Depicting CH4 fluxes and drivers dynamics
Abstract
Since the advancement in CH4 eddy covariance flux measurements, monitoring of CH4 emissions is becoming more widespread. Since CH4 fluxes are not as predictable or as easily interpretable as CO2 fluxes, understanding their emission patterns often still challenging. As these are spatially (ecosystem and latitudinal) and temporal very divers and often event based, a better understanding or interpretation of results is required. An improvement in understanding does also increase the reliability of gap-filling methods as annual greenhouse gas budgets rely on high quality data. There are generalised additive models (Wood 2001) that can easily be applied to sites, models where a relationship between the response variable, in this case CH4 and explanatory variables (drivers) is established. Relevant for CH4flux dynamics are the smoothing function that is applied, where each predictor variable is separated into sections and a polynomial function fitted. On the one hand such models are rarely used as they are difficult to interpret since no parameter values are retuned. On the other hand, such models are very good for prediction and explanatory analysis in estimating the functional nature of a response. Applying such models to CH4 eddy flux data does improve our understanding of the dynamics of CH4 emissions and the respective meteorological drivers. Furthermore, such models combined with tree models (interactions between the explanatory variables), can visualise precise dynamics and easily applied to individual sites. These models are simple tools in understanding of these complex fluxes, as they can include a variety of drivers, and their relevance tested by the model. Model input variables should be as independent as possible (avoiding cross-correlation), avoiding redundant inputs, as models should follow the principle of parsimony of being simple but not too simple. Wood SN (2001). mgcv: GAMs and generalized ridge regression for R. R news.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFM.B41B0423D
- Keywords:
-
- 0414 Biogeochemical cycles;
- processes;
- and modeling;
- BIOGEOSCIENCESDE: 0426 Biosphere/atmosphere interactions;
- BIOGEOSCIENCESDE: 0428 Carbon cycling;
- BIOGEOSCIENCESDE: 0452 Instruments and techniques;
- BIOGEOSCIENCES