Abstract

The problem of four-dimensional data assimilation in the tropics has been studied using a limited-area primitive equation model. Of prime concern is the relative importance of different update variables and their impact on data assimilation. Five new experiments complement a set of ten previously reported experiments that investigate the feasibility of four-dimensional data assimilation in the monsoon region using only the wind observations. In addition to assessing the relative importance of update variables, the present study investigates the role of model physics in data assimilation.

The assimilation experiments are carried out for the onset vortex case of the 1979 Indian summer monsoon for which many special FOGE/MONEX datasets are available. The assimilation-forecast system for all of the experiments comprises a 12-h assimilation phase followed by a 24-h forecast period. In all experiments, updating is done via the Newtonian nudging approach which, in our previous study, was found to be more effective than other methods of updating.

It is found that at least for this dataset, the wind data were wore valuable than the temperatures. Although temperature assimilation alone had some unexpected positive results, it did not offer appreciable improvement over wind-only assimilations when the two variables were inserted together. On the other hand, a combination of wind and moisture data produced the most positive results. This confirms the importance of wind and moisture data in the tropics. Finally, it has been found that the incorporation of physical parameterizations during the assimilation period is important for a proper spinup of the model and its smooth transition into the forecast stage.

Abstract

A limited primitive equation model has been used to study the feasibility of four-dimensional data assimilation in the monsoon region and, further, to study the applicability of several assimilation techniques currently being employed in global models. The two fundamental objectives of this research are

(i) to understand how the model atmosphere responds to the insertion of asynchronous data and its impact on the assimilation-prediction cycle, and

(ii) to determine what assimilation strategies work best for limited-area models in the tropics.

A sequence of ten assimilation experiments are performed using different update procedures; all insertions are carried out with only the wind observations. The model is initialized with the ECMWF FGGE level III-b data for the onset vortex case of 17 June 1979, and assimilations are carried out using the summer MONEX level II-b data during the first 12 hours. From these assimilated states, 24-h forecasts are then made.

The results lend support to the premise that the required initial conditions can be obtained by the process of four-dimensional updating of the prognostic variables. The results also clearly demonstrate the superiority of the continuous assimilation approach via Newtonian nudging over that by indirect insertion. Furthermore, the insertion shocks are significantly minimized by assimilating only the rotational component of the winds. On the other hand, the application of noise control measures only marginally alleviate the insertion shocks accompanying continuous indirect insertion.

Abstract

The current major expansion in observational capability of line National Weather Service is principally in the volume of asynchronous data rather than synchronous observations at the standard synoptic times. Hence, the National Meteorological Center is considering a continuous data assimilation system to replace at some time the intermittent system now used by its regional and global operational models.

We describe this system, based on the Newtonian relaxation technique, as developed for the eta model. Experiments are performed for the first intensive observing period of the Genesis of Atlantic Lows Experiment (GALE) in January 1986, when strong upper-level cyclogenesis occurred, with a pronounced tropopause fold but only modest surface development. The GALE level IIIb dataset was used for initializing and updating the model.

Issues addressed in the experiments include choice of update variable, number, and length of update segments; need for updating moisture and surface pressure information; nudging along boundaries; and noise control. Assimilation of data from a single level was also studied.

Use of a preforecast assimilation cycle was found to eliminate the spinup problem almost entirely. Multiple, shorter assimilation segments produced better forecasts than a single, longer cycle. Updating the mass field was less effective than nudging the wind field but assimilating both was best. Assimilation of moisture data, surprisingly, affected the spinup adversely, but nudging the surface pressure information reduced the spurious pillow effect. Assimilation of single-level information was ineffective unless accompanied by increased vertical coupling, obtained from a control integration.

Abstract

A conjugate-gradient variational blending technique, based on the method of direct minimization, has been developed and applied to the problem of initialization in a limited-area model in the summer monsoon region. The aim is to blend gridded winds from a high-resolution limited-area analysis with a lower-resolution global analysis for use in a limited-area model that uses the, global analyst for boundary conditions. The ability of the variational matching approach in successfully blending meteorological analyses of varying resolutions is demonstrated. Reasonable agreement is found between the blended analyses and the imposed weak constraints, together with an adequate rate of convergence in the unconstrained minimization procedure. The technique is tested on the 1979 onset vortex vortex case using data from the FGGE Summer MONEX campaign. The results indicate that the forecasts made from the variationally matched analyses show positive impact and perform better than those from the unblended analyses.

Abstract

A winter snowstorm developed on 10–11 February 1988 over the midwestern United States and produced several inches of snowfall locally over east-central Illinois. Analysis of the mesoscale organization of the storm revealed the presence of complex banded structure throughout its 17-h evolution. Three distinct types of mesoscale precipitation bands were identified during the course of the storm using a 10-cm Doppler radar as part of the University of Illinois Winter Precipitation Program. The bands had different orientations, directions of movement, relationships to synoptic-scale frontal zones, and mechanisms for development.

The mesoscale organization of this storm system is reviewed. Mesoscale, synoptic-scale, and Doppler analyses of the storm structure are presented. The role of boundary-layer convergence, conditional symmetric instability, and frontogenetical forcing in the formation and maintenance of the different mesoscale precipitation bands is discussed.

Abstract

On 14 February 1992, a long-lived moderate-amplitude mesoscale gravity wave formed in Kansas during the Storm-scale Operational and Research Meteorology-Fronts Experiment Systems Test (STORM-FEST). Wave formation was evident in correlated surface pressure and wind fields. The wave of depression, accompanied by a weak rainband, tracked across the state. A wealth of data was collected on the mature wave as it passed over the STORM-FEST dual-Doppler domain. However, the mechanism of genesis remained difficult to ascertain, since wave formation occurred in a region of less comprehensive observations.

The genesis of the STORM-FEST gravity wave is successfully simulated using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (Penn State–NCAR) Mesoscale Model (MM5), which was run at 6-km grid spacing in the innermost domain. The lee cyclone movement, dry airmass development, and gravity wave formation over Kansas were successfully captured by the model. Results presented here indicate that evaporative processes associated with a rainband resulted in subsidence warming and depression of the underlying warm-frontal inversion. The reduced inversion height produced surface pressure falls, the surface manifestation of a developing gravity wave. Numerical experiments with and without evaporative processes have isolated the key importance of evaporatively driven downdrafts in wave genesis. A conceptual model of the development and evolution of the wave is presented that is consistent with both observations and the findings of the numerical experiments.

Abstract

This paper presents observations of the finescale three-dimensional kinematic and thermodynamic structure of a long-lived mesoscale gravity wave that occurred on 14–15 February 1992 during the Storm-scale Operational and Research Meteorology-Fronts Experiment Systems Test (STORM-FEST). In Part I of this series of papers, it was shown that the wave was generated just behind the leading edge of an advancing dry air mass that originated as a foehnlike downslope flow over the Rocky Mountains of southern Colorado and New Mexico. Surface pressure signatures of wave motion began as the dry air mass ascended a warm front east of a lee cyclone and a rainband developed along its leading edge. The wave and rainband intensified, remaining near the leading edge of the dry air mass as the dry air mass advanced northeastward over the warm frontal inversion. After 7 h of evolution, the leading edge of the dry air mass passed over the dual-Doppler network in northeast Kansas. It was at this point in the evolution that the relationships among the mesoscale gravity wave, the leading edge of the dry air mass, and the rainband were determined using dual-Doppler kinematic and thermodynamic retrieval analyses. These analyses demonstrate temporal and spatial consistency, and were verified to the degree possible with independent datasets.

Over the dual-Doppler network, the leading edge of the dry air mass was characterized by a strong gradient in horizontal momentum. Convection occurred at the leading edge of the dry air mass. Within the dry air mass, a four-quadrant region of divergence and convergence was observed. Associated with these patterns of convergence and divergence were downdrafts and updrafts. The downdraft contributed to the depression of the warm frontal inversion, creating the wave signature in the surface barograms. The downdraft branch of the circulation, which acted to depress the inversion height, was associated with net cooling due to evaporation of precipitation above the inversion and with circulations associated with the deceleration of air at the leading edge of the advancing dry air mass. The wavelength of the wave over the Doppler domain was determined by the scale of the internal circulations within the dry air and the scale of the convection.

Pressure, p′, and virtual potential temperature, θ ^′ _υ , perturbations were generated by the advancing dry air mass and associated convection. Strong horizontal p′ gradients were associated with both the abrupt deceleration of air at the leading edge of the dry air mass and with the divergent flow at the top of the updraft. Vertical p′ gradients were associated with convective updrafts and downdrafts. The total p′ distribution was the sum of these effects. Because the convection was weak, the vertical accelerations were small, and the retrieved distribution of θ ^′ _υ was such that the buoyancy force exactly balanced the vertical perturbation pressure gradient force.

A conceptual model for the generation and maintenance of this wave is proposed based on the analysis of the observations and the larger-scale measurements presented in Part I.

Abstract

A severe freezing rainstorm produced as much as 4.5 cm of freezing rain during an 18-h period at Champaign, Illinois, on 14–15 February 1990, resulting in over $12 million in damage, week-long power outages, and a federal disaster declaration. The ice storm occurred during the University of Illinois Winter Precipitation Program based in Champaign. The early mesoscale evolution of this storm was documented for several hours with a 10-cm Doppler radar and Cross-chain Loran Atmospheric Sounding System soundings launched every 3 h. The freezing rain event occurred when convective bands developed over a slow-moving warm front during a period of strong overrunning. The strongest convection developed in a period of about 1 h, with a narrow elongated band northwest of the radar producing very heavy sleet and a band just south of the radar producing heavy freezing rain, along with in-cloud lightning.

An analysis of conditional symmetric instability yielded no evidence that centrifugal accelerations were important to the development of convection in this storm. Frontogenetic forcing was strongest several hours before the development of the bands but apparently was also insufficient to trigger convection until the local atmosphere became marginally unstable to upright convection. The transition from a conditionally stable to an unstable atmosphere in the vicinity of the bands was directly associated with locally strong warm advection above the warm frontal surface.

Forecast guidance, including the nested grid model (NGM) thickness, precipitation, and 850-mb temperature forecasts, and model output statistics of both the limited fine mesh (LFM) model and the NGM all predicted that the warm front would progress northward and that freezing rain would convert to rain before significant glaze accumulations occurred. Forecasts of midtropospheric parameters such as 1000–500-mb thickness and 850-mb temperature indeed verified; however, surface temperature forecasts were significantly in error, with errors ranging from 5° to 10°C during the period of heaviest glaze accumulation. The observed surface temperature never rose above 0°C during the period of ice accumulation or throughout the following day. The isothermal conditions observed during and after the storm appeared to be the result of sublimation and melting of ice that had accumulated on surface objects. The available evidence suggested that ice sublimation and melting, in addition to cooling the boundary layer, maintained a small wedge of cold air at the surface over which warmer air rose as it advected northward. The result of ice sublimation and melting was to retard the movement of the surface warm front, although warm air aloft was free to move over the narrow wedge of cooled surface air. By maintaining the surface temperature near 0°C, diabatic processes extended the duration of time that heavy glaze accumulations remained on trees and wires, allowing more damage to occur.

Abstract

Two large-amplitude gravity waves were observed over the midwestern United States on 5 and 14 January 1989 during the University of Illinois Winter Precipitation Program. On both days, an extensive amount of data was recorded, including data from two radars and a radiosonde facility. The waves originated near Missouri, registered pressure fluctuations as large as 10 mb, and produced distinct precipitation bands along their updraft regions.

The waves were long-lived and maintained their identity over 1000 km, a distance several times their wave-lengths. The synoptic features at the surface were dissimilar. A deep cyclone was present on 5 January, while a leeside trough was present on 14 January. However, the middle- and upper-tropospheric flow patterns were similar. In both cases, the axis of a trough was immediately upstream of the gravity-wave genesis area and a jet streak had just propagated through the base of the trough toward a downstream ridge. Soundings taken near the gravity waves were remarkably similar, with both soundings showing a surface inversion capped by a deep layer of near-neutral stability. However, the relationship between the location of the gravity wave and the region of large-scale precipitation differed in the two cases. The 5 January wave occurred at the back edge of the precipitation associated with a comma cloud, while the wave on 14 January was observed at the leading edge of the synoptic-scale precipitation region.

The gravity wave had the structure of a solitary wave of elevation on 5 January, while it appeared as an undular bore with an embedded pressure jump on 14 January. A critical level, with small Richardson numbers, was present in both the cases. A well-defined duct, formed by an inversion below and critical level above, contributed to the maintenance of waves. Shearing instability and geostrophic adjustment were the likely generation mechanisms, though it was difficult to discount the role of convection.

Abstract

The successful deployment of many different observing systems during the summer Monsoon Experiment of 1979 provides a unique opportunity to perform extensive observing system experiments. These numerical studies, accomplished here with a ten-level, limited-area primitive equation model, allow the assessment of the value of individual or combined observing systems to the model's four-dimensional data assimilation system as well as to its subsequent forecasts. The specific objectives of this work include the investigation of (i) the relative merit of ten different data platforms, (ii) the relative role of wind and mass field data, (iii) the effect of different vertical distributions of single-level wind data, and (iv) the dynamical response of the model to different modes of data insertion.

Eight experiments are summarized, all of which involved a 12-h data assimilation period based on the Newtonian relaxation procedure followed by a 36-h forecast. Predictions using all of the data produced very good forecasts of the June 1979 onset vortex over the Arabian Sea. The dropwindsonde data were found to be most responsible for this success, primarily because they resolve the rotational modes of the system and cover a significant depth of the troposphere. While the winds were more important, the dropsonde thermodynamic data were beneficial. All datasets, when tested individually, had a positive impact on the forecasts. When used in combination, however, some datasets became less important or even redundant. The influence of satellite winds was enhanced greatly by spreading the wind increments over a larger vertical depth. It is shown that the dynamical response of the model to the various distributions and amounts of new data is consistent with geostrophic adjustment theory and provides guidance for future observing system.