Quantification of carbon (C) budgets at the landscape or catchment scale is generally achieved using process-based models as scaling tools. Such models require some metric of the exchange surface capability (e.g. Leaf Area Index, LAI) and a set of rate parameters for C processing. The net C exchange is then determined by driving the model with meteorological observations. Regional fields of parameters and drivers may be derived by upscaling site level measurements, constrained using Earth Observation data such as vegetation indices and digital elevation models (DEMs). I explore issues of error and uncertainty when upscaling C model parameters and drivers, and the effect of these uncertainties on the final analysis of the carbon budget. Two study areas focus the research: a region of tundra in Arctic Sweden and a ponderosa pine stand in Oregon. I use geostatistical techniques to develop fields of LAI and meteorology, complete with error statistics, whilst the distributions of rate parameters for a C model are derived via the Ensemble Kalman filter (EnKF). I report that the use of DEM data can provide LAI fields with an r<sup>2</sup> ~50% greater than those derived from EO data alone. In particular I find strong relationships between LAI, elevation and topographic exposure. I explore the use of spatio-temporal geostatistics to improve meteorological fields, but report a better interpolation skill when temporal autocorrelations are ignored. Variation in parameters has a much larger effect on the uncertainty of the carbon budge (~50%) than driver uncertainty (~10%). The combined uncertainty in parameterisation and meteorology may result in a 53% uncertainty in total C uptake.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:662298 |
| Date | January 2008 |
| Creators | Spadavecchia, Luke |
| Publisher | University of Edinburgh |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/1842/14462 |
Page generated in 0.0026 seconds