborate with PI, and students to deploy deep learning models to identify and monitor water in satellite imagery time series using sensors such as MODIS, VIIRS, Sentinel-1, and Planetscope.
• Writing articles for publication (co-authored and first-authored).
• Support organization of meetings, collaboration, and lead NASA funded projects related to flood mapping in Bangladesh and near real time flood mapping using VIIRS.
• Present results at AGU, AAG, or other relevant conferences.
• Support grant proposal writing efforts and developing proof of concepts (including opportunities to lead ones’ own).
• Collaborate with partners at NASA to implement flood water detection models in satellite imagery.

Additional Information:

• Researcher will gain skills in fundraising (grant writing), mentoring, and research administration essential for future faculty, government, and industry research positions, with hands on mentoring from the PI to help you achieve your career goals. The researcher will also be given at least 20% of their time to devote to personal projects, publications, and funding proposals. The University of Arizona is a Hispanic-Serving Institution, and the Social Pixel lab is committed to diversity, inclusion, and social justice inside the lab and in our research practices- see more- https://beth-tellman.github.io/Empowerment%20Vision%20and%20Operation%20(1).pdf

• PhD in Geography, Environmental Sciences, Computer Science, or related field upon hire.
• Experience in Remote Sensing and GIS.
• Experience in Machine Learning preferably using Python.
• At least one first authored publication in a referred journal.

• PhD in Geography or Environmental Science.
• Experience programming in Google Earth Engine.
• Experience with data fusion (radar, optical imagery, of various spatial and temporal resolutions).
• Experience working with hydrological data (such as stage/discharge measurements, surface water measurements).
• Experience in training deep learning models, such as convolutional neural networks, using Pytorch or Fastai.
• Experience in data engineering- curating and maintaining labeled training datasets for deep learning.
• Experience with transformers, LSTMs, and other deep learning modalities.