Description: We propose a further refinement of the the non-homogeneous Gaussian regression (NGR) approach for temperature, which transforms the output of an ensemble prediction system (EPS) into predictive Gaussian distributions at each location of interest. Model fitting is partly done within a regression framework using a penalized version of the least squares loss function. This is conceptually simpler than the original approach and at the same time able to prevent overfitting. While calibration is initially performed at observation locations only, geostatistical methods are used to provide predictive distributions on the entire grid. The incorporation of land use information in this interpolation scheme further improves predictive performance even though a simpler statistical model than in the original approach is used. The assessment of predictive performance and calibration is carried out with dynamical forecasts of 2-meter temperatures by the COSMO-DE-EPS, an application of the COSMO (Consortium for Small-scale Modeling) model system that covers the area of Germany and neighbouring countries.

Description: The representation of rainfall and in particular its diurnal cycle, is generally poor in general circulation models (GCMs). Nonetheless, studies make use of GCMs in future climate projections in regions where precipitation has a strong diurnal cycle. In this study we evaluate whether one GCM (ACCESS1.3) can represent the rainfall in such a region (Australia) where the diurnal cycle of rainfall is produced as a result of both the destabilisation of the boundary layer (convection) and a larger-scale re-organisation of the low level flow, the latter of which may be resolved by a GCM. In northern and eastern Australia, where the diurnal cycle of rainfall is controlled by convective processes, the GCM produces rain 3–6 hours too early in the day. Nevertheless, the model represents the continental-scale reorganisation of the low-level circulation that results from the diurnal cycle of surface heating and cooling. A nocturnal low-level jet forms over the western half of the continent with strong convergence at the jet exit, which initiates rain overnight in the continental north-west in agreement with previous work. The model also captures the change in the air-flow direction, from a southeasterly to a northeasterly, responsible for bringing the necessary moisture for precipitation to occur. Thus, while the model may have a tendency to initiate convection too early, it is able to represent the larger scale nocturnal re-organisation of the flow and the associated rainfall.

Description: Over enclosed and semi-enclosed bodies of waters, the land-breeze/sea-breeze circulation is expected to be modified by the presence of opposing coastlines. These effects are studied using satellite scatterometer surface wind observations from the QuikSCAT and ADEOS-2 tandem mission from April to October 2003. Winds are studied for six bodies of water: the Red Sea, the Gulf of California, the Mediterranean, the Adriatic Sea, the Black Sea, and the Caspian Sea. These bodies of water are large enough for the geographic orientation of the diurnal winds relative to the coastline to match the expected orientation for a straight coastline. Land breezes from opposite coastlines converge in the middle of these bodies of water, and in some cases the convergence line is shifted substantially away from the mid-point between opposite coastlines. Displacements in the convergence line appear likely to be explained by differences in the strength of the diurnal winds emanating from opposite coastlines, associated with differential heating or with island/peninsula effects, and by geographic displacements associated with large-scale mean wind patterns.

Description: One of the prerequisites for achieving skill in decadal climate prediction is to successfully initialise and predict the circulation in the Atlantic Ocean. The RAPID array measures the Atlantic Meridional Overturning Circulation (MOC) at 26°N. Here we develop a method to include these observations in the UK Met Office Decadal Prediction System (DePreSys). The proposed method uses covariances of overturning transport at 26°N with ocean temperature and salinity throughout the ocean to create the density structure necessary to reproduce the observed transport. Assimilating transport alone in this way effectivly reproduces the observed transport at 26°N, and is better than using basin-wide temperature and salinity observations only. However, when the transport observations are combined with insitu tempertature and salinity observations in the analysis, the transport is not currently reproduced as well. The reasons for this are investigated using pseudo-observations in a twin experiment framework. Sensitivity experiments are used to show that the MOC at monthly time scales in this model is modulated by a very specific mechanism where non-local density anomalies appear to be more important for transport variability at 26°N in this model than local density gradients. This may not be the case in a different model .

Description: The impact of Radio Occultation observations from Global Positioning System satellites (GPSRO) on global Numerical Weather Prediction has been analysed with a recent version of the ECMWF Integrated Forecasting System. As in previous studies, the use of GPSRO was found to improve the NWP forecast skill and to drastically decrease model induced temperature biases in the analysis. The maximum forecast impact is in the lower and middle stratosphere, where the GPSRO observations have the smallest errors, but it is also visible in the troposphere. The tropospheric impact of GPSRO comes in part from direct tropospheric measurements and in part from stratosphere-troposphere interactions: This second mechanism is found to be particularly important during the northern hemisphere winter. The forecast impact of GPSRO observations is compared with that of conventional and hyperspectral satellite nadir sounders. It is found that while GPSRO data have a smaller impact than those of either class of nadir sounders, they are still able to account for a considerable fraction (30% to 70%) of the global forecast error reduction afforded by the use of the full observing system over a system which only uses conventional observations. When forecast verification is performed against radiosonde observations, GPSRO is found to be the most valuable satellite observing system in the lower stratosphere. This is remarkable in view of the relative sparseness of the GPSRO spatial and temporal coverage and an indication of the potential improvements that a denser GPSRO observing network would be able to provide. The forecast impact of GPSRO observations is also evaluated in the context of a data denial assimilation experiment with respect to the full observing system. Results are found to be consistent with those from the reduced baseline observational network and also indicate a statistically significant positive impact on tropospheric synoptic skill scores.

Description: Low-frequency variability and trends in temperature from 1979 to 2012 are examined. Observational improvements are noted and near-surface behaviour of the ERA-Interim reanalysis is reviewed. Attention is then focussed on how closely ERA-Interim fits the upper-air data it assimilates, the bias adjustments it infers for satellite data, and its agreement with the ERA-40, MERRA and JRA-55 reanalyses and with model simulations. Global-mean fits to independently homogenised radiosonde temperatures and variationally adjusted satellite brightness temperatures are mainly within 0.1K in the troposphere, with some degradation over time from assimilating varying amounts of aircraft and rain-affected microwave-radiance data, and from a change in source of sea-surface-temperature analysis. Lower-tropospheric warming appears to be somewhat underestimated. Temperature variations in the tropical upper troposphere correlate well with those at the surface, but amplitude is more than doubled, in agreement with modelling. Specific humidity varies in concert; relative humidity is largely uniform, but dips during El Niño events. Agreement with the other reanalyses is particularly close in the lower stratosphere, where radiance data and the background model constrain cooling to be slightly slower than in the homogenised radiosonde data. Perturbations to global-mean temperatures from underestimating warming following the El Chichón and Pinatubo volcanic eruptions and from assimilating recent GPSRO data are at most 0.2K, less than 20% of the net change since 1979 at 50 hPa. Middle-stratospheric variations are more uncertain. Recent cooling appears to be underestimated by assimilating increasing amounts of unadjusted radiosonde data, but results do not support a recent reprocessing of earlier sounding data that suggests stronger middle-stratospheric cooling than previously indicated. Strong analysed upper-stratospheric cooling agrees quite well with model simulations if occasional jumps due to unadjusted bias changes in high-sounding satellite data are discounted. Producing ERA-Interim in two separate streams caused only minor discontinuities where streams join at the start of 1989.

Description: Divergence and vorticity are well known to be geometrically invariant quantities in that their mathematical forms are independent of the orientation of the coordinate axes. Various other functions of the elements of the horizontal velocity gradient tensor are invariants in the same sense: examples are the resultant deformation and the determinant and Frobenius norm of the tensor. A brief account of these quadratic invariants is given, including expressions relating them to divergence and vorticity and to one another, and noting their occurrence in the divergence equation. Assuming shallow water dynamics with background rotation, time evolution equations for the resultant deformation and the other quadratic invariants are derived and compared. None rivals the vorticity and potential vorticity equations for compactness, but each may be written quite concisely in terms of familiar quantities . Corresponding time evolution equations under quasi-geostrophic shallow water dynamics are also derived, and lead to a simple prognostic equation for the ageostrophic vorticity.

Description: Dynamical imbalances can induce spurious variability which can be diagnosed from the physical tendencies observed in the first moments of short-term forecasts using as initial conditions analyses obtained from an assimilation system using this model. In this paper this approach is taken to investigate differences in the balance obtained from 3D and 4D-Var analyses, using the forecast-assimilation system of the Meteorological Service of Canada (MSC). The results indicate that the model is then in good balance globally but the 4D-Var analyses slightly upsets the balance in the Tropics thereby altering the characteristics of the Inter-tropical Convergence Zone (ITCZ). As the assimilation is driven by a particular model, the resulting analyses keep an imprint of the dynamics of that model and using this analysis with another model may not be as well in balance due to the differences between the two models. To study this point, ERA-interim 4D-Var reanalyses were used as initial conditions first at a lower horizontal and vertical resolution, and then at a resolution closer to that of the Global Environmental Multiscale (GEM) model. The higher resolution reanalyses led to a better balance than that with a lower resolution version of the ERA-interim reanalyses. The coarser analyses create significant imbalances in the Canadian global model which persist for more than 5 days. In particular, it was noted that convection is nearly absent early on as if at a lower resolution, the ERA-interim analyses did not inject sufficient humidity to trigger convection. It was also noted that reducing the vertical resolution is more damaging than using a coarser horizontal resolution. In limited-area regional climate models, external analyses are used to define the boundary conditions and the Canadian Regional Climate Model (CRCM) was used to assess the impact of different ways to define the boundary conditions. The CRCM is a limited-area configuration of the GEM global model used in the 3D and 4D-Var assimilation. Experiments were conducted in which the boundary conditions driving the CRCM are provided every 6-h as is usually done for the CRCM climate simulations. When using 4D-Var analyses and ERA-interim reanalyses (coarse and full resolution) to define the boundary conditions, the results indicate that imbalances persist even after 15 days and are more significant for the coarser analyses. Moreover, even though the model exhibits relatively good balance initially, after 5-days imbalances appear gradually in the interior of the regional model domain.

Description: This paper aims to understand to what extent the winds fields from an advanced Numerical Weather Prediction (NWP) system and from satellite scatterometer describe the same spatial and temporal features of the wind in the Mediterranean Sea. We investigated wind fields for the period February 2010 to February 2012 using the ASCAT scatterometer data with 12.5 km wind vector cells and the analysis wind fields from the ECMWF T1279 global model. The ASCAT-ECMWF mean relative bias and centered Root Mean Square deviation of wind speed, normalized by scatterometer wind speed , and , have been found to be 7 % and 23 %. An interesting result is the identification of dependence of both and on the distance from the coast, indicating the coastal areas as the main source of discrepancy between the two data sets. From 50 km to 200 km away from coast, decreases from 40 % to 25 % and from 8 % to 4 %. These results gain more importance considering that the Mediterranean Sea is essentially a coastal sea (50 % of its surface lies within 50 km from the coast). Both and have been found to depend non linearly on the wind speed. The seasonal variation of and shows that they are in phase opposition, with higher values of during the warm season (April to October). It is hypothesized that local coastal circulations like land/sea breezes could explain the observed mismatch between model and observations. The reported findings emphasize a common feature of the present atmospheric models forecasting winds over regional basins like the Mediterranean.

Description: The EarthCARE satellite mission plans to perform a continuous closure experiment to assess the quality of retrieved cloud and aerosol properties. It will do so by comparing top-of-atmosphere (TOA) broadband (BB) fluxes to simulated values produced by 3D radiative transfer models that act on the 2D retrieved cross-section and a 3D atmosphere around it produced by a scene construction algorithm (SCA). This study proposes and tests a method for estimating errors in simulated TOA BB fluxes due to the SCA. Two methods for estimating SCA-related errors for TOA fluxes are presented. The primary one relies on computation of errors for reconstructed narrowband imager nadir radiances. A-train satellite data were used to show that for constructed domains measuring (11 km) 2 , approximately the size of EarthCARE assessment domains, with total cloud fractions 〉 0.2, errors for reflected BB shortwave fluxes due to the SCA are smaller than ±4.2 W m − 2 and ±11.5 W m − 2 for 66% and 90% of the domains, respectively. Corresponding values for outgoing longwave fluxes are ±1.2 W m − 2 and ±3.0 W m − 2 . Largest and smallest errors are associated with fields of broken convective cloud and overcast stratiform cloud, respectively. The SCA was applied to simulated measurements for a (153 km) 2 field of deep convective clouds produced by a cloud system-resolving model. Actual and estimated TOA BB shortwave flux errors due to the SCA agree well and are smaller than ±22 W m − 2 and ±40 W m − 2 for 66% and 90% of the (11 km) 2 sampled sub-domains. Assuming that errors due to the SCA are purely bias errors, they were subtracted from fluxes estimated for the constructed domains. This resulted in TOA BB shortwave flux errors smaller than ±7 W m − 2 and ±25 W m − 2 for 66% and 90% of the sampled sub-domains. This suggests that estimated errors due to the SCA should be removed directly from simulated TOA BB fluxes before executing the closure assessment.