[Paper Review] Mapping bathymetry of inland water bodies on the North Slope of Alaska with Landsat using Random Forest

Mark Carroll, Margaret Wooten+7 more|ArXiv.org|Feb 14, 2025

Climate change and permafrost3 citations

TL;DR

The paper trains a Random Forest Regressor to predict waterbody depth from Landsat multispectral data, using synthetic training data to overcome scarce in situ measurements, and produces a per-pixel depth map for the North Slope of Alaska with r^2 = 0.76 on validation.

ABSTRACT

The North Slope of Alaska is dominated by small waterbodies that provide critical ecosystem services for local population and wildlife. Detailed information on the depth of the waterbodies is scarce due to the challenges with collecting such information. In this work we have trained a machine learning (Random Forest Regressor) model to predict depth from multispectral Landsat data in waterbodies across the North Slope of Alaska. The greatest challenge is the scarcity of in situ data, which is expensive and difficult to obtain, to train the model. We overcame this challenge by using modeled depth predictions from a prior study as synthetic training data to provide a more diverse training data pool for the Random Forest. The final Random Forest model was more robust than models trained directly on the in situ data and when applied to 208 Landsat 8 scenes from 2016 to 2018 yielded a map with an overall $r^{2}$ value of 0.76 on validation. The final map has been made available through the Oak Ridge National Laboratory Distribute Active Archive Center (ORNL-DAAC). This map represents a first of its kind regional assessment of waterbody depth with per pixel estimates of depth for the entire North Slope of Alaska.

Motivation & Objective

• Address scarcity of in situ depth measurements for North Slope waterbodies.
• Develop a machine learning approach to estimate bathymetry from Landsat images.
• Create a regional, per-pixel depth map for the North Slope waterbodies.
• Evaluate model robustness with synthetic training data to supplement limited observations.

Proposed method

• Train a Random Forest Regressor to predict depth from multispectral Landsat data.
• Use modeled depth predictions from a prior study as synthetic training data to enrich the training set.
• Validate the model on Landsat 8 scenes from 2016 to 2018 and report per-pixel depth estimates.
• Publish the resulting depth map via the ORNL-DAAC data center for broader use.

Experimental results

Research questions

• RQ1 Can Random Forest depth predictions from Landsat imagery accurately estimate bathymetry for small inland waterbodies on the North Slope of Alaska?
• RQ2 Does incorporating synthetic training data improve model robustness and predictive performance compared to using in situ data alone?
• RQ3 What is the per-pixel depth mapping performance (validation r^2) across a regional extent and multiple scenes?
• RQ4 Is the resulting bathymetric map suitable for open data dissemination and regional assessment?

Key findings

• The final Random Forest model achieved an r^2 of 0.76 on validation across 208 Landsat 8 scenes from 2016–2018.
• Synthetic training data derived from a prior study helped enhance model robustness over using only in situ data.
• The resulting bathymetric map provides per-pixel depth estimates for the entire North Slope of Alaska.
• The depth map has been made available through the ORNL-DAAC data center for broader use.