From emissions to source allocation: Synergies and trade-offs between top-down and bottom-up information

Calibrating the CAMS European multi-model air quality forecasts for regional air pollution monitoring

“…A second question was also answered in relation to the possible added value of higher-resolution forecast strategies ingesting detailed information from the inventory emission. When comparing the raw CAMS forecasts against the higher-resolution strategy by Sartini et al (2020) the aforementioned added value hardly emerges. The situation of substantial equivalence is however totally washed out when our dynamic calibration D-EMOS +4r is applied to the raw CAMS forecasts, thus bringing out the intrinsic strength of CAMS multi-model ensemble strategy.…”

“…The main conclusion we can draw from this first step of our analysis is that CAMS forecasts need to be calibrated using a training set of past observation-forecast pairs. Before to discuss this important issue, let us try to understand how CAMS multi-model ensemble performs with respect to the model strategy, here referred to as S2020, used in Sartini et al (2020) having higher spatial resolution and accounting more detailed emission information from the regional inventory emissions.…”

“…2 reports the statistical error indices for a sub-set of five control stations, classified as industrial (IS), rural (RS) and urban (US) stations. We focused on such a sub-set so to have the possibility to compare our results with those obtained by Sartini et al (2020) using the high-resolution numerical strategy, called LINEA, implemented and managed by ARPA Liguria to forecast the concentration of photochemical pollutants. In the table only daily means are considered as done in Sartini et al (2020).…”

“…We focused on such a sub-set so to have the possibility to compare our results with those obtained by Sartini et al (2020) using the high-resolution numerical strategy, called LINEA, implemented and managed by ARPA Liguria to forecast the concentration of photochemical pollutants. In the table only daily means are considered as done in Sartini et al (2020). The added value brought by the concomitant action of high-resolution and more detailed The question addressed here is the following: can a calibration strategy properly set for ensemble prediction systems substantially increase the forecast skill thereby overcoming persistence?…”

“…2. The table also includes the error indices for the same stations associated to the forecast model used by Sartini et al (2020). The conclusion is that the potential benefits brought by the combined use of a high-resolution model and a more detailed inventory information can be significantly exceeded upon implementing a proper dynamic calibration of the coarser model strategy.…”