CALJET Goals and Experimental Design 25 September 1997
F. M. Ralph, NOAA/ERL/Environmental Technology Laboratory
1. Motivation
Land falling cyclones cause extensive damage along the US west coast by producing >8 inches of rain in 24 h (e.g., 1/10/95, 3/10/95, 11/20/96, and 1/1/97), and >50 m/s surface winds (e.g., 12/11/95 and 11/20/96) as they come ashore. Although accurate warnings of these events are needed to alert the public, military, and emergency services, the timing is often off by several hours and the location by 300 km along shore. Rhea (1996) showed that 6-h QPF from the NWS California and Nevada River Forecast Center for 2 heavy precipitation periods had a significantly lower correlation with observed precipitation than did 24-h QPF. The need for improvements in this type of forecast is reflected in the objectives of the National Weather Service and the USWRP. These views are also supported by recommendations of operational weather forecasters from NOAA and the military, and from river forecasters, that short-range forecasts (0-6 h) of mesoscale winds and precipitation are critical forecast problems along the United States west coast for which improvements are needed. A key player involved in creating this severe weather is the prefrontal low-level jet (LLJ), within extratropical cyclones. This jet often contains much moisture, and can cause extreme coastal rains when it encounters mountains (e.g., Smith 1979; Banta 1990). Damaging coastal winds can be created either by low-level blocking, or by mountain wave behavior. Operational forecasters have long used the LLJ's strength, position, and moisture content to help
predict precipitation rates and coastal winds. Because significant errors in these key parameters are
possible due to the limited amount of data available offshore and aloft along the coast, it appears that
improved observations of the LLJ could improve forecasts. Adjoint techniques have also been used to
identify limited regions where additional data could help forecasts downstream (e.g., Pailleux 1990;
Rabier et al. 1994; Langland et al. 1995; Bao and Bresch 1996; Stensrud et al. 1996). These
approaches suggest that improved knowledge of the position, and strength of the LLJ offshore, 0-18 h
before landfall could allow mesoscale numerical models to better predict the location and intensity of
flooding rains and damaging winds. Predictions of the overall motion and development of the entire
cyclone will also affect the location and timing of the heavy rains, and because the LLJ is a key player in
the self-development process in cyclogenesis through thermal advections, better observations of the
LLJ could also improve the overall cyclone prediction (Langland et al. 1995).
2. Primary goals
The experiment focuses on improving prediction and understanding of key mesoscale phenomena in the
coastal zone of the western United States related to the land fall of the LLJ in winter storms. Data will
be gathered offshore up to 12 h before land fall, and along shore during land fall. The specific goals are
as follows:
3. Relationship to U.S. Weather Research Program and U.S. National Weather Service
scientific objectives
CALJET addresses two of the three initial scientific foci of the U.S. Weather Research Program
(Emanuel et al. 1995, Dabbert et al. 1996), which encourage studies related to:
Similarly, CALJET addresses two of the major priorities set within the National Weather Service aimed
at improving operational weather forecasts (Uccellini 1996). Within the Office of Meteorology's 10-y
Strategic Plan it is recommended that "The NWS science priorities include studies of processes
involved with
CALJET also directly addresses two major NOAA programs regarding the importance and mix of observations: the North American Observing System (NAOS), and Pacific Coastal Forecast System (Pac-CFS) programs. Pac-CFS has deployed dozens of drifting buoys off the U.S. west coast, and has supported the deployment and testing of a profiler on the Farallon Islands west of San Francisco that will be included in our research. It has also been noted by the USWRP (Dabbert et al. 1996) that advances in both observational
techniques and numerical simulations have created a unique opportunity to improve weather forecast
accuracy through integrated studies of physical processes, predictability, and observing system
capabilities. This conclusion was found to be especially relevant to the subjects of coastal meteorology
and mountain weather (Rotunno et al. 1996; Smith et al. 1996).
4. Relationship to the COAST and FASTEX experiments
Two experiments on cyclones help provide the experience needed to carry out CALJET:
COAST explored the modification of mesoscale structures by coastal mountains in land falling extratropical cyclones in the Pacific Northwest using the P-3 (Bond et al. 1996). The experiment successfully measured the land falling phase of several cyclones, focusing mostly on the modification of fronts, and the distribution of precipitation and microphysical conditions related to orographic precipitation enhancement. Although COAST contained a significant mesoscale modeling component, and the P-3 provided invaluable data along and on shore, little data was gathered far enough offshore to aid in initializing the models (Colle and Mass 1996). FASTEX aimed at improving understanding and prediction of secondary cyclogenesis in the North Atlantic (Joly et al. 1995), and used the P-3 and other research aircraft. Observations of both the precursors to the storms and the storms themselves were made. A key objective was to test the use of adaptive observing strategies in improving 24-48 h weather predictions. This approach was based on techniques that have shown strong sensitivity of the cyclone forecasts to a few observations from within a limited domain marking a dynamically sensitive part of the flow (e.g., Pailleux 1990, and Rabier et al. 1994). The data from several research aircraft and their dropsondes were incorporated directly into real-time numerical weather predictions. CALJET will capitalize on these techniques, i.e., targeting (FASTEX) of the LLJ 0-18 h before land
fall, and documenting the land falling phase of the LLJ for use in verifying the mesoscale simulations
(COAST). Because CALJET is sharply focused on the LLJ it is possible to carry out the planned
objectives without the much larger facilities required for FASTEX.
5. Experimental design
CALJET will take place from 1 December 1997 to 31 March 1998, when numerous wind profilers and
a cloud radar will be in place along the coast. A special observing period will focus on 18 January to
28 February 1998, when NOAA's P-3 and the Univ. of Oklahoma's Doppler on Wheels will be
operated out of Monterey, California. A weather briefing will be presented each day at the Naval
Postgraduate School. The dates of the experiment correspond to the time of maximum monthly
averaged precipitation along the California coast (Fig. 1), which peaks in early January in northern
California, and in early February in central California. The experiment incorporates both numerical
modeling and observations.
a) Observations CALJET will make use of a wide variety of special observing systems to augment the existing operational network. The special observations, which are shown in Fig. 2 and listed in Appendix 1, will include additional offshore data from the NOAA P-3 (GPS dropsondes, dual-Doppler-capable tail radar, gust probe, C-SCAT, etc...), drifting buoys, and three island-mounted
Fig. 1. Climatology of precipitation at sites in northern (Eureka), central (San Francisco), and southern
(Santa Barbara) California. The data are derived from the California Water Supply Outlook, and
represent 30-day average rainfall over 70 y. Precipitation (inches) and central dates (mm/dd) of each
30-d average are marked.
wind profilers. The coastal network will be enhanced, ultimately including 20 wind profilers with RASS (mostly boundary layer profilers continuously measuring winds from 0.3 to 3 km AGL, and virtual temperature from 0.3 to 1.0 km AGL). GPS dropsonde data from the P-3 will be sent hourly via satellite for use in nowcasting and numerical modeling. Profiler and RASS data will be sent to NCEP via NOAA's Forecast Systems Laboratory using procedures recently established for providing boundary layer profiler data from dozens of sites around the country to NCEP for possible use in operational numerical weather prediction. The drifting buoys will report via the ARGOS satellite roughly 4 times per day. Operational surface and sounding networks and satellite-derived sea surface winds and cloud-tracked winds will be included.
i) Three special observing areas (SOA) will be created along the coast.
Fig. 2. A map of CALJET's coastal domain showing key observing systems deployed for the
experiment. Coastal and drifting buoys are not shown. Topography is shown by shading: light gray is
0.3 - 1.0 km MSL, and dark gray is >1.0 km MSL.
ii) The P-3 "Pre-land fall" flight strategy (Fig. 3a): As storms approach the California and Oregon coasts, potential IOPs will be identified based on both operational and mesoscale experimental model guidance, as well as satellite (cloud, water vapor, surface winds, and cloud-tracked winds) and buoy observations. If the available evidence suggests that a LLJ maximum associated with a cyclone, or with secondary waves on the polar front is approaching the California or Oregon coasts, an IOP will be called. It will be necessary to make this decision roughly 12-24 h before the flight would begin, i.e., 24-36 h before the LLJ makes land fall. This decision will be aided by quasi-operational 36-h mesoscale simulations from the Naval Research Laboratory's Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS; Hodur 1997) and the Penn State/NCAR MM5 (Grell et al. 1994) using 12-36-km grid sizes. These simulations will be initialized with operational global-scale models, and experimental data such as from the P-3, drifting buoys, satellite image-tracked winds (Veldon et al. 1997), and profilers when possible. These runs should be completed by 9 AM PDT each day, which is 1 h before CALJET's daily weather briefing. Predictions from both the operational and experimental models would likely be improved through cooperation with the E-PAC experiment that is being designed to explore targeted observing for 48-72 h forecasts using dropsonde data from two C-130 aircraft over the eastern Pacific ocean for two weeks sometime during late January or February 1998. CALJET's P-3 flights will target the LLJ and its environs roughly 1000 km offshore (Fig. 3a). Dropsondes will be deployed over a wide enough area to include not only the LLJ, but also features that contribute to the LLJ and its evolution on the mesoscale. These include the cold pool behind the surface cold front, the region ahead of the LLJ, and the warm front. After close examination of the operational and experimental data before take off, the P-3's radar, in-situ, dropsonde, and C-SCAT (surface winds beneath the aircraft) data will help guide the flight. A typical flight will initially emphasize 400 mb flight legs crossing the warm front, the warm sector, the LLJ, and the cold front. This provides a mapping for later low-level flight legs, and gives the most warning to operational forecasters based on transmission of the dropsonde data via satellite. The low-level flight legs emphasize radar and in-situ data gathering to precisely measure the LLJ jet position, strength, and moisture content, as well as the boundary layer.
Fig. 3. a) Schematic description of P-3 observing strategy for the pre-land fall phase of the
experiment. The pre-cold frontal LLJ and frontal positions shown are loosely based on analysis of a
storm on 6-7 January 1995 (Doyle 1997) and approximate the meteorological scales and structures
involved. The one-way ferry time of the P-3 at 400 mb from Monterey is shown (light dashed). Total
flight time is 10 h. b) The observing strategy for the land falling phase. The coast and coastal
mountains (~0.8 km tall), the Doppler on Wheels (radar dishes), wind profilers (solid dots), a vertically
pointing 3-cm radar (+) and a GPS integrated water vapor sensor (open box) are shown. A cold
frontal rain band offshore and the region of orographic precipitation enhancement are shown with rain
rate contours of 2 and 10 (shaded) mm h-1. The 100 km range ring (light dashed) of each DOW is
marked (light dashed).
iii) Observing strategy for the land falling phase (Fig. 3b): If the pre-land fall flight is close enough to shore that some ferry time can be saved, and the LLJ is already making land fall, then the P-3 could also observe conditions at land fall at the end of the flight. This would require 2-4 h after working farther offshore with the pre-land fall flight strategy. Another scenario would consist of a pre-land fall flight in a slower-moving storm, followed by a coastal flight after the required 16 h of down time for the P-3 crew. This logistical limitation of the experimental design is partially overcome by the deployment of the two
DOW radars and a mobile sounding system along the coast. These tools could be deployed roughly 12
h before land fall, and could respond rapidly to changing conditions or errors in the forecast position of
heavy precipitation by moving north or south along the coast. [In the event of severe flooding the
coastal roads may be blocked, but several routes exist across the coastal mountains that could allow
these tools to continue documenting the event.] In an ideal case the deployments will capture the
passage of the heavy precipitation across the microphysical observing area (SOA-1). This is shown in
Fig. 3b, where the P-3 is used to first measure conditions well upstream of the coast, and then to
perform stacked flight legs across the mountain to measure microphysical conditions, and finally to
measure these conditions along shore just a few kilometers offshore. This last objective would also
provide turbulence and flux measurements in the boundary layer just offshore. The boundary layer and
turbulence measurements are motivated by the fact that surface moisture fluxes contribute to the
moisture cycle. Also, recent research has shown the impact of differential surface friction between land
and sea versus topographic effects (Doyle and Warner 1993; Doyle 1997), and the sensitivity of
cyclones to surface fluxes and boundary layer parameterizations is well known (Kuo et al. 1991;
Persson et al. 1995; Doyle 1995).
b) Numerical modeling
i) Pre-experiment simulations to help guide the design of CALJET's observing strategies The use of variational methods in numerical modeling has grown significantly in the last decade.
Varational methods, such as the adjoint technique, have been developed for data assimilation (e.g.,
Lewis and Derber 1985), model sensitivity (e.g., Errico and Vukicevic 1992), and model tuning (e.g.,
Derber 1989). Of direct relation to the proposed work is the application of the dry version of an
adjoint of a tangent-linear model developed by Errico et al. (1994) as part of the Mesoscale Adjoint
Modeling System (MAMS). The dry version of the MAMS nonlinear model is based on the model
designated MM4 developed jointly at the Pennsylvania State University and NCAR (Anthes et al.
1987). This adjoint has been used to study model sensitivities to initial/boundary conditions (e.g.,
Errico and Vukicevic 1992) and model parameters (e.g., Bao and Errico 1997) in which the sensitivity
of a defined functional of model state at some point in time and space with respect to perturbations of
model state and parameters at earlier times in the model simulation. Errico and Vukicevic (1992) show
that the tangent-linear model accurately produces the evolution of perturbations in comparison with the
full nonlinear version of the model. Therefore, the use of the adjoint of the tangent-linear model prior to
the development of heavy precipitation event should provide useful and interesting results about the
model sensitivity to various initial condition perturbations.
Working Hypothesis: We assume that the development of heavy rainfall is the dominant, energetically active mode in the model simulation. We alter this mode as guided by the adjoint sensitivity analysis and our understanding of how heavy rainfall processes evolve both in reality and in the model framework. Therefore, the technique that we use for identifying the sensitivity area in the initial condition is based upon the following hypothesis: By modifying the initial condition based upon the adjoint sensitivity analysis with respect to known mesoscale flow structure that influences heavy rainfall development (e.g., low-level jet, model resolvable-scale vertical motion, inversion strength, and low-level temperature), such that the atmosphere near the region of heavy rainfall development in both space and time is altered, the model predictability can be improved. Methodology: A control simulation of California's New Years Day storm of 1997 will be performed using the full physics version of the Penn State/NCAR mesoscale model. We will then produce simulations from the same initial condition with the dry version of the model. It is expected that the differences between the dry version and the full physics version will be small before the onset of heavy rainfall. The dry version of the MAMS adjoint model can then be used to determine the sensitivity of the nonlinear model output to changes in the initial conditions. In particular, we use the adjoint model to guide us in altering the initial condition in such a way as to change the model atmosphere near the initial region of heavy rainfall development, while raining within the error characteristics of the observations. The altered initial condition is expected to provide better predictability of mesoscale flow structure that is crucial to the heavy rainfall development in the model. In this approach, we use the adjoint technique to selectively change the model initial state to produce the greatest influence on the subsequent development of heavy rainfall. In addition, the MM5 adjoint will also be employed to explore the sensitivity of conditions at land fall to
conditions offshore 12 h earlier. This is illustrated in Fig. 4, which shows that the vertical vorticity
associated with the LLJ at the time of its land fall near 0000 UTC 1 January 1997 was sensitive to
errors in the zonal (u) wind component at = 0.925 roughly 300 to 1200 km to the south and
southwest 12 h earlier. This region is well within the range of the P-3, and its western portion
corresponds to the LLJ.
Special real-time simulations of MM5 and COAMPS out to 36 h will be performed. They will be
based on large-scale analyzed fields from NCEP's or the Navy's operational global-scale models, but
we will attempt to incorporate as much of the available experimental data as possible: i.e., from drifting
buoys, profilers, special soundings, the P-3's dropsondes, and cloud-tracked winds. These simulations
serve two major purposes: 1) to provide additional guidance to the operational weather forecasters
responsible for issuing warnings to the public about approaching storms, and 2) to help guide the
deployment of the P-3, Doppler on Wheels and special soundings. These mesoscale simulations will
use nested domains with roughly 36 km and 12 km grid sizes. It may be possible to use MM5 to
simulate the larger meso--scale domain (up to 2000 km offshore) with courser grids, and to use
COAMPS to focus on the meso--scale conditions at landfall using finer grids.
iii) Post-experiment sensitivity studies After the experiment these models will be used to explore several research topics that CALJET will
make possible, including the value of offshore data in mesoscale data assimilation, the behavior of the
LLJ at landfall, the role of warm rain processes in this region during heavy precipitation events, and to
assess the accuracy of the boundary layer and moisture cycles in the models.
6. Summary
The CALJET experiment will use the NOAA P-3, tropospheric wind profilers, the University of Oklahoma's Doppler on Wheels, and drifting buoys to measure conditions along and off the California and Oregon coasts during the 1997/98 winter. The goals are to improve quantitative precipitation forecasts (QPF) in land-falling winter storms, and to better understand the underlying physical processes and how they are represented in mesoscale numerical models. The LLJ is a major focus for damaging weather when oceanic winter storms make land fall. Uncertainty in its position, strength, and moisture content just 12 h before land fall contributes to errors in mesoscale QPF. This view is supported by the MM5 adjoint analysis of a strong case from 1997, which also showed that the range of the P-3 is adequate to capture a major event 12 h before land fall (Fig. 4) using the planned pre-land fall flight strategy (Fig. 3a). The data from CALJET's pre-land fall flights will help assess the potential value of such offshore data. If the conclusion is that such data is of value, it would provide motivation to further improve affordable unattended aircraft and buoy-mounted wind profiler technologies that already show some promise. Another unique aspect of CALJET is that we intend for the P-3 dropsonde data to be transmitted nearly real time for direct use by operational weather forecasters. Also, the data will be ingested into mesoscale research numerical models (COAMPS and MM5) in a way that should provide improved short-term model guidance to forecasters. These have the potential to improve operational warnings to the public 0-18 h before the land fall of winter storms during the experiment. In post analysis these data will greatly improve our ability to measure the usefulness of offshore vertical profile data in testing future operational observing system options. The array of 20 coastal wind profilers represents the most extensive coastal deployment of wind profilers with RASS to date, and provides valuable real-time data concerning the position and intensity of the land falling LLJ and other features. As
Fig. 5. a) Composite El Nino index showing the strongest El Ninos of this century. Current conditions
are shown for comparison. (From NOAA's Climate Diagnostics Center, http://www.cdc.noaa.gov).
b) Correlation between warm SST anomalies in the equatorial Pacific ocean and precipitation in the
Continental United States in Jan/Feb/Mar. Contour interval is 10 mm, no zero contour is shown, and
negative contours are dashed. Light and dark shading represent areas of statistically significant
correlations. (From Livezey et al. 1997). The star marks CALJET's operations center and the center
of its coastal observing domain.
pointed out from a study using 13 radiosonde sites along the U.S. west coast during STORMFEST (Hirschberg et al. 1995), such dense and frequent data captured more information on energy fluxes across the coast than the standard operational network, and thus would likely benefit forecasts farther downstream. With its lesser height coverage (1-4 km versus 15 km), but more frequent (1 h versus 6 h) profiles, CALJET's profiler data could possibly improve the 0-72 h forecasts over the rest of the Continental U. S. This experiment builds on the techniques and results of the COAST, and FASTEX experiments, but focuses more sharply on mesoscale prediction of coastal flooding rains and damaging winds. This focus reflects the current objectives of two major research agendas from the U.S. National Weather Service, and the U. S. Weather Research Program. Evidence of a major El Nino in the equatorial Pacific is shown in Fig. 5a, based on a composite of
several El Nino indices produced by NOAA's Climate Diagnostics Center (http://www.cdc.noaa.gov).
It has long been suggested that this anomaly is correlated with wet winters in California. Some of this
inference is based on the winter of 1982/83, which was both the wettest winter on record in much of
California and the strongest El Nino of the century. This correlation has recently been established more
firmly by Livezey et al. (1997), who found a statistically significant correlation between warm equatorial
sea surface temperature anomalies in the Pacific and increased rainfall in California in
January/February/March (Fig. 5b). In fact, the greatest precipitation anomaly is located well within the
CALJET domain, almost precisely where the special microphysical observing area will be sited.
However, recall that any particular realization of this phenomenon can differ substantially from its
statistical norm. Nonetheless, the fact that the current El Nino is likely to be even stronger than the
1982/83 event, and the correlations found by Livezey et al. 1997) indicate that there is a greater than
normal likelihood that this winter will produce many of the storms of the type CALJET is designed to
study.
Participants:
Appendix 1: Special observing systems for CALJET
References
Anthes, R.A., E.-Y. Hsie, and Y.-H. Kuo, 1987: Description of the Penn State/NCAR Mesoscale
model version 4 (MM4). NCAR Technical Note, NCAR/TN-282+STR, 66 pp.
Banta, R. M., 1990: The role of mountain flows in making clouds. Atmospheric Processes Over
Complex Terrain, W. Blumen, Ed., Amer. Meteor. Soc., Boston, MA, 229-283.
Bond, N. A., C. F. Mass, B. F. Smull, R. A. Houze, M.-J. Yang, B. A. Colle, S. A. Braun, M. A.
Shapiro, B. R. Colman, P. J. Neiman, J. E. Overland, W. D. Neff, and J. D. Doyle, 1996: The
Coastal Observation and Simulation with Topography (COAST) experiment. Bull. Amer. Meteor.
Soc., (submitted).
Bao, J.-W., and R.M. Errico, 1997: An adjoint examination of a nudging method for data assimilation.
Mon. Wea Rev., in press.
Colle, B. A., and C. F. Mass, 1996: An observational and modelling study of the interaction of low-level southwesterly flow with the Olympic Mountains during COAST IOP 4. Mon. Wea. Rev., 124,
2152-2175.
Dabbert, W. F., T. W. Schlatter, C. A Davis, R. J. Fleming, R. M. Hodur, G. J. Holland, S. E. Koch,
S. J. Lord, W. D. Neff, R. A. Pielke Sr., L. J. Pietrafesa, D. Raymond, R. B. Smith, and D. S. Zrni,
1996: Research opportunities from emerging atmospheric observing and modeling capabilities: Report
of the second prospectus development team of the U.S. Weather Research Program. Bull. Amer.
Meteor. Soc., 77, 305-323.
Derber, J.C, 1989: A variational continuous assimilation technique. Mon. Wea. Rev., 117, 2437-2446.
Doyle, J. D., and T. T. Warner, 1993: A three-dimensional numerical investigation of a Carolina
coastal low-level jet during GALE IOP 2. Mon. Wea. Rev., 121, 1030-1047.
Doyle, J. D., 1995: Coupled ocean wave/atmosphere mesoscale model simulations of cyclogenesis.
Tellus, 47A, 766-778.
Doyle, J. D., and M. A. Shapiro, 1996: Mesoscale characteristics of a topographically modulated
frontal zone over Norway. Preprints, 7th Conf. on Meosocale Processes, 9-13 September, Reading,
U.K., 319-321.
Emanuel, K., D. Raymond, A. Betts, L. Bosart, C. Bretherton, K. droegemeier, B. Farrell, J. M.
Fritsch, R. Houze, M. Le Mone, D. Lilly, R. Rotunno, M. Shapiro, R. Smith, and A. Thorpe, 1995:
Report of the first prospectus development team of the U.S. Weather Research Program to NOAA
and the NSF. Bull. Amer. Meteor. Soc., 76, 1194-1208.
Errico, R.M., and T. Vukicevic, 1992: Sensitivity analysis using an adjoint of the PSU-NCAR mesoscale model. Mon. Wea. Rev., 120, 1644-1660.
Errico, R.M., K. Raeder and T. Vukicevic, 1994: Mesoscale adjoint modeling system: version 1.
NCAR Technical Note, NCAR/TN-410+IA, 224 pp.
Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth-generation Penn
State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR.
Hirschberg, P. A., R. J. Lind, S. J. Bolduc, and R. L. Elsberry, 1995: The west coast Picket fence
experiment during STORMFEST. Bull. Amer. Meteor. Soc., 76, 1741-1757.
Hodur, R. M., 1997: The Naval Research Laboratory's Coupled Ocean/Atmosphere Mesoscale
Prediction System (COAMPS). Mon. Wea. Rev., 125, 1414-1430.
Joly, A., K. A. Browning, A. J. Thorpe, Y. Lemaitre, and J. P. Cammas, 1995: FASTEX: Outline of
the field programme. Report available from FASTEX Project Office.
Kuo, Y.-H., R. J. Reed, and S. Low-Nam, 1991: Effects of surface energy fluxes during the early
development and rapid intensification stages of seven explosive cyclones in the western Atlantic. Mon.
Wea. Rev., 119, 457-476.
Langland , R. H., R. L. Elsberry, and R. M. Errico, 1995: Evaluation of physical processes in an
idealized extratropical cyclone using adjoint sensitivity. Quart. J. Roy. Meteor. Soc., 121, 1349-1386.
Lewis, J.M., and J.C. Derber, 1985: The use of adjoint equations to solve a variational adjustment
problem with advective constraints. Tellus, 37, 309-327.
Livezey, R. E., M. Masutani, A. Leetmaa, H. Rui, M. Ji, and A. Kumar, 1997: Teleconnective
response of the Pacific-North American region atmosphere to large central equatorial Pacific SST
anomalies. J. Climate, 10, 1787-1820.
Pailleux, J., 1990: The impact of North Atlantic TEMPSHIP radiosonde data on the ECMWF analysis
and forecast. ECMWF Research Department Technical Memorandum No. 174, 36 pp.
Persson, P. O. G., J.-W. Bao, and S. D. Burk, 1995: Tests of the Burk-Thompson PBL scheme.
Abstracts, Fifth PSU/NCAR Mesoscale Model Users Workshop, Aug.14-16, Boulder, CO.
Rabier, F., E. Klinker, P. Courtier, and A. Hollingswoth, 1994: Sensitivity of twoday forecast errors
over the Northern Hemisphere to initial conditions. ECMWF Research Department Technical
Memorandum No. 203, 38 pp.
Rhea, J. O., 1996: An objective orographically based QPF aid for California. National Weather
Service Western Region Technical Attachment No. 96-02, 13 pp.
Rotunno, R., 1992: Coastal Meteorology. National Academy Press, 99 pp.
Rotunno, R., L. J. Pietrafesa, J. Allen, B. Coleman, C. Dorman, C. Krietzberg, S. Lord, M. McPhee,
G. Mellor, C. Mooers, P. Niiler, R. Pielke Sr., M. Powell, D. Rogers, J. Smith, and L. Xie, 1996:
Report of the third prospectus development team of the U.S. Weather Research program to NOAA
and NSF. Bull. Amer. Meteor. Soc., (in press).
Smith, R. B., 1979: The influence of the mountains on the atmosphere. Advances in Geophysics, B.
Saltzman, Ed., Vol. 21, Academic Press, 87-230.
Smith, R. B., and co-authors, 1996: Research needs and opportunities in mountain meteorology:
Report of the fourth prospectus development team of the U.S. Weather Research program to NOAA
and NSF. Bull. Amer. Meteor. Soc., (in press).
Stauffer, D. R., and N. L. Seaman, 1992: Use of four-dimensional data assimilation in a limited-area
mesoscale model. Part I: Experiments with synoptic-scale data. Mon. Wea. Rev., 118, 1250-1277.
Uccellini, L. W., 1996: Office of Meteorology 1996-2005 Strategic Operating Plan. Available from
the United States National Weather Service, Office of Meteorology, 75 pp.
Veldon, C. S., C. M. Hayden, S. J. Neiman, W. P. Menzel, S. Wanzong, and J. S. Goerss, 1997: Upper-tropospheric winds derived from geostationary satellite water vapor observations. Bull. Amer. Meteor. Soc., 78, 173-195. |