2026 is under way, which means the annual Sevilleta LTER All-hands meeting just took place at the comfortable UNM field station near la Jolla, New Mexico, about an hour south of Albuquerque. As always, exciting dryland science was shared in talks, a field trip, and a poster session; fresh collaborations were sparked, and old friends hand a chance to mingle; and dinners were chile-smothered.
Jacob attended the meeting and presented a poster on a new PhenoCam data processing workflow he is building, which uses the GRIME AI open source software to separate vegetation from bare soil in hundreds of thousands of PhenoCam images from the SEV Mean-Variance Experiment. The questions Jacob is investigating are: 1) does processing images using dynamic masks that separate vegetation and soil improve results compared to static masks that include both vegetation and soil? 2) how well does an AI-based image segmentation model trained on data from one study extrapolate to other study sites?
Jacob’s poster was a hit, and he spent so long answering questions that he almost missed dinner. He is already looking forward to the 2027 All-Hands Meeting!
