Silvia Caldararu

PhD student

Project

Understanding and predicting biogenic isoprene emissions at global scales

Supervisors

Dr. Paul Palmer, Edinburgh University

Dr. Drew Purves, Microsoft Research Cambridge

Research group

The Tropospheric Chemistry and Earth Observation Modelling Group

Project details

Isoprene, a highly chemically reactive hydrocarbon emitted by terrestrial vegetation, plays important direct and indirect roles in (1) the production of tropospheric ozone, a pollutant and greenhouse gas comparable with methane, and (2) the formation of organic aerosol. Consequently, isoprene plays a significant role in climate. Available evidence suggests that the magnitude and geographic pattern of global isoprene emission rates will respond to future changes in CO2 and climate. Emission rates per unit leaf area are highly sensitive to incident radiation, temperature, and CO2, and vary by orders of magnitude among different plant functional types (e.g. grasses vs trees) and among species within a functional type (e.g. different species of deciduous tree), the spatial distributions of which are also expected to respond to change in climate.

The only way to predict future changes in isoprene emission rates is to use bottom-up isoprene emission models, which scale up from sparse leaf-level and canopy flux observations to emissions at spatial scale of 10-100s km. Scaling up these sparse measurements relies on making assumptions about poorly understood physiological and biomechanical processes, e.g., the effect of leaf age and recent weather on emission rates. These assumptions result in large uncertainties in the magnitude and spatial distributions of isoprene emissions, compromising model predictions of future isoprene fluxes.

Estimates of regional isoprene emissions, inferred from satellite observations, are available for most of the globe at relatively fine temporal resolution (~100 km). Canopy-level isoprene emission data are also becoming increasingly available. Together, these data offer a potentially powerful way to test and improve bottom-up emission models, and associated model parameters. Significant discrepancies between emissions models and satellite data already identified are leading to better understanding of the underlying physical processes. However, to date, model improvement using the satellite data has been ad hoc. A better approach to model improvement is to use a Bayesian approach to formally estimate bottom-up emission model parameters using all the available leaflevel, canopy-scale, and space-borne data. The student will parameterize a suite of bottom-up isoprene emission models, and formally compare the predictive ability of each of these models using model selection criteria. The three key results will be: (1) a formal test of a variety of hypotheses about the processes governing isoprene emissions at regional to global scales; (2) one or more new parameterized isoprene emission models, which are consistent with all known data; (3) an estimate of uncertainty, in both model formulation and parameter values, which can be propagated through to isoprene emission estimates for the current climate, and for future climate scenarios.