COHO Project

  • Field measurements to parameterize and evaluate the SPA model
  • Micrometeorological network deployment

Regional Analyis of CO2 and H2O Exchange Over Heterogenous Terrain

Funding provided by NASA

Investigators: Larry Mahrt, Bev Law, and Mathew Williams


1. Modelling the soil-plant-atmosphere (SPA) continuum
We parameterized and applied the SPA model at each of the five field sites within the study area (3 ponderosa pine sites, a fir site, and a juniper site). We assembled the most complete time series of carbon and water fluxes and stocks for all sites. We used the SPA model to explore the relative importance of ecosystem structure (i.e., LAI) versus local environment (i.e., change in precipitation) in determining environmental controls on productivity across the study area (Schwarz et al. 2004).

2. Improving models of C cycling via data assimilation
For the young pine site we combined C flux and stock data with the C box model (called DALEC) and an ensemble Kalman filter to produce estimates of key rate parameters and a filtered estimate of C cycling (Williams et al. 2005). This paper describes the first application of data assimilation techniques for investigating and filtering eddy covariance and other flux data. In this paper we show, via sensitivity analysis, exactly how assimilating an estimate of photosynthesis – that might be provided indirectly by remotely sensed data – improves a model prediction of net ecosystem exchange.

3. Coupled C and water cycling
In (2), we identified the importance of including correct drought constraints on DALEC. We created a new version of DALEC which couples C-water fluxes to provide simulations of developing summer drought stress and its impacts on productivity. The modelling of C-water linkages is derived from the SPA work undertaken in (1), above. We constructed simple aggregated daily models of gross primary production and evapotranspiration. The aggregated model is capable of describing the complex non-linear response surfaces of these fluxes. We conducted a full C-water data assimilation (DA) exercise across all ponderosa pine sites, fir and juniper sites. The DA provides estimates of critical parameters and rate constants across these different locations.

4. Regional assessment of C stocks and dynamics
The goal was to generate a landscape estimate of carbon dynamics for a region in central Oregon spanning the large precipitation gradient. The estimate will use our novel data assimilation techniques, assimilating earth observation (EO) data into the DALEC model operating in spatial mode. The exercise in (3) generated parameter sets for different aged stands of pine, fir and juniper. Predictions across the region rely on maps of species and stand age, so that the correct parameter sets can be spatially allocated. A connected problem is setting the initial conditions – the C and water stocks in all pools, including soils - and attaching confidence intervals to these estimates. The region ranges from fir through pine to juniper, but is a mix of ages and soil types. The assessment of C dynamics must assimilate the detailed study site data (flux towers), extensive surveys (stand biomass and age data) and the full range of available EO products into the spatial model. We conducted ground truthing in summer 2005 to improve interpretation of MODIS, Landsat and aerial photography data.

Generating the detailed meteorological drivers for the model is also a major task. We assembled met data for 2000, for 61 stations across western and central Oregon, and we applied geostatistical techniques to generate maps of temperature, humidity, precipitation and radiation, with confidence intervals.

This work is continuing with funding from the UK NERC Centre for Terrestrial Carbon Dynamics.


Schwarz, P.A., B. E. Law, M. Williams, J. Irvine, M. Kurpius & D. Moore (2004). Climatic versus biotic constraints on carbon and water fluxes in seasonally drought-affected ponderosa pine ecosystems. Global Biogeochemical Cycles 18, GB4007, doi:10.1029/2004GB002234.

Quaife, T., P. Lewis, M. De Kauwe, M. Williams, B.E. Law, M. Disney, P. Bowyer. 2008. Assimilating canopy reflectance data into an ecosystem model with an ensemble Kalman filter. Remote Sensing of Environment, doi:10.1016/j.rse.2007.05.020.

Williams, M., P.A. Schwarz, B.E. Law, J. Irvine & M. Kurpius (2005). An improved analysis of forest carbon dynamics using data assimilation. Global Change Biology 11, 89-105.