California Precipitation Simulation in the NMC Nested Spectral Model: 1993 January Event

Shyh-Chin Chen and John 0. Roads Climate research Division, Scripps Institution of oceanography

University of California, San Diego 0224, La Jolla, CA 92093-0224

Henry H.-M. Juang and Masao Kanamitsu National Meteorological Center, National Weather Service, Washington, D.C.

I. Introduction

Over California there is very noticeable low frequency variability in the precipitation and surface temperature extreme events. The winter-mean time series of precipitation along the west slope of the Sierra Nevadas in Northern California from 1895 through 199 1 is shown in Fig. 1. Several multi-winter wet and dry spells have occurred with anomalous amplitudes comparable to climatological mean. For example, the state of California has had a severe drought since 1987 (State of California 1992). In fact, during January and February of 1993, record breaking precipitation fell over Southern California and effectively ended a near-record, long-term drought. However, the drought reappeared again in 1994.

Although some precipitation extreme events occurred during intense El NiiiosISouthern Oscillation ENS0 periods, the overall statistical relation (not shown) is quite blurred. To complicate matters, both dry and wet winters occurred during the more recent warm events. For example, 1987 was dry and 1993 was wet. Also, many California precipitation extremes, for example, 1991, 1990, 1984, 1986 and 1956) did not occur during pronounced ENS0 events (Ropelewski and Halpert 1986; Cayan and Peterson 1989). However, California's climate is also influenced by mid-latitude central Pacific variations (Namias 1978; Klein and Bloom 1987; and Cayan and Peterson 1989). In a modeling study, Chen and Cayan (1994) showed that regional surface temperature and precipitation extremes over California could be attributed to seasonal time scale variations of large-scale circulation patterns which are induced solely by the model internal nonlinear'dynamics. Background variations or forcings from many aspects are probably important for exciting large-scale internal oscillation modes and determining precipitation extremes over the West Coast.

Interpreting large-scale variations on regional scales has been hampered by a limited spatial observations, which are often limited to populated land areas (e.g. Roads et al., 1991). Although regional meteorological characteristics can be related to large-scale circulations; and described by the derived divergence of moisture fluxes (Roads et a1 1993); the quality and resolution of this data is usually insufficient for studying low frequency regional atmospheric hydrological variations (Rasmusson 1967). Therefore to increase our understanding of regional climates, we are using a regional model nested within a global model.

I1 Model

The nested spectral model (NSM) was developed at the National Meteorological Center (NMC) by Juang and Kanamitsu (1994). The NSM is a high resolution regional spectral model (RSM) nested in an NMC's global spectral model (GSM). The GSM is a frozen and lower resolution version of NMC's operational medium range forecast (MRF) global model. The nesting is done by first integrating the GSM for a nesting period, currently set at 6 hours. The GSM thus provides initial and low spatial resolution model parameters as lateral boundary conditions of the RSM. Both models (GSM and RSM) use the same comprehensive set of physical parameterization modular packages as described in NMC Development Division Staff (1988) and Kalnay et a1 (1990). Both models use the same 18 terrain following sigma layers. The major

difference between the RSM and GSM lies in the horizontal basis functions; the global model expresses its solutions in spherical harmonics with a triangular truncation of 62 (T62) whereas the regional component uses 30 cosine waves with a horizontal resolution of 25 by 27 km for our California study. The regional horizontal domain is chosen so that it covers all of California and Nevada and parts of the surrounding states.

Basically, the RSM predicts regional deviations from time varying large-scale circulations as depicted by either a global forecast or analysis. Briefly, the non-linear advection is first computed by the global and regional spectral methods in an exact manner so that the model is free of aliasing and phase error. The global-scale tendency is then removed, so that in effect only the portion affecting the perturbation is retained. At the horizontal boundaries, the perturbation forcing is set to zero, which ensures that the boundary perturbations remain zero. The regional model thus produces only those smaller scales which are due to higher resolution orography and land-sea distributions.

The kinetic energy (K. E.) spectra shown in Figure 2a gives an example of how the RSM and GSM model complement each other. The kinetic energy in the global model was calculated as a function of zonal wave number, only for those latitude regions encompassing the California region. Although this eliminates non-relevant latitude belts, there are still some slight inconsistencies in that the global model includes longitude regions not considered for the regional model. Anyway, after the relatively flat planetary wave portion of the spectra, the global model shows a rapid drop-off consistent with the -3 power law of geostrophic motion. Due to small scale diffusion, an even more rapid drop-off occurs near the end of the global resolution. In the RSM model rapid drop-off can really only be seen beyond wave 300. The area mean (first mode) and the lowest wave-number of perturbations in the region model also display a much smaller amplitudes relatively to those from global model. This suggests that the regional model does not modify the large-scale fields; it only augments scales beyond which the global model can currently adequately describe.

The continuation of the spectra is also evident for other model parameters. Fig. 2b shows the power spectra for the vertical column integrated precipitable water (water vapor). The same behavior is evident here. The major difference is that the regional spectrum seems to be slightly more intense than that extrapolated from the global one. This is due to the fact that part of the model domain covers the mountainous Sierra where the integrated water vapor is much less than those over flat Pacific ocean. Perhaps the major surprise here, as shown in Fig. 2c, is for the precipitation spectra, which the RSM spectrum is much larger for all wavelengths. Although, again, the spectra calculation is not over the same areas, the higher amounts are a slight cause for worry.

I11 Preliminary Precipitation Simulation

As a particular example, we show in Figs. 3a and 3b the GSM and nested RSM 6-hour accumulated rainfall from 24-hr forecast initialized on January 15, 1993. Note that (Fig. 3a) the GSM displays a coarsely distributed precipitation pattern whereas the regional model (Fig. 3b) shows more realistic fine-scale features. The 24-hour forecast indicates heavy precipitation along the western-side of the Sierra-Nevadas and over the Los Angeles basin.

For the 30-day accumulated precipitation pattern, we believe the RSM simulation produces more realistic fine-scale details than the GSM. Shown in Figs. 3c and 3d are the 30-day accumulated precipitation from GSM and RSM, respectively. The simulation was done by reinitializing the model daily (every day in January, 1993) with NMC sigma-layer analysis. Compared to the observations shown in Fig. 4, the RSM captures many major features over heavy rainfall areas including Southern California. However, the observations are too coarse to even reveal such fine-scale structure. Detailed inter-annual variations of these fine-scale precipitation patterns are

vitally important to California. For example, characteristics of wintertime precipitation over the Sierra Nevadas determine the amount and timing of the subsequent spring runoff, which the California water supply relies heavily upon.

Further details and comparisons to individual stations will be shown at the meeting.

References:

Cayan, D.R., and D.H. Peterson, 1989: The influence of North Pacific atmospheric circulation on streamflow in the West. Geophysical Monograph, 55, available from American Geophysical Union, 2000 Florida Avenue, NW, Washington, D. C. 20009.

Chen, S.-C., and D.R. Cayan, 1994: Low Frequency aspects of the large-scale circulation and West Coast U.S. temperaturelprecipitation fluctuations in a simplified general circulation model, J. Climate, in press.

Juang, H.-M. H., and M. Kanamitsu, 1994: The NMC nested regional spectral model. Mon. Wea. Rev., 122, 3-26.

Kalnay, E., M. Kanarnitsu, and W. E. Baker, 1990: Global numerical weather prediction at the National Meteorological Center. Bull. Amer. Meteor. Soc., 71, 14 10- 1428.

Klein, and Bloom, 1987: Specification of monthly precipitation over the United States from he surrounding 700 mb height field. Mon. Wea. Rev., 115, 2 1 18-2 132.

Namias, J. 1-978, Multiple causes of the North American abnormal winter 1976-77, Mon. Wea. Rev., 106, 279-295.

NMC Development Division Staff, 1988: Documentation of the research version of the medium Range Forecasts model. NMC Documentation Series #1. (Available from the Development Division, NMC, Washington DC, 20233)

Rasmusson, E. M., 1967: Atmospheric water vapor transport and the water balance of North America, Characteristics of the water vapor flux field. Mon. Wea. Rev., 95,403-426.

Roads, J.O. T.N. Maisel and J. Alpert, 1991 : Further evaluation of the National Meteorological Center's Medium Range Forecast Model Precipitation Forecasts. Weather and Forecasting, 6(4), 483-497.

Roads, J.O., S.-C. Chen, J. Kao, D. Langley, and G. Glatzmaier, 1993: Global aspects of the Los Alamos general circulation model hydrologic cycle, J. Geophys. Res, 97, 10051- 10068.

State of California, 1992: Water conditions in California, Report 3, April, 1992, State of California, The Resource Agency and Department of Water Resources.

TOTAL PRECIPITATION (inches) JAN l a 9 3

CLlUAir AUALISlS CfhTfR. N O M

Fig. 4 Observed precipitation over continental U.S. (from Weekly Weather and Crop Bulletin, February 9, 1993)

O B S DJF SAC-SJQ T E M P & P R E C I P ANOMALIES

-50 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

YEAR

Fig. 1 Wintertime (DJF) precipitation anomalies from 1895 through 1991 over the Sacramento and San Joaquin Valley. The wintertime precipitation average is 37 cm.

RN**2 S p e c t r a

I

K . E . S p e c t r a R Q * * 2 S p e c t r a

Global Zonal Wave m + l Global Zonal Wave m+l

0 1E.04. I

Fig. 2a, 2b, 2c (a) Kinetic energy (KE) as a function of global zonal wave number. The global model KE spectrum is area averaged over the latitude bands of the RSM California region. The regional domain average perturbation KE from the RSM is shown by the dashed line. The KE is averaged from

i each 6 hour output of a 24-hour forecast for each day of January 1993. (b) Same as (a), except vertically integrated water vapor spectra are shown. (c) Same as (a), except spectra are shown only for the monthly

. mean precipitation. 1000

, ,. .; I

:., :,;{:.. ' <$!:.

( % ,>

9; Regional ~ 4 9 ~ 1

, .

Global Zonal Uave mil

0.1E-09=

O.lE-In_

0.1E-ll

, Regional Model . ,' \I . % . . :": '..

3

10 100

0.1E-02:

0.1E-03-

0.lE-04

7

0.lE-02

10 100 1000 10 100 1000

,,$SM PRECIP INIT. DAY= 0 1 15 9 3 &F2$pa7&4m, I,\ JAN 1 9 9 3 GSM ACCUM. PRECIP. , INCH

(hFSM PRECIP INIT. DAY= 0 1 15 9 3 &?2$,pa=/i4HR) /a\ JAN 1 9 9 3 RSM ACCUM. PRECIP. ,,= , ,,,,

Fig. 3. Last 6-hour precipitation forecast from a 24-hour forecast initialized on 00Z January 15, 1993 by (a) the GSM and (b) the RSM. Contours are shown every 0.2 inches. January accumulations are shown on the right for (c) the GSM and (d) the RSM. Here, (on the right), the contours are plotted every 2 inches starting from 4 inches. The orography in the global and regional models is shaded every 600 meters starting at 200 meters; the darkest shading indicates orography higher than 2000 meters.