Team 10 Home: Generating Finer Cropland Soil Moisture Content Maps Using Earth Observation Embeddings

Team hero image

Team Photo

Team photo

People

Name	Affiliation	Contact	Github
Yi Yang	Colorado State University	yi.yang@colostate.edu	@y1y9ng
Aashish Gautam	Jackson State University	aashish.gautam@students.jsums.edu	@aashish66
Mariella Carbajal Carrasco	North Carolina State University	mcarbaj@ncsu.edu	@carbajalmariella
Alice Heiman	Stanford University	aheiman@stanford.edu	@aliceheiman
Amos Abdulai	Livingstone College	abdulaiamos716@outlook.com	aabdulai116
Mohammad Shahriar Saif	Colorado State University	ms.saif@colostate.edu	@saif8091
Yu Peng	Indiana University	yp24@iu.edu	@eco-yupeng
Shashi Konduri	NEON/Battelle	konduri@battelleecology.org	@Shashi-konduri
Johnie plentybull	jplenty28637@olc.edu
Tatiana Acero-Cuellar	University of Delaware	taceroc@udel.edu	taceroc

Our question(s) 📣

Our working question:

Can Earth Observation Embeddings estimate soil moisture content at higher spatial and temporal resolutions than traditional ML/RS approaches?
Can Earth Observation Embeddings also help improve the prediction of soil mositrure in deeper depths

Our final product:

Data Product: Higher spatial and temporal resolution soil moisture map of agricultural areas in California. (starting 2017-2025)
Academic Product: Paper
Academic Product: a new framework that can gap fill fine temporal and spatial resolution soil mositure with existing field soil moisture measurements

What would count as progress:

Specific question
Roadmap and timeline for future work
Potentially trying to produce some initial maps with Alpha Earth Foundations model

Hypotheses/Intentions

Our hypotheses is that: earth embeddings (which harmonize many different remote sensing data sources) could help us produce higher resolution soil moisture content maps

Study Area (Soil moisture Station Density Map)

Exploration figure

Why this matters

This matters because:

For food security, we need to
High-value crops like grapes and corn are important for nutrition and agricultral export
Soil moisture information allow farmers and state-level officials to make more proactive management strategies, for instance in irrigation and drought-preparedness
California contains the Central Valley, one of the most productive agricultural regions in the US

People who could use this:

Farmers, land-managements, state-level agriculture officials, food- and beverage industry

Promising data sources:

Data source 1: SMAP L4 Global:https://nsidc.org/data/spl4smgp/versions/7
Data source 2: SMOS: https://earth.esa.int/eogateway/missions/smos
Data source 3: USGS In-Situ Soil Moisture sensor network for validation
Data source 4: 30m Crop LULC Regions
Data source 5: Alpha Earth Embeddings (which includes bands (C&L-bands) which are sensitive to soil moisture)
Data source 6: Terra Torch Prithvi, Clay, and Terra Mind earth embedding models