2022 ESA Annual Meeting (August 14 - 19)

OOS 7-6 CANCELLED - Near-term agricultural forecasts of carbon dynamics to improve carbon sequestration and decision support at Field Observatory Network (FiON) sites

9:15 AM-9:30 AM
520C
Istem Fer, Finnish Meteorological Institute;Olli Nevalainen,Finnish Meteorological Institute;Henri Kajasilta,Finnish Meteorological Institute;Laura Heimsch,Finnish Meteorological Institute;Henriikka Vekuri,Finnish Meteorological Institute;Åsa Stam,Finnish Meteorological Institute;Stephanie Gerin,Finnish Meteorological Institute;Toni Viskari,Finnish Meteorological Institute;Julius Vira,Finnish Meteorological Institute;Juha-Pekka Tuovinen,Finnish Meteorological Institute;Tuomas Laurila,Finnish Meteorological Institute;Annalea Lohila,Finnish Meteorological Institute;Liisa Kulmala,Finnish Meteorological Institute;Jari Liski,Finnish Meteorological Institute;
Background/Question/Methods

Agricultural soils constitute a large carbon pool that has been depleted since the beginning of agricultural cultivation and proposed to have the potential to be (partially) refilled following carbon farming practices (Paustian et al., 2019). However, rates of change in agricultural soil carbon due to carbon farming are hard to detect and quantify with respect to the large background stock especially considering the large heterogeneity and variability in the soil ecosystems. Furthermore, implementing the right practices at the right situations in day-to-day operations proved to be challenging.Towards addressing these challenges, a network of benchmark fields is being monitored and modeled as part of the Field Observatory Network (FiON) (Nevalainen et al., 2022). Within FiON and its online web service Field Observatory (fieldobservatory.org), we operationalized a near-term iterative forecasting system based on the Predictive Ecosystem Analyzer (pecanproject.org) model-data integration cyberinfrastructure (Fer et al., 2020) where gap-filled fluxes are integrated to daily values and assimilated into agro-ecosystem models jointly with satellite derived observations. With this approach, we aim to improve i) predictive carbon science by studying variability patterns in space and time and ii) provide decision support on near term events that can potentially have long-term effects on carbon storage.

Results/Conclusions

Models that were initialized and calibrated with the pre-forecasting period site-level data had improved predictions in terms of both uncertainty reduction and accuracy. From Spring 2021 onwards, every day a 15-day 250 member ensemble forecast is made from the agro-ecosystem models, mainly assimilating the gap-filled daily CO2 fluxes and satellite derived LAI estimations. Related carbon states were also updated within the analysis step as relations among different ecosystem processes are encoded in the models. The weather forecasting data that were used to drive models at each time step, as well as the forecast and analysis states were archived. Post-hoc analyses of the results pointed out underlying sources driving variability. While 15-day forecasts provided limited decision support within a cropping cycle, post-hoc experiments concerning field activities that may have 1–2 weeks of flexibility such as harvest and fertilization were found informative and potentially impactful on carbon balance in the mid-term. Overall, our operative iterative near-term forecasting system presents a framework to explore the impacts of short-term interventions dynamically, systematically, and quantitatively and in return devise more reliable and comprehensive decision criteria with potential to include seasonal, annual, and longer-term forecasts in the future.