Skip to main content
2023 ESA Annual Meeting Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Back

Contributed Poster

Session: Modeling

PS 32-181 - Iterative near-term ecological forecasts provide new insight into the predictability of multiple water quality variables over time and space

Tuesday, August 8, 2023
5:00 PM – 6:30 PM PDT
Location: ESA Exhibit Hall
Abstract: Near-term iterative ecological forecasting has great potential for providing new insights into our understanding of environmental predictability. Predictability likely varies across time and space; however, these patterns have largely been unexamined across multiple ecosystem variables. To explore the fundamental limits of predictability for two important freshwater ecosystem variables, we produced forecasts of water temperature and dissolved oxygen (DO) at multiple depths and over two open-water seasons in Lake Sunapee, NH, USA. Our forecasting system, FLARE, uses a 1-D hydrodynamic model and daily data assimilation to update initial conditions of temperature and DO and fit parameters over time. FLARE also calculates the contribution of several sources of uncertainty to produce probabilistic forecasts of these variables up to 35 days into the future. Overall, we observed that the limits of predictability (assessed as forecast skill) varied substantially among ecological variables, depth, and time into the future. We found that forecast skill for both temperature and DO was similar between years with mean continuous ranked probability score (CRPS) of 0.5-0.6 mg/L for oxygen and 0.9-1.1°C for temperature across both years. However, the range in forecast skill across days was more variable deeper in the water column for both variables. Specifically, we found that from 1-day-ahead to 35-day-ahead forecast horizons, skill in temperature forecasts decreased much more rapidly than oxygen (slope = 0.033 and 0.009, respectively) at 1 m lake depth, but that forecast skill decreased similarly into the future at 10 m for both variables (slope = 0.05 and 0.03, respectively). In addition, forecast skill was highly variable over the open-water period indicating that external environmental variables which changed from year to year were important in driving forecast skill (e.g., weather, hydrology). Finally, our data show that including multiple sources of uncertainty (e.g., initial condition, weather, model process, parameter) influenced forecast skill, with forecast skill decreasing substantially when uncertainty was not included. Overall, this work provides important insight into the predictability of multiple ecological variables over time and space, laying the groundwork for understanding the limits of predictability for managing critical water quality resources.