Seminar: Kseniia Ivanova

We investigate the drivers of Arctic carbon exchange by integrating chamber measurements and renotely sensed data. We evaluate how CH₄ and CO₂ fluxes respond differently to local microtopography versus regional climate across nested scales. To optimize landscape scale budgets, we test whether machine learning can independently capture tundra mosaicism or if accuracy requires models stratified by functional types. Our results demonstrate that explicitly accounting for local wet-dry gradients is essential to reduce systematic biases in upscaling carbon estimates across permafrost landscapes.