Search Results

You are looking at 1 - 10 of 63 items for

  • Author or Editor: Juanzhen Sun x
  • Refine by Access: All Content x
Clear All Modify Search
Juanzhen Sun

Abstract

The feasibility of initializing a numerical cloud model with single-Doppler observations and predicting the evolution of thunderstorms has been tested using an observed case of a supercell storm during the Severe Thunderstorm Electrification and Precipitation Study (STEPS). Single-Doppler observations from the Weather Surveillance Radar-1988 Doppler (WSR-88D) at Goodland, Kansas, are assimilated into a cloud-scale numerical model using a four-dimensional variational data assimilation (4DVAR) scheme. A number of assimilation and short-range numerical prediction experiments are conducted. Both the assimilation and prediction results are compared with those of a dual-Doppler synthesis. The prediction results are also verified with reflectivity observations. It is shown that the analysis of the wind field captures the major structure of the storm as revealed by the dual-Doppler synthesis. Thermodynamical and microphysical features retrieved through the dynamical model show consistency with expectations for a deep convective storm. The predicted storm evolution represented by the reflectivity field correlates well with the observations for a 2-h prediction period. The relative importance of the initial fields on the subsequent prediction of the storm evolution is examined by alternately removing the perturbation in each of the initial fields. It is shown that the prediction is most sensitive to the initialization of wind, water vapor, and temperature perturbations.

A number of sensitivity experiments for initialization are conducted to show how the initial analysis depends on the application of a cycling procedure, the weights of the smoothness constraint, and the relative importance between the radial velocity and the reflectivity observations. It is found that the application of the cycling procedure improves the analysis and the subsequent forecast. Greater smoothness coefficients of the penalty term in the cost function result in a larger rms difference in the wind analysis, but help spread the information out and improve the forecast slightly. The radial velocity observations play a more important role than the reflectivity in terms of the wind analysis and the subsequent precipitation forecast.

Full access
Juanzhen Sun

Abstract

Previous experiments with the adjoint technique for determining the three-dimensional wind and thermodynamic fields from single-Doppler radar data have assumed that the radar observes one component of the velocity in Cartesian coordinates. This technique is generalized to radial velocity observations by fitting a Cartesian prediction model to interpolated radial velocity data in Cartesian coordinates. The three-dimensional wind and temperature are determined by minimizing the difference between simulated single-Doppler observations of radial velocity and reflectivity in Cartesian coordinates and their predictions from a numerical model. An application of this technique to a simulation of dry convection gives successful results.

Full access
Juanzhen Sun
and
Ying Zhang

Abstract

This paper presents a case study on the assimilation of observations from multiple Doppler radars of the Next Generation Weather Radar (NEXRAD) network. A squall-line case documented during the International H2O Project (IHOP_2002) is used for the study. Radar radial velocity and reflectivity observations from four NEXRADs are assimilated into a convection-permitting model using a four-dimensional variational data assimilation (4DVAR) scheme. A mesoscale analysis using a supplementary sounding, velocity–azimuth display (VAD) profiles, and surface observations from Meteorological Aerodrome Reports (METAR) are produced and used to provide a background and boundary conditions for the 4DVAR radar data assimilation. Impact of the radar data assimilation is assessed by verifying the skill of the subsequent very short-term (5 h) forecasts.

Assimilation and forecasting experiments are conducted to examine the impact of radar data assimilation on the subsequent precipitation forecasts. It is found that the 4DVAR radar data assimilation significantly reduces the model spinup required in the experiments without radar data assimilation, resulting in significantly improved 5-h forecasts. Additional experiments are conducted to study the sensitivity of the precipitation forecasts with respect to 4DVAR cycling configurations. Results from these experiments suggest that the forecasts with three 4DVAR cycles are improved over those with cold start, but the cycling impact seems to diminish with more cycles. The impact of observations from each of the individual radars is also examined by conducting a set of experiments in which data from each radar are alternately excluded. It is found that the accurate analysis of the environmental wind surrounding the convective cells is important in successfully predicting the squall line.

Full access
Qingnong Xiao
and
Juanzhen Sun

Abstract

The impact of multiple–Doppler radar data assimilation on quantitative precipitation forecasting (QPF) is examined in this study. The newly developed Weather Research and Forecasting (WRF) model Advanced Research WRF (ARW) and its three-dimensional variational data assimilation system (WRF 3DVAR) are used. In this study, multiple–Doppler radar data assimilation is applied in WRF 3DVAR cycling mode to initialize a squall-line convective system on 13 June 2002 during the International H2O Project (IHOP_2002) and the ARW QPF skills are evaluated for the case. Numerical experiments demonstrate that WRF 3DVAR can successfully assimilate Doppler radial velocity and reflectivity from multiple radar sites and extract useful information from the radar data to initiate the squall-line convective system. Assimilation of both radial velocity and reflectivity results in sound analyses that show adjustments in both the dynamical and thermodynamical fields that are consistent with the WRF 3DVAR balance constraint and background error correlation. The cycling of the Doppler radar data from the 12 radar sites at 2100 UTC 12 June and 0000 UTC 13 June produces a more detailed mesoscale structure of the squall-line convection in the model initial conditions at 0000 UTC 13 June. Evaluations of the ARW QPF skills with initialization via Doppler radar data assimilation demonstrate that the more radar data in the temporal and spatial dimensions are assimilated, the more positive is the impact on the QPF skill. Assimilation of both radial velocity and reflectivity has more positive impact on the QPF skill than does assimilation of either radial velocity or reflectivity only. The improvement of the QPF skill with multiple-radar data assimilation is more clearly observed in heavy rainfall than in light rainfall. In addition to the improvement of the QPF skill, the simulated structure of the squall line is also enhanced by the multiple–Doppler radar data assimilation in the WRF 3DVAR cycling experiment. The vertical airflow pattern shows typical characteristics of squall-line convection. The cold pool and its related squall-line convection triggering process are better initiated in the WRF 3DVAR analysis and simulated in the ARW forecast when multiple–Doppler radar data are assimilated.

Full access
Frédéric Fabry
and
Juanzhen Sun

Abstract

Data assimilation is used among other things to constrain the initial conditions of weather forecasting models by fitting the model fields to observations made over a certain time interval. In particular, it tries to tie incomplete data with model constraints to detect and correct for initial condition errors. This is possible only if initial condition errors leave their signature on the data assimilated and if the model is capable of faithfully reproducing such signatures. Using simulations of the evolution of convective storms in the Great Plains over an active 6-day period, the propagation of initial condition errors to other variables as well as their effect on the accuracy of the forecasts were investigated. Increasing the assimilation time window boosts the ability of assimilation systems to detect a variety of initial condition errors; however, limits to the predictability of convective events impose a maximum assimilation period that is a function of the type of measurements assimilated as well as of the type of errors one tries to correct for. These findings are then used to suggest changes in assimilation approaches to take into account the different predictability times of the model fields constrained by assimilation.

Full access
Juanzhen Sun
and
Hongli Wang

Abstract

The Weather Research and Forecasting Model (WRF) four-dimensional variational data assimilation (4D-Var) system described in Part I of this study is compared with its corresponding three-dimensional variational data assimilation (3D-Var) system using a Great Plains squall line observed during the International H2O Project. Two 3D-Var schemes are used in the comparison: a standard 3D-Var radar data assimilation (DA) that is the same as the 4D-Var except for the exclusion of the constraining dynamical model and an enhanced 3D-Var that includes a scheme to assimilate an estimated in-cloud humidity field. The comparison is made by verifying their skills in 0–6-h quantitative precipitation forecast (QPF) against stage-IV analysis, as well as in wind forecasts against radial velocity observations. The relative impacts of assimilating radial velocity and reflectivity on QPF are also compared between the 4D-Var and 3D-Var by conducting data-denial experiments. The results indicate that 4D-Var substantially improves the QPF skill over the standard 3D-Var for the entire 6-h forecast range and over the enhanced 3D-Var for most forecast hours. Radial velocity has a larger impact relative to reflectivity in 4D-Var than in 3D-Var in the first 3 h because of a quicker precipitation spinup. The analyses and forecasts from the 4D-Var and 3D-Var schemes are further compared by examining the meridional wind, horizontal convergence, low-level cold pool, and midlevel temperature perturbation, using analyses from the Variational Doppler Radar Analysis System (VDRAS) as references. The diagnoses of these fields suggest that the 4D-Var analyzes the low-level cold pool, its leading edge convergence, and midlevel latent heating in closer resemblance to the VDRAS analyses than the 3D-Var schemes.

Full access
Juanzhen Sun
and
Andrew Crook

Abstract

The adjoint method to retrieve the three-dimensional wind and thermodynamic fields is applied to single-Doppler observations of a gust front measured during the Phoenix II experiment. This method uses a fluid dynamics model and its adjoint, and combines the retrieval with data assimilation into the prediction model. The wind and thermodynamic variables are determined by minimizing the difference between the model solution and the observations. Experiments are conducted first with radial velocity alone and then with both radial velocity and reflectivity to examine the quality of the retrieval with respect to radar location, boundary conditions, length of assimilation, data filtering, smoothing enforced by penalty functions, and model accuracy. Verification of these experiments is provided by a dual-Doppler analysis.

Test results show that the adjoint method is able to retrieve the wind and thermodynamic fields of the gust front. The retrieved horizontal wind agrees very well with the dual-Doppler analysis. This study also demonstrates that the adjoint method has the ability to resolve finescale features along the loading edge of the gust front.

Full access
Eunha Lim
and
Juanzhen Sun

Abstract

A Doppler velocity dealiasing algorithm is developed within the storm-scale four-dimensional radar data assimilation system known as the Variational Doppler Radar Analysis System (VDRAS). The innovative aspect of the algorithm is that it dealiases Doppler velocity at each grid point independently by using three-dimensional wind fields obtained either from an objective analysis using conventional observations and mesoscale model output or from a rapidly updated analysis of VDRAS that assimilates radar data. This algorithm consists of three steps: preserving horizontal shear, global dealiasing using reference wind from the objective analysis or the VDRAS analysis, and local dealiasing. It is automated and intended to be used operationally for radar data assimilation using numerical weather prediction models.

The algorithm was tested with 384 volumes of radar data observed from the Next Generation Weather Radar (NEXRAD) for a severe thunderstorm that occurred during 15 June 2002. It showed that the algorithm was effective in dealiasing large areas of aliased velocities when the wind from the objective analysis was used as the reference and that more accurate dealiasing was achieved by using the continuously cycled VDRAS analysis.

Full access
F. Zhang
,
Chris Snyder
, and
Juanzhen Sun

Abstract

The ensemble Kalman filter (EnKF) uses an ensemble of short-range forecasts to estimate the flow-dependent background error covariances required in data assimilation. The feasibility of the EnKF for convective-scale data assimilation has been previously demonstrated in perfect-model experiments using simulated observations of radial velocity from a supercell storm. The present study further explores the potential and behavior of the EnKF at convective scales by considering more realistic initial analyses and variations in the availability and quality of the radar observations. Assimilation of simulated radial-velocity observations every 5 min where there is significant reflectivity using 20 ensemble members proves to be successful in most realistic observational scenarios for simulated supercell thunderstorms, although the same degree of success may not be readily expected with real observations and an imperfect model, at least with the present EnKF implementation. Even though the filter converges toward the truth simulation faster from a better initial estimate, an experiment with the initial estimate of the supercell displaced by 10 km still yields an accurate estimate of the storm for both observed and unobserved variables within 40 min. Similarly, radial-velocity observations below 2 km are certainly beneficial to capturing the storm (especially the detailed cold pool structure), but in their absence the assimilation scheme can still achieve a comparably accurate estimate of the state of the storm given a slightly longer assimilation period. An experiment with radar observations only above 4 km fails to assimilate the storm properly, but, with the addition of a hypothetical surface mesonet taking wind and temperature observations, the EnKF can again provide a good estimate of the storm. The supercell can also be successfully assimilated in the case of radar observations only below 4 km (such as those from the ground-based mobile radars). More frequent observations can help the storm assimilation initially, but the benefit diminishes after half an hour. Results presented here indicate that the vertical resolution and the uncertainty of observations, for the typical range of most of the observational radars, would have little impact on the overall performance of the EnKF in assimilating the storm.

Full access
Juanzhen Sun
and
N. Andrew Crook

Abstract

The adjoint technique for retrieval of the thermodynamic fields is compared with the traditional technique of Gal-Chen and Hane. The comparison is performed using both Doppler radar observations and simulated data. The real dataset is a gust-front case observed during the Phoenix II experiment. The simulated data are from a numerical experiment of a collapsing cold pool. In the simulated data study, we assume that observations of the horizontal velocity are available, either from a dual-Doppler synthesis or from a single-Doppler retrieval. Tests are performed on data that have been degraded in various ways to replicate real data. These tests include the sensitivity to the temporal sampling frequency, random error, spatially correlated error, and divergent/rotational error in the forcing terms. In most of the cases examined, it is found that the adjoint method is able to retrieve the buoyancy field more accurately than the traditional technique.

Full access