Session Description: This short course introduces participants to fundamental concepts and techniques related to remote sensing image processing in the R Studio programming environment. The fundamental skills learned in this course are directly applicable to other areas of ecological research, including those not focused on the course's case study. Applicable skills learned include: how to access publicly available remote sensing imagery; writing algorithms to process and analyze the imagery; integrating other spatially explicit data with remote sensing imagery; and implementing deep learning models to classify important features of a study area. This course is also designed to help participants overcome some of the challenges posed by barriers to computationally intensive processes by using a free cloud computing platform, Google Colaboratory (Python). This platform allows users to build deep learning models on supercomputers at home and for free.
For the case study, participants will work with Light Detection and Ranging (LiDAR) and very high resolution (VHR) imagery to calculate Normalized Difference Vegetation Index (NDVI), canopy height models, and segment individual trees and tree crowns from a complex environment. Participants will also engage with deep learning integration into ecological research through finetuning an existing fully convolutional neural network in Python for tree species classification.
The intended audience is graduate students and early-career ecologists who are interested in learning how to use some basic geospatial programming in their research. Participants need not know the basics of R or Python in order to effectively participate in this course, though some background in either will be helpful.