Satellite observations may help to improve nitrogen deposition estimates from models. In this thesis, we aim to improve modelled nitrogen deposition estimates from the LOTOS-EUROS chemical transport model by integration and assimilation of different types of satellite measurements. Several approaches to use atmospheric satellite observations of NH3 concentrations and satellite-derived land surface parameters (leaf-area-index, roughness length) for nitrogen deposition modelling are presented. In the first study, presented in Chapter 2, NH3 total column observations from the IASI satellite instrument are used in combination with information from LOTOS-EUROS to compute satellite-derived NH3 surface concentration and dry deposition fields. The IASI-derived NH3 surface concentrations were used to identify regions with systematic over-, or underestimations in modelled NH3 concentrations, indicating potential errors in the current emission inventory for NH3. The comparison with available in-situ observations in Europe, however, showed no significant or consistent improvement in the IASI-derived concentrations compared to the originally modelled concentrations from LOTOS-EUROS. Here, our study illustrated that the application of this method might not be viable at this time, particularly due to the relatively low near-surface sensitivity of the current NH3 observing instruments and the uncertainty in surface-atmosphere exchange schemes for NH3. In the second study, presented in Chapter 3, several satellite products were used to derive more realistic, dynamic input values for the roughness length (z0) and the leaf-area-index (LAI) values. The satellite-derived z0 values were validated with z0 values at FLUXNET sites, showing an overall improvement compared to the default values used in LOTOS-EUROS. The z0 and LAI values were integrated in LOTOS-EUROS for the computation of deposition fluxes, instead of the fixed and land-use specific default values. Compared to the default model runs, this led to distinct changes in the modelled total Nr deposition of up to ∼30 % and an overall shift from wet to dry deposition. Our results illustrated that the Nr deposition fields were especially sensitive to changes in the LAI input values. The changes for land use specific deposition fluxes were even greater, with particularly large changes in the modelled deposition fluxes over coniferous and deciduous forests. In the third study, presented in Chapter 4, CrIS-NH3 satellite observations were integrated into LOTOS-EUROS in two different ways. In the first method, the NH3 surface concentrations from CrIS were used to fit spatially variant NH3 emission time factors. This method proved to be especially useful in agricultural-intensive regions during spring, where it can be successfully used to estimate the onset and duration of the NH3 spring peak. In the second method, the NH3 input emissions were refined in a top-down approach, using a Local Ensemble Transform Kalman Filter (LETKF) as data assimilation algorithm. The strength of this method primarily lies in fine-tuning existing NH3 emission patterns, and in that way improving the spatial distribution of the modelled NH3 fields. Both these methods, especially when combined, led to distinct improvements in the comparison with in-situ observations and showed strong potential to improve the NH3 input emissions, and herewith modelled nitrogen deposition. Lastly, in the fourth study, presented in Chapter 5, long-term, land-use specific, modelled nitrogen deposition estimates from LOTOS-EUROS were used to examine the interaction of nitrogen deposition and drought as co-stressors on gross primary production (GPP) at European FLUXNET forest sites. Our results showed that nitrogen deposition is a clear driver of GPP in forests. However, due to the differential response of various dominant forest types to drought and the limited variation in nitrogen deposition levels found at the FLUXNET sites, no consistent positive nor negative nitrogen deposition effects on the drought response of GPP could be isolated.
|12 Jan 2022
|Published - 12 Jan 2022