Model output statistics (MOS) approaches relying on machine learning
algorithms were applied to downscale regional air quality forecasts produced by CAMS (Copernicus Atmosphere Monitoring Service) at ...hundreds of monitoring sites across Europe. Besides the CAMS forecast, the predictors in the MOS typically include meteorological variables but also ancillary data. We explored first a “local” approach where specific models are trained at each site. An alternative “global” approach where a single model is trained with data from the whole geographical domain was also investigated. In both cases, local predictors are used for a given station in predictive mode. Because of its global nature, the latter approach can capture a variety of meteorological situations within a very short training period and is thereby more suited to cope with operational constraints in relation to the training of the MOS (frequent upgrades of the modelling system, addition of new monitoring sites). Both approaches have been implemented using a variety of machine learning algorithms: random forest, gradient boosting, and standard and regularized multi-linear models. The quality of the MOS predictions is evaluated in this work for four key pollutants, namely particulate matter (PM10 and PM2.5), ozone (O3) and nitrogen dioxide (NO2), according to scores based on the predictive errors and on the detection of pollution peaks (exceedances of the regulatory thresholds). Both the local and the global approaches significantly improve the performances of the raw ensemble forecast. The most important result of this study is that the global approach competes with and can even outperform the local approach in some cases. This global approach gives the best RMSE scores when relying on a random forest model for the prediction of daily mean, daily max and hourly concentrations. By contrast, it is the gradient boosting model which is better suited for the detection of exceedances of the European Union regulated threshold values for O3 and PM10.
Ensemble-based techniques have been widely utilized in estimating uncertainties in various problems of interest in geophysical applications. A new cloud retrieval method is proposed based on the ...particle filter (PF) by using ensembles of cloud information in the framework of Gridpoint Statistical Interpolation (GSI) system. The PF cloud retrieval method is compared with the Multivariate Minimum Residual (MMR) method that was previously established and verified. Cloud retrieval experiments involving a variety of cloudy types are conducted with the PF and MMR methods with measurements of infrared radiances on multi-sensors onboard both geostationary and polar satellites, respectively. It is found that the retrieved cloud masks with both methods are consistent with other independent cloud products. MMR is prone to producing ambiguous small-fraction clouds, while PF detects clearer cloud signals, yielding closer heights of cloud top and cloud base to other references. More collections of small-fraction particles are able to effectively estimate the semi-transparent high clouds. It is found that radiances with high spectral resolutions contribute to quantitative cloud top and cloud base retrievals. In addition, a different way of resolving the filtering problem over each model grid is tested to better aggregate the weights with all available sensors considered, which is proven to be less constrained by the ordering of sensors. Compared to the MMR method, the PF method is overall more computationally efficient, and the cost of the model grid-based PF method scales more directly with the number of computing nodes.
Clouds play a key role in radiation and hence O3 photochemistry by
modulating photolysis rates and light-dependent emissions of biogenic
volatile organic compounds (BVOCs). It is not well known, ...however, how much
error in O3 predictions can be directly attributed to error in cloud
predictions. This study applies the Weather Research and Forecasting with
Chemistry (WRF-Chem) model at 12 km horizontal resolution with the Morrison
microphysics and Grell 3-D cumulus parameterization to quantify uncertainties
in summertime surface O3 predictions associated with cloudiness over
the contiguous United States (CONUS). All model simulations are driven by
reanalysis of atmospheric data and reinitialized every 2 days. In sensitivity
simulations, cloud fields used for photochemistry are corrected based on
satellite cloud retrievals. The results show that WRF-Chem predicts about
55 % of clouds in the right locations and generally underpredicts cloud
optical depths. These errors in cloud predictions can lead to up to 60 ppb
of
overestimation in hourly surface O3 concentrations on some days. The
average difference in summertime surface O3 concentrations derived from
the modeled clouds and satellite clouds ranges from 1 to 5 ppb for maximum
daily 8 h average O3 (MDA8 O3) over the CONUS. This represents up to
∼ 40 % of the total MDA8 O3 bias under cloudy conditions in
the tested model version. Surface O3 concentrations are sensitive to
cloud errors mainly through the calculation of photolysis rates (for
∼ 80 %), and to a lesser extent to light-dependent BVOC emissions.
The sensitivity of surface O3 concentrations to satellite-based cloud
corrections is about 2 times larger in VOC-limited than NOx-limited
regimes. Our results suggest that the benefits of accurate predictions of
cloudiness would be significant in VOC-limited regions, which are typical of
urban areas.
Clouds play a critical role in modulating tropospheric radiation and thus photochemistry. We develop a methodology for calculating the vertical distribution of tropospheric ultraviolet (300–420 nm) ...actinic fluxes using satellite cloud retrievals and a radiative transfer model. We demonstrate that our approach can accurately reproduce airborne‐measured actinic fluxes from the 2013 Studies of Emissions and Atmospheric Composition, Clouds and Climate Coupling by Regional Surveys (SEAC4RS) campaign as a case study. The results show that the actinic flux is reduced below moderately thick clouds with increasing cloud optical depth and can be enhanced by a factor of 2 above clouds. Inside clouds, the actinic flux can be enhanced by up to 2.4 times in the upper part of clouds or reduced up to 10 times in the lower parts of clouds. Our study suggests that the use of satellite‐derived actinic fluxes as input to chemistry‐transport models can improve the accuracy of photochemistry calculations.
Key Points
Satellite cloud retrievals are used in a radiative transfer model to simulate cloudy‐sky actinic flux in the U.S.
Clouds significantly attenuate and/or enhance actinic fluxes relative to cloud‐free skies
Satellite‐constrained actinic fluxes capture well in situ airborne measurements from SEAC4RS where cloud‐sky conditions represent more than 60% of the total data
Accuracy of cloud predictions in numerical weather models can considerably impact ozone (O3) forecast skill. This study assesses the benefits in surface O3 predictions of using the Rapid Refresh ...(RAP) forecasting system that assimilates clouds as well as conventional meteorological variables at hourly time scales. We evaluate and compare the WRF‐Chem simulations driven by RAP and the Global Forecast System (GFS) forecasts over the Contiguous United States (CONUS) for 2016 summer. The day 1 forecasts of surface O3 and temperature driven by RAP are in better agreements with observations. Reductions of 5 ppb in O3 mean bias error and 2.4 ppb in O3 root‐mean‐square‐error are obtained on average over CONUS with RAP compared to those with GFS. The WRF‐Chem simulation driven by GFS shows a higher probability of capturing O3 exceedances but exhibits more frequent false alarms, resulting from its tendency to overpredict O3. The O3 concentrations are found to respond mainly to the changes in boundary layer height that directly affects the mixing of O3 and its precursors. The RAP data assimilation shows improvements in the cloud forecast skill during the initial forecast hours, which reduces O3 forecast errors at the initial forecast hours especially under cloudy‐sky conditions. Sensitivity simulations utilizing satellite clouds show that the WRF‐Chem simulation with RAP produces too thick low‐level clouds, which leads to O3 underprediction in the boundary layer.
Key Points
The performance for day 1 O3 forecasts driven by RAP is generally improved than that driven by GFS
The reduction in day 1 O3 forecast errors is mainly associated with the boundary layer height
The cloud assimilation improves cloud forecasts for the first few hours and reduces errors in 1‐hr O3 forecast under cloudy conditions
Most of the annual rainfall in the Southeastern Mediterranean falls in the wet season from November to March. It is associated with Mediterranean cyclones, and is sensitive to climate variability. ...Predicting the wet season precipitation with a few months advance is highly valuable for water resource planning and climate-associated risk management in this semi-arid region. The regional water resource managements and climate-sensitive economic activities have relied on seasonal forecasts from global climate prediction centers. However due to their coarse resolutions, global seasonal forecasts lack regional and local scale information required by regional and local water resource managements. In this study, an analog statistical-downscaling algorithm, k-nearest neighbors (KNN), was introduced to bridge the gap between the coarse forecasts from global models and the needed fine-scale information for the Southeastern Mediterranean. The algorithm, driven by the NCEP Climate Forecast System (CFS) operational forecast and the NCEP/DOE reanalysis, provides monthly precipitations at 2–4months of lead-time at 18 stations within the major regional hydrological basins. Large-scale predictors for KNN were objectively determined by the correlations between the station historic daily precipitation and variables in reanalysis and CFS reforecast. Besides a single deterministic forecast, this study constructed sixty ensemble members for probabilistic estimates. The KNN algorithm demonstrated its robustness when validated with NCEP/DOE reanalysis from 1981 to 2009 as hindcasts before applied to downscale CFS forecasts. The downscaled predictions show fine-scale information, such as station-to-station variability. The verification against observations shows improved skills of this downscaling utility relative to the CFS model. The KNN-based downscaling system has been in operation for the Israel Water Authority predicting precipitation and driving hydrologic models estimating river flow and aquifer charge for water supply.
► Downscaled global scale seasonal precipitation prediction to fine scale. ► Downscaling algorithm enhanced the usefulness of global seasonal forecasts. ► Multi-realization ensemble probabilistic forecasts for operation.