Abstract

The local ensemble tangent linear model (LETLM) provides an alternative method for creating the tangent linear model (TLM) and adjoint of a nonlinear model that promises to be easier to maintain and more computationally scalable than earlier methods. In this paper, we compare the ability of the LETLM to predict the difference between two nonlinear trajectories of the Navy’s global weather prediction model at low resolution (2.5° at the equator) with that of the TLM currently used in the Navy’s four-dimensional variational (4DVar) data assimilation scheme. When compared to the pair of nonlinear trajectories, the traditional TLM and the LETLM have improved skill relative to persistence everywhere in the atmosphere, except for temperature in the planetary boundary layer. In addition, the LETLM was, on average, more accurate than the traditional TLM (error reductions of about 20% in the troposphere and 10% overall). Sensitivity studies showed that the LETLM was most sensitive to the number of ensemble members, with the performance gradually improving with increased ensemble size up to the maximum size attempted (400). Inclusion of physics in the LETLM ensemble leads to a significantly improved representation of the boundary layer winds (error reductions of up to 50%), in addition to improved winds and temperature in the free troposphere and in the upper stratosphere/lower mesosphere. The computational cost of the LETLM was dominated by the cost of ensemble propagation. However, the LETLM can be precomputed before the 4DVar data assimilation algorithm is executed, leading to a significant computational advantage.

Abstract

A methodology is presented for objectively optimizing nonorographic gravity wave source parameters to minimize forecast error for target regions and forecast lead times. In this study, we employ a high-altitude version of the Navy Global Environmental Model (NAVGEM-HA) to ascertain the forcing needed to minimize hindcast errors in the equatorial lower stratospheric zonal-mean zonal winds in order to improve forecasts of the quasi-biennial oscillation (QBO) over seasonal time scales. Because subgrid-scale wave effects play a large role in driving the QBO, this method leverages the nonorographic gravity wave drag (GWD) parameterization scheme to provide the necessary forcing. To better constrain the GWD source parameters, we utilize ensembles of NAVGEM-HA hindcasts over the 2014–16 period with perturbed source parameters and develop a cost function to minimize errors in the equatorial lower stratosphere compared to analysis. Thus, we may determine the set of GWD source parameters that yields a forecast state that most closely agrees with observed QBO winds over each optimization time interval. Results show that the source momentum flux and phase speed spectrum width are the most important parameters. The seasonal evolution of optimal parameter value, specifically a robust semiannual periodicity in the source strength, is also revealed. Changes in optimal source parameters with increasing forecast lead time are seen, as the GWD parameterization takes on a more active role as QBO driver at longer forecast lengths. Implementation of a semiannually varying source function at the equator provides RMS error improvement in QBO winds over the default constant value.

Abstract

The 3D version of the Chemical Lagrangian Model of the Stratosphere (CLAMS) is used to study the transport of CH₄ and O₃ in the Antarctic stratosphere between 1 September and 30 November 2002, that is, over the time period when unprecedented major stratospheric warming in late September split the polar vortex into two parts. The isentropic and cross-isentropic velocities in CLAMS are derived from ECMWF winds and heating/cooling rates calculated with a radiation module. The irreversible part of transport, that is, mixing, is driven by the local horizontal strain and vertical shear rates with mixing parameters deduced from in situ observations.

The CH₄ distribution after the vortex split shows a completely different behavior above and below 600 K. Above this potential temperature level, until the beginning of November, a significant part of vortex air is transported into the midlatitudes up to 40°S. The lifetime of the vortex remnants formed after the vortex split decreases with the altitude with values of about 3 and 6 weeks at 900 and 700 K, respectively.

Despite this enormous dynamical disturbance of the vortex, the intact part between 400 and 600 K that “survived” the major warming was strongly isolated from the extravortex air until the end of November. According to CLAMS simulations, the air masses within this part of the vortex did not experience any significant dilution with the midlatitude air.

By transporting ozone in CLAMS as a passive tracer, the chemical ozone loss was estimated from the difference between the observed [Polar Ozone and Aerosol Measurement III (POAM III) and Halogen Occultation Experiment (HALOE)] and simulated ozone profiles. Starting from 1 September, up to 2.0 ppmv O₃ around 480 K and about 70 Dobson units between 450 and 550 K were destroyed until the vortex was split. After the major warming, no additional ozone loss can be derived, but in the intact vortex part between 450 and 550 K, the accumulated ozone loss was “frozen in” until the end of November.

Abstract

In 1992 and 1993, the authors made measurements of the marine boundary layer off the coast of Oregon from an airship. In 1992, these measurements consisted of coherent microwave backscatter measurements at K_u band taken from the gondola of the airship and micrometeorological and wave height measurements made from an airborne platform suspended by a cable 65 m below the gondola so that it was between 5 and 20 m above the sea surface. In 1993, an infrared imaging system was added to the suite of instruments operated in the gondola and two narrowbeam infrared thermometers were mounted in the suspended platform. In both years, a sonic anemometer and a fast humidity sensor were carried on the suspended platform and used to measure surface layer fluxes in the atmosphere above the ocean. A laser altimeter gave both the altitude of the suspended platform and a point measurement of wave height. By operating all these instruments together from the slow-moving airship, the authors were able to measure atmospheric fluxes, microwave cross sections and Doppler characteristics, air and sea surface temperatures, and wave heights simultaneously and coincidentally at much higher spatial resolutions than had been possible before. Here the authors document the methods and present observations of the neutral drag coefficient between wind speeds of 2 and 10 m s⁻¹, the relationship between the wind vector and the microwave cross section, and the effect of a sharp sea surface temperature front on both the wind vector and the microwave cross section. The drag coefficients first decrease with increasing wind speed, then reach a minimum and begin to increase with further increases in the wind speed. The values of the drag coefficient at very low wind speeds are higher than those given by Smith, however, and the minimum drag coefficient seems to occur somewhat above the wind speed he indicates. The authors show that their measured azimuthally averaged cross sections fall somewhat below the SASS II model function of Wentz et al. at low wind speeds but are rather close to that model at higher wind speeds. Coefficients describing the dependence of the cross section on azimuth angle are generally close to those of SASS II. The azimuthally averaged cross sections generally fall within the 90% confidence interval of the model function based on friction velocity recently proposed by Weissman et al. but are often near the upper limit of this interval. Somewhat surprisingly, a residual dependence on atmospheric stratification is found in the neutral drag coefficients and in the microwave cross sections when plotted against a neutral wind speed obtained using the Businger–Dyer stability corrections. This indicates that these corrections are not adequate over the ocean for stable conditions and the authors suggest that wave-induced shear near the surface may be the reason. Finally, it is shown that winds around a sea surface temperature front can rapidly change direction and that the microwave cross section follows this change except very near the front where it becomes more isotropic than usual.

Abstract

An essential component of four-dimensional variational data assimilation is the tangent linear model (TLM), which is a linearized version of the full nonlinear forecast model. A relatively new approach to calculating the TLM is a regression model called the ensemble tangent linear model (ETLM). Here we validate the ETLM for linearizing a nonorographic gravity wave drag (NGWD) subgrid-scale model. The regression is applied to an ensemble created by perturbing the atmospheric state and calculating one time step of the NGWD model. The ETLM is validated using independent perturbations based on archived analysis increments. We examine how the skill of the NGWD ETLM depends on the choice of ensemble perturbation, ensemble size, amount of ensemble inflation/deflation, and the size of the localization stencil. After examining the nearly perfect results using a large ETLM ensemble (100 000 members), optimal tuning is then performed for 150–500 members. For smaller ETLM ensembles, spurious noise due to sampling error could be reduced either by downscaling the perturbations or by localizing the ETLM. The impact of localization decreases as the ETLM ensemble size increases. We then validate the ETLM using one year of archived DA analysis increments. The skill varies over time with percentage errors relative to persistence forecasts (where 100% is no skill, 0% is a perfect forecast) generally ranging from ∼50% to 90% (∼40% to 80%) for ETLMs with 150 (500) members. The ETLM is also shown to propagate small increments (1% of the size of analysis increments) with fractional errors of ∼10%.

Abstract

Gravity wave (GW) momentum and energy deposition are large components of the momentum and heat budgets of the stratosphere and mesosphere, affecting predictability across scales. Since weather and climate models cannot resolve the entire GW spectrum, GW parameterizations are required. Tuning these parameterizations is time-consuming and must be repeated whenever model configurations are changed. We introduce a self-tuning approach, called GW parameter retrieval (GWPR), applied when the model is coupled to a data assimilation (DA) system. A key component of GWPR is a linearized model of the sensitivity of model wind and temperature to the GW parameters, which is calculated using an ensemble of nonlinear forecasts with perturbed parameters. GWPR calculates optimal parameters using an adaptive grid search that reduces DA analysis increments via a cost-function minimization. We test GWPR within the Navy Global Environmental Model (NAVGEM) using three latitude-dependent GW parameters: peak momentum flux, phase-speed width of the Gaussian source spectrum, and phase-speed weighting relative to the source-level wind. Compared to a baseline experiment with fixed parameters, GWPR reduces analysis increments and improves 5-day mesospheric forecasts. Relative to the baseline, retrieved parameters reveal enhanced source-level fluxes and westward shift of the wave spectrum in the winter extratropics, which we relate to seasonal variations in frontogenesis. The GWPR reduces stratospheric increments near 60°S during austral winter, compensating for excessive baseline nonorographic GW drag. Tropical sensitivity is weaker due to significant absorption of GW in the stratosphere, resulting in less confidence in tropical GWPR values.

Abstract

Upper atmosphere sounding (UAS) channels of the Special Sensor Microwave Imager/Sounder (SSMIS) were assimilated using a high-altitude version of the Navy Global Environmental Model (NAVGEM) in order to investigate their potential for operational forecasting from the surface to the mesospause. UAS radiances were assimilated into NAVGEM using the new Community Radiative Transfer Model (CRTM) that accounts for Zeeman line splitting by geomagnetic fields. UAS radiance data from April 2010 to March 2011 are shown to be in good agreement with coincident temperature measurements from the Sounding of the Atmosphere Using Broadband Emission Radiometry (SABER) instrument that were used to simulate UAS brightness temperatures. Four NAVGEM experiments were performed during July 2010 that assimilated (i) no mesospheric observations, (ii) UAS data only, (iii) SABER and Microwave Limb Sounder (MLS) mesospheric temperatures only, and (iv) SABER, MLS, and UAS data. Zonal mean temperatures and observation − forecast differences for the UAS-only and SABER+MLS experiments are similar throughout most of the mesosphere, and show large improvements over the experiment assimilating no mesospheric observations, proving that assimilation of UAS radiances can provide a reliable large-scale constraint throughout the mesosphere for operational, high-altitude analysis. This is confirmed by comparison of solar migrating tides and the quasi-two-day wave in the mesospheric analyses. The UAS-only experiment produces realistic tidal and two-day wave amplitudes in the summer mesosphere in agreement with the experiments assimilating MLS and SABER observations, whereas the experiment with no mesospheric observations produces excessively strong mesospheric winds and two-day wave amplitudes.

Abstract

An ensemble-based tangent linear model (TLM) is described and tested in data assimilation experiments using a global shallow-water model (SWM). A hybrid variational data assimilation system was developed with a 4D variational (4DVAR) solver that could be run either with a conventional TLM or a local ensemble TLM (LETLM) that propagates analysis corrections using only ensemble statistics. An offline ensemble Kalman filter (EnKF) is used to generate and maintain the ensemble. The LETLM uses data within a local influence volume, similar to the local ensemble transform Kalman filter, to linearly propagate the state variables at the central grid point. After tuning the LETLM with offline 6-h forecasts of analysis corrections, cycling experiments were performed that assimilated randomly located SWM height observations, based on a truth run with forced bottom topography. The performance using the LETLM is similar to that of the conventional TLM, suggesting that a well-constructed LETLM could free 4D variational methods from dependence on conventional TLMs. This is a first demonstration of the LETLM application within a context of a hybrid-4DVAR system applied to a complex two-dimensional fluid dynamics problem. Sensitivity tests are included that examine LETLM dependence on several factors including length of cycling window, size of analysis correction, spread of initial ensemble perturbations, ensemble size, and model error. LETLM errors are shown to increase linearly with correction size in the linear regime, while TLM errors increase quadratically. As nonlinearity (or forecast model error) increases, the two schemes asymptote to the same solution.

Abstract

An ensemble-based linearized forecast model has been developed for data assimilation applications for numerical weather prediction. Previous studies applied this local ensemble tangent linear model (LETLM) to various models, from simple one-dimensional models to a low-resolution (~2.5°) version of the Navy Global Environmental Model (NAVGEM) atmospheric forecast model. This paper applies the LETLM to NAVGEM at higher resolution (~1°), which required overcoming challenges including 1) balancing the computational stencil size with the ensemble size, and 2) propagating fast-moving gravity modes in the upper atmosphere. The first challenge is addressed by introducing a modified local influence volume, introducing computations on a thin grid, and using smaller time steps. The second challenge is addressed by applying nonlinear normal mode initialization, which damps spurious fast-moving modes and improves the LETLM errors above ~100 hPa. Compared to a semi-Lagrangian tangent linear model (TLM), the LETLM has superior skill in the lower troposphere (below 700 hPa), which is attributed to better representation of moist physics in the LETLM. The LETLM skill slightly lags in the upper troposphere and stratosphere (700–2 hPa), which is attributed to nonlocal aspects of the TLM including spectral operators converting from winds to vorticity and divergence. Several ways forward are suggested, including integrating the LETLM in a hybrid 4D variational solver for a realistic atmosphere, combining a physics LETLM with a conventional TLM for the dynamics, and separating the LETLM into a sequence of local and nonlocal operators.

Abstract

A data assimilation system (DAS) is described for global atmospheric reanalysis from 0- to 100-km altitude. We apply it to the 2014 austral winter of the Deep Propagating Gravity Wave Experiment (DEEPWAVE), an international field campaign focused on gravity wave dynamics from 0 to 100 km, where an absence of reanalysis above 60 km inhibits research. Four experiments were performed from April to September 2014 and assessed for reanalysis skill above 50 km. A four-dimensional variational (4DVAR) run specified initial background error covariances statically. A hybrid-4DVAR (HYBRID) run formed background error covariances from an 80-member forecast ensemble blended with a static estimate. Each configuration was run at low and high horizontal resolution. In addition to operational observations below 50 km, each experiment assimilated 10⁵ observations of the mesosphere and lower thermosphere (MLT) every 6 h. While all MLT reanalyses show skill relative to independent wind and temperature measurements, HYBRID outperforms 4DVAR. MLT fields at 1-h resolution (6-h analysis and 1–5-h forecasts) outperform 6-h analysis alone due to a migrating semidiurnal (SW2) tide that dominates MLT dynamics and is temporally aliased in 6-h time series. MLT reanalyses reproduce observed SW2 winds and temperatures, including phase structures and 10–15-day amplitude vacillations. The 0–100-km reanalyses reveal quasi-stationary planetary waves splitting the stratopause jet in July over New Zealand, decaying from 50 to 80 km then reintensifying above 80 km, most likely via MLT forcing due to zonal asymmetries in stratospheric gravity wave filtering.