Roothaert, Ritten (2022) Improving Data Assimilation Approach for Estimating CO2 Surface Fluxes Using ML. Master's Thesis / Essay, Artificial Intelligence.
|
Text
mAI_2022_RoothaertHM.pdf Download (2MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (170kB) |
Abstract
The environment is changing due to anthropogenic carbon emissions, and so is the carbon cycle regulating the exchange of CO2 (i.e. fluxes) between the Earth’s surface and the atmosphere. Measuring these changes is difficult, as it would require enormously dense observation networks to capture the strongly heterogeneous underlying flux-landscape. Through a combination of carbon exchange (CE) models and data assimilation (DA), the CarbonTracker data assimilation shell (CTDAS) generates a flux-landscape estimate which optimally matches the available observations. The current implementation of this DA approach is static; flux-landscape estimates produced in the past are not used for estimating new flux-landscapes. However, preliminary research has shown that seasonal, currently unused, patterns are present within the estimates of the DA approach. We propose three different methods for utilizing these patterns: a simple monthly mean model, a seasonal autoregressive integrated moving average (SARIMA) model, and a seasonal autoregressive integrated moving average with exogenous factors (SARIMAX) model. Preliminary results strongly indicate that the monthly mean model provides a substantial improvement over the current DA implementation once incorporated within CTDAS. In contrast, the SARIMA and SARIMAX models struggle to capture the non-stationary seasonal patterns.
| Item Type: | Thesis (Master's Thesis / Essay) |
|---|---|
| Supervisor name: | Lawrence, C.P. and Peters, W. and Woude, A.M. van der |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Master's Thesis / Essay |
| Language: | English |
| Date Deposited: | 29 Nov 2022 15:11 |
| Last Modified: | 29 Nov 2022 15:11 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/28997 |
Actions (login required)
 |
View Item |