Michael Humber

Michael Humber is an Assistant Research Professor and Instructor who has been at the University of Maryland since Fall 2007. He started research with the Department of Geographical Sciences in Fall 2010 as an undergraduate intern. Initially, his work was focused on photointerpreting wildfire burn scars in Landsat images to create a validation dataset for the MODIS Collection 5.1 MCD45A1 Burned Area product. During this time, he developed the workflows for generating vector files from the MCD45A1 Burned Area product which are distributed to the public and contributed to developing a NASA World Wind Java-based 3-D visualization tool for the MODIS Burned Area and Active Fire products. 

As a graduate research assistant, Michael’s research shifted to identifying the role of semantic overlap in object pair identification for preliminary classifications of very high resolution satellite and aerial imagery. Quantifying the semantic overlap is a necessary step for validating such classification maps in cases when the map and ground truth legend terminologies do not coincide in a 1:1 or 1:M fashion and is also required for identifying meaningful land cover / land use change over time. His master’s capstone centered on developing a Flex-based web application for the Central Africa Regional Program for the Environment (CARPE) known as CARPE Mapper which allowed users to visualize forest cover datasets, interact with the data, and download reports in .pdf or .csv formats.

In 2013, Michael joined the professional track faculty as a Faculty Research Assistant (Faculty Specialist) supporting the Group on Earth Observations Global Agricultural Monitoring (GEOGLAM) initiative and the MODIS Fire projects. He worked on the initial development of the GEOGLAM Crop Monitor’s Crop Assessment Tool, which was awarded a “Special Achievement in GIS” by Esri in 2014. The Crop Monitor has since expanded to nine separate implementations with dozens of partner organizations at the global, region, and national scales including its two most visible implementations – Crop Monitor for AMIS and Crop Monitor for Early Warning. As an Assistant Research Professor, Michael leads technical development of the Crop Monitors and works with Dr. Louis Giglio to maintain and distribute GIS-ready products derived from the MODIS Collection 6 MCD64A1 Burned Area products. His research focuses on validation of burned area and cropland maps using both pixel-based and object-based approaches.