AbstractUncertainty on carbon fluxes is determined by the uncertainties of ecosystem model structure, data and model parameter uncertainties and the temporal and spatial inaccuracy of the input data retrieval. The objective of this paper is to understand the error propagation and uncertainty of evaporative fraction (EF), soil moisture content (SMC) and water limited net ecosystem productivity (NEP). In this respect, C-Fix and spaceborne remote sensing are used for the ‘Brasschaat’ pixel. A simple model based on error theory and a Monte-Carlo approach are used. Different error scenarios are implemented to assess input uncertainty on EF, SMC and NEP as estimated with C-Fix.The minimum and maximum relative errors on the time averaged EF values from the simple error modelling approach amount 11% and 54%. Applying the Monte-Carlo approach, the minimum and maximum relative errors on EF are 8% and 34%, respectively. The minimum and maximum relative errors on the averaged SMC from the simple error approach are 4% and 18%, respectively. From the Monte-Carlo approach, the minimum and maximum relative errors on SMC are 4% and 12%. The minimum and maximum absolute errors on daily NEP (of 0.22gCm−2d−1) estimated from the simple error approach are 1.28 and 4.51gCm−2d−1. From the Monte-Carlo approach, the minimum and maximum absolute errors are 0.86 and 1.85gCm−2d−1.The simple error modelling and Monte-Carlo approach lead to error estimates of the same order of magnitude, though the error values derived from the Monte-Carlo approach are lower. For ecosystem carbon fluxes, both error assessment approaches lead to large differences. The complexity of the model and hence the correlation between model parameters might be responsible for this.Conclusively, the contribution of the error on soil respiration produces the largest uncertainty on NEP. This carbon flux is the most difficult one to measure and model. Improvement in NEP flux estimation can only be expected when more insight is gained in the process of soil respiration and hence, better (sub-)models are obtained.
On uncertainties in carbon flux modelling and remotely sensed data assimilation: The Brasschaat pixel case
Advances in Space Research ; 41 , 1 ; 20-35
2007-08-16
16 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Data analysis and remotely sensed data
AIAA | 1970
|The analysis of remotely sensed data
NTRS | 1977
Processing and Applications of Remotely Sensed Data
Springer Verlag | 2017
|