Transferring Deep Learning Knowledge for Scaling Up Hydrographic Feature Extraction

The U.S. Geological Survey (USGS) 3D Elevation Program (3DEP) provides high accuracy, high-resolution (HR) elevation data for the United States. The USGS has also been coordinating efforts to derive hydrography from high-resolution 3DEP elevation data, including interferometric synthetic aperture radar (IfSAR) data in Alaska, and lidar data in the conterminous United States.

Deriving hydrography from elevation through traditional flow routing and interactive methods is a complex, time-consuming process that must be tailored for different geographic and hydrologic conditions. The large volume of surface water features and HR remote sensing data make manual annotation of the water features over the entire nation infeasible. Furthermore, annual and seasonal variations of surface waters warrant some level of periodic updates to hydrography.

Advances in deep learning technologies provide an opportunity to automate hydrography extraction and scale up the process. One major challenge, however, is the effect of spatial heterogeneity due to varying geographic conditions. In other words, it is unclear how a deep learning model pretrained in one geographic region can be effectively applied to other regions for hydrographic feature extraction.

Research presented here aims to fill the gap by testing automated deep learning for extracting hydrography from digital elevation model (DEM) data and its transferability to a range of geographic conditions in Alaska. In transfer learning, the knowledge (e.g., neural network weights) from one domain is transferred to other domains to help reduce training requirements. This work examines whether applying transfer learning methods can improve machine learning predictions for Alaska. If time permits transfer learning research to extract hydrography from lidar in a conterminous US study area will be presented. 

Agenda

• 00:00 – Start 
• 1:44 –Transferring Deep Learning Models for Hydrographic Feature Extraction from IfSAR Data in Alaska presentation begins 
• 22:28 Transfer Learning with Convolutional Neural Networks for Hydrological Streamline Delineation presentation begins 
• 38:24 - Presentation ends/Conclusion