• Barnston, A. G., and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev.,115, 1083–1126.

  • Blasing, T. J., and G. R. Lofgren, 1980: Seasonal climatic anomaly types for the North Pacific sector and western North America. Mon. Wea. Rev.,108, 700–719.

  • Buell, C. E., 1971: Integral equation representation for “factor analysis.” J. Atmos. Sci.,28, 1502–1505.

  • Cayan, D. R., 1991: “Cool” versus “warm” winter precipitation and its effect on streamflow in California. Proc. Seventh Annual Pacific Climate (PACLIM) Workshop, Asilomar, CA, California Dept. of Water Resources, 91–99.

  • ——, and J. O. Roads, 1984: Local relationships between United States west coast precipitation and monthly mean circulation parameters. Mon. Wea. Rev.,112, 1276–1282.

  • ——, and D. H. Peterson, 1989: The influence of North Pacific atmospheric circulation on streamflow in the west. Aspects of Climate Variability in the Pacific and the Western Americas, Geophys. Monogr., No. 55, Amer. Geophys. Union, 375–397.

  • Chang, F.-C., 1986: On the seasonality of climate fluctuations over the contiguous United States. Ph.D. dissertation, University of Washington, 161 pp. [Available from University Microfilms, 300 N. Zeeb Rd., Ann Arbor, MI 48106.].

  • Chen, T.-C., J.-M. Chen, and C. K. Wikle, 1996: Interdecadal variation in U.S. Pacific coast precipitation over the past four decades. Bull. Amer. Meteor. Soc.,77, 1197–1205.

  • Cheng, X., G. Nitsche, and J. M. Wallace, 1995: Robustness of low-frequency circulation patterns derived from EOF and rotated EOF analyses. J. Climate,8, 1709–1713.

  • Davis, J. C., 1986: Statistics and Data Analysis in Geology. John Wiley and Sons, 550 pp.

  • Deser, C., and J. M. Wallace, 1990: Large-scale atmospheric circulation features of warm and cold episodes in the tropical Pacific. J. Climate,3, 1254–1281.

  • Englehart, P. J., and A. V. Douglas, 1985: A statistical analysis of precipitation frequency in the conterminous United States, including comparisons with precipitation totals. J. Climate Appl. Meteor.,24, 350–362.

  • Fovell, R. G., and M.-Y. Fovell, 1993: Climate zones of the conterminous United States defined using cluster analysis. J. Climate,6, 2103–2135.

  • Harman, H. H., 1976: Modern Factor Analysis. 3d ed. University of Chicago Press, 487 pp.

  • Holton, J. R., 1992: An Introduction to Dynamic Meteorology. 3d ed. Academic Press, 511 pp.

  • Horel, J. D., 1981: A rotated principal component analysis of the interannual variability of the Northern Hemisphere 500-mb height field. Mon. Wea. Rev.,109, 2080–2092.

  • ——, and J. M. Wallace, 1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Karl, T. R., A. J. Koscielny, and H. F. Diaz, 1982: Potential errors in the application of principal component (eigenvector) analysis to geophysical data. J. Appl. Meteor.,21, 1183–1187.

  • ——, L. K. Metcalf, M. L. Nicodemus, and R. G. Quayle, 1983: Statewide average climatic history. Historical Climatology Series 6-1, NOAA National Climatic Data Center, Asheville, NC, vi–vii. [Available from National Climatic Data Center, Federal Building, 151 Patton Ave., Asheville, NC 28801.].

  • Klein, W. H., and H. J. Bloom, 1987: Specification of monthly precipitation over the United States from the surrounding 700-mb height field. Mon. Wea. Rev.,115, 2118–2132.

  • Kushnir, Y., and J. M. Wallace, 1989: Low-frequency variability in the Northern Hemisphere winter: Geographical distribution, structure and timescale dependence. J. Atmos. Sci.,46, 3122–3142.

  • Lambert, S. J., 1990: Discontinuities in the long-term Northern Hemisphere 500-millibar heights dataset. J. Climate,3, 1479–1484.

  • Lau, N.-C., G. H. White, and R. L. Jenne, 1981: Circulation statistics for the extratropical Northern Hemisphere based on NMC analyses. National Center for Atmospheric Research Tech. Note NCAR/TN-171+STR, Boulder, CO, 138 pp. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307.].

  • Leith, C. E., 1973: The standard error of time-average estimates of climatic means. J. Appl. Meteor.,12, 1066–1069.

  • Magaña, V. O., 1991: Tropical–extratropical atmospheric interactions. Ph.D. dissertation, University of California, Los Angeles, 194 pp. [Available from University Microfilms, 300 N. Zeeb Rd., Ann Arbor, MI 48106.].

  • Meko, D. M., C. W. Stockton, and W. R. Boggess, 1980: A tree-ring reconstruction of drought in southern California. Water Res. Bull.,16, 594–600.

  • Michaelson, J., L. Haston, and F. W. Davis, 1987: 400 years of central California precipitation variability reconstructed from tree-rings. Water Res. Bull.,23, 809–818.

  • Mitchell, T. P., and J. M. Wallace, 1996: ENSO seasonality: 1950–78 versus 1979–92. J. Climate,9, 3149–3161.

  • Namias, J., 1978a: Recent drought in California and western Europe. Rev. Geophys. Space Phys.,16, 435–458.

  • ——, 1978b: Multiple causes of the North American abnormal winter 1976–1977. Mon. Wea. Rev.,106, 279–295.

  • ——, 1979: Premonitory signs of the 1978 break in the west coast drought. Mon. Wea. Rev.,107, 1675–1681.

  • ——, 1981: The heavy California winter rains of 1979–80 as a manifestation of macroscale air/sea coupling. Proc. Fifth Annual Climate Diagnostics Workshop, Seattle, WA, U.S. Dept. Commerce, NOAA, 35–50.

  • ——, 1982: Meteorological and oceanographic conditions for the enhancement or suppression of winter rains over California. Storms, Floods, and Debris Flows in Southern California and Arizona 1978 and 1980, National Academy Press, 25–41.

  • North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev.,110, 699–706.

  • O’Lenic, E. A., and R. E. Livezey, 1988: Practical considerations in the use of rotated principal component analysis (RPCA) in diagnostic studies of upper-air height fields. Mon. Wea. Rev.,116, 1682–1689.

  • Parker, D. E., and C. K. Folland, 1991: Worldwide surface air temperature trends since the mid-19th century. Greenhouse-Gas Induced Climatic Change: A Critical Appraisal of Simulations and Observations, M. E. Schlesinger, Ed., Elsevier, 173–193.

  • ——, P. D. Jones, C. K. Folland, and A. Bevan, 1994: Interdecadal changes of surface temperature since the late nineteenth century. J. Geophys. Res.,99(D7), 14373–14399.

  • Pazan, S. E., and G. Meyers, 1982: Interannual fluctuations of the tropical Pacific wind field and the Southern Oscillation. Mon. Wea. Rev.,110, 587–600.

  • Preisendorfer, R. W., 1988: Principal Component Analysis in Meteorology and Oceanography. Elsevier, 425 pp.

  • ——, and T. P. Barnett, 1977: Significance tests for empirical orthogonal functions. Preprints, Fifth Conf. on Probability and Statistics in Atmospheric Sciences, Las Vegas, NV, Amer. Meteor. Soc., 169–172.

  • Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev.,110, 354–384.

  • Richman, M. B., 1981: Obliquely rotated principal components: An improved meteorological map typing technique? J. Appl. Meteor.,20, 1145–1159.

  • ——, 1986: Rotation of principal components. J. Climatol.,6, 293–335.

  • Rogers, J. C., 1981: The North Pacific oscillation. J. Climatol.,1, 39–57.

  • Ropelewski, C. F., and M. S. Halpert, 1986: North American precipitation and temperature patterns associated with the ENSO. Mon. Wea. Rev.,114, 2352–2362.

  • ——, and ——, 1989: Precipitation patterns associated with the high index phase of the Southern Oscillation. J. Climate,2, 268–284.

  • Schonher, T., and S. E. Nicholson, 1989: The relationship between California rainfall and ENSO events. J. Climate,2, 1258–1269.

  • Sellers, W. D., 1968: Climatology of monthly precipitation patterns in the western United States, 1931–1966. Mon. Wea. Rev.,96, 585–595.

  • Stidd, C. K., 1954: The use of correlation fields in relating precipitation to circulation. J. Meteor.,11, 202–213.

  • Swetnam, T. W., and J. L. Betancourt, 1990: Fire–Southern Oscillation relations in the southwestern United States. Science,249, 1017–1020.

  • Trenberth, K. E., 1990: Recent observed interdecadal climate changes in the Northern Hemisphere. Bull. Amer. Meteor. Soc.,71, 988–993.

  • Wallace, J. M., C. Smith, and C. S. Bretherton, 1992: Singular value decomposition of wintertime sea surface temperature and 500-mb height anomalies. J. Climate,5, 561–576.

  • Walsh, J. E., M. B. Richman, and D. W. Allen, 1982: Spatial coherence of monthly precipitation in the United States. Mon. Wea. Rev.,110, 272–286.

  • Weaver, R. C., 1962: Meteorology of hydrologically critical storms in California. Hydrometeorological Rep. 3, Dept. of Commerce, Weather Bureau, Washington, DC, 207 pp.

  • Willmott, C. J., 1977: A component analytic approach to precipitation regionalization in California. Arch. Meteor. Geophys. Bioklimatol. Ser. B,24, 269–281.

  • Woodruff, S. D., R. J. Slutz, R. L. Jenne, and P. M. Steurer, 1987: A comprehensive ocean–atmosphere data set. Bull. Amer. Meteor. Soc.,68, 1239–1250.

  • ——, S. J. Lubker, K. Wolter, S. J. Worley, and J. D. Elms, 1993: Comprehensive Ocean Atmosphere Data Set (COADS) Release 1a: 1980–92. Earth Sys. Monitor,4, 1–8.

  • Yarnal, B., and H. F. Diaz, 1986: Relationships between extremes of the Southern Oscillation and the winter climate of the Anglo–American Pacific coast. J. Climatol.,6, 197–219.

  • Yin, Z.-Y., 1996: Discontinuities in the NMC winter 500-mb and 700-mb geopotential height data. J. Climate,9, 786–802.

  • Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal-to-century scale variability: 1900–93. J. Climate, 1004–1020.

  • View in gallery

    Climatic divisions of California.

  • View in gallery

    Eigenvalues (cm2 month−2) (circles), range of eigenvalues due to sampling errors (cm2 month−2) (North et al. 1982) (shaded bars), and 95th percentile eigenvalues from Monte Carlo simulations (cm2 month−2) (Preisendorfer and Barnett 1977) (dashed line) for (a) seasonal- and (b) monthly mean PCs.

  • View in gallery

    Correlation maps of December–February seasonal-mean precipitation anomalies derived for (a) RPC 1 (the Oregon pattern) and (b) RPC 5 (the southern California pattern). The contour interval is 0.1, with negative (zero) contours dashed (thickened). Digital values for the three California coastal climatic divisions are plotted to supplement the contour analysis. Light, medium, and dark shading indicates elevations >1, 2, and 3 km, respectively.

  • View in gallery

    Wintertime seasonal-mean (a) Oregon RPC (line) and north coast climatic division precipitation anomalies (bars), and (b) southern California RPC (line) and south coast climatic division precipitation anomalies (bars). The climatic division time series are for the California climatic divisions that are closest to the core of each loading vector, and the RPCs are scaled to be the regressed values for those grid points.

  • View in gallery

    Correlation coefficients between the southern California seasonal-mean precipitation RPC and Pacific SST anomalies for the period 1931–88. The contour interval is 0.2, with negative, zero, and positive contours indicated by dashed, thick solid, and thin solid lines, respectively. Light (dark) shading indicates correlation magnitudes between 0.2 and 0.4 (0.4 and 0.6).

  • View in gallery

    Reconstructed (a) 500-hPa geopotential height and (b) SLP fields for an extreme dry month of the north coast RPC. See text for details on the construction of the maps. Isopleth intervals are 6 dam in (a) and 2 hPa in (b). The 564-dam contour in (a) and the 1020-hPa isobar in (b) are thickened, and the SLP isobars are labeled with the last two digits of the pressure in hPa (e.g., “1020” as “20”).

  • View in gallery

    As in Fig. 6 but for an extreme wet month of the north coast RPC.

  • View in gallery

    Correlation coefficients between the north coast RPC and monthly values of (a) 500-hPa geopotential height and (b) SLP. Contour interval is 0.1, with negative, zero, and positive contours indicated by dashed, thick solid, and thin solid lines, respectively.

  • View in gallery

    (a) 500 hPa and (b) SLP in December 1955, the wettest “north coast” month. Isopleth intervals of 6 dam in (a) and 2 hPa in (b).

  • View in gallery

    As in Fig. 9 but for February 1986, the third wettest north coast month.

  • View in gallery

    As in Fig. 9 but for December 1951, the 20th wettest south coast month.

  • View in gallery

    As in Fig. 9 but for January 1972, the 169th wettest south coast month (out of 173).

  • View in gallery

    Fig. A1. As in Fig. 3 but for RPC analysis of December–February monthly mean precipitation anomalies for the contiguous United States, with (a) RPC 1 (the north coast pattern) and (b) RPC 8 (the south coast pattern) capturing the predominant portion of California precipitation variability. Only the part of the domain with significant correlations is plotted.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 24 24 4
PDF Downloads 26 26 2

The Variability of Wintertime Precipitation in the Region of California

View More View Less
  • 1 Universities Space Research Association, Inc., NASA/Goddard Space Flight Center, Greenbelt, Maryland
  • | 2 Department of Atmospheric Sciences, University of California, Los Angeles, Los Angeles, California
© Get Permissions
Full access

Abstract

Rotated principal component (RPC) analysis, subject to the varimax criterion and including area weighting, is applied to a 58-yr record (1931–88) of monthly- and seasonal-mean Climatic Division precipitation anomalies for the contiguous United States to document wintertime precipitation variability in the region of California. Rotated principal components (time series) derived from this analysis are related to anomalies of seasonal-mean global sea surface temperature, and monthly mean Northern Hemisphere 500-hPa geopotential height and sea level pressure (SLP).

Wintertime seasonal-mean precipitation in California is captured by two RPCs. The first RPC documents coherent precipitation anomalies centered in northern California, Oregon, southern Idaho, and eastern Washington, and explains the largest portion of area-averaged variance of any of the patterns in the decomposition. A second RPC captures coherent precipitation variability in the south coast and southeast desert regions of California, southern Nevada, southern Utah, and northern Arizona. Fluctuations in the first RPC correlate poorly with Pacific Ocean SST anomalies. However, wet winters in the region of the second RPC correlate modestly with simultaneous cool western subtropical Pacific Ocean SST anomalies and weakly with warm SST anomalies over a broad region of the central and eastern tropical Pacific. The spatial scale of the tropical SST correlations and the prominent multidecadal timescale signal of the RPC are consistent with ENSO fluctuations on this timescale influencing southern California precipitation.

Consistent with the results of earlier studies, significant correlations are found between California wintertime monthly mean precipitation variability and regional 500-hPa geopotential height and SLP anomalies. Linear regression analysis is used to construct estimates of the total 500-hPa geopotential height and SLP fields (climatology plus anomaly) that are representative of the extreme wet and dry California winter months; these are then compared with the observed conditions in the individual extreme months. Several different flow patterns appear capable of producing anomalously large monthly precipitation totals in California.

*thinsp;Current affiliation: Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, Seattle, Washington.

Corresponding author address: Dr. Todd P. Mitchell, Joint Institute for the Study of the Atmosphere and Ocean, Box 354235, University of Washington, Seattle, WA 98195.

Abstract

Rotated principal component (RPC) analysis, subject to the varimax criterion and including area weighting, is applied to a 58-yr record (1931–88) of monthly- and seasonal-mean Climatic Division precipitation anomalies for the contiguous United States to document wintertime precipitation variability in the region of California. Rotated principal components (time series) derived from this analysis are related to anomalies of seasonal-mean global sea surface temperature, and monthly mean Northern Hemisphere 500-hPa geopotential height and sea level pressure (SLP).

Wintertime seasonal-mean precipitation in California is captured by two RPCs. The first RPC documents coherent precipitation anomalies centered in northern California, Oregon, southern Idaho, and eastern Washington, and explains the largest portion of area-averaged variance of any of the patterns in the decomposition. A second RPC captures coherent precipitation variability in the south coast and southeast desert regions of California, southern Nevada, southern Utah, and northern Arizona. Fluctuations in the first RPC correlate poorly with Pacific Ocean SST anomalies. However, wet winters in the region of the second RPC correlate modestly with simultaneous cool western subtropical Pacific Ocean SST anomalies and weakly with warm SST anomalies over a broad region of the central and eastern tropical Pacific. The spatial scale of the tropical SST correlations and the prominent multidecadal timescale signal of the RPC are consistent with ENSO fluctuations on this timescale influencing southern California precipitation.

Consistent with the results of earlier studies, significant correlations are found between California wintertime monthly mean precipitation variability and regional 500-hPa geopotential height and SLP anomalies. Linear regression analysis is used to construct estimates of the total 500-hPa geopotential height and SLP fields (climatology plus anomaly) that are representative of the extreme wet and dry California winter months; these are then compared with the observed conditions in the individual extreme months. Several different flow patterns appear capable of producing anomalously large monthly precipitation totals in California.

*thinsp;Current affiliation: Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, Seattle, Washington.

Corresponding author address: Dr. Todd P. Mitchell, Joint Institute for the Study of the Atmosphere and Ocean, Box 354235, University of Washington, Seattle, WA 98195.

1. Introduction

California experiences precipitation in a single season centered in the winter months of December, January, and February, with the largest precipitation amounts (annual totals > 1 m) found in the northern portions of the state. Over the last 10 years California has experienced widely varying annual precipitation; for 6 consecutive years (August 1986–July 1992) most of the state received approximately 80% of normal precipitation, while in another year (August 1994–July 1995) precipitation totals were approximately 180% of normal statewide.1 Precipitation fluctuations of this magnitude and over this large a region constitute one of the dominant forms of climate variability in the United States. The dramatic swings in annual precipitation in California affect its residents through flood damage in heavy rain years and water rationing in drought years. As California produces approximately 12% of the agricultural output of the United States, the impact of precipitation variations in California is felt throughout the country. On a longer timescale, years of significantly below-normal precipitation intensify the ongoing political competition between California agricultural, industrial, and domestic users of the scarce water resources.

The recent fluctuations in California seasonal precipitation amounts have resulted in increased interest in documenting the spatial and temporal scales of its variability, as well as on understanding the underlying mechanisms. The present study contributes to this effort by characterizing the monthly and seasonal-mean precipitation variability in California during December, January, and February, the calendar months during which the state receives approximately 50% of its annual precipitation. Atmospheric and oceanic conditions associated with the anomalously wet and dry winter months and winter seasons will also be documented.

In order to identify patterns of coherent precipitation fluctuations that explain the majority of California winter season precipitation variability, Rotated principal component (RPC) analysis is applied to Climatic Division precipitation data for the period 1931–88. Data for the contiguous United States are analyzed in recognition of the fact that the spatial scales of climate variability are often larger than those of individual states. This study revisits the RPC analysis of wintertime monthly-mean precipitation anomalies for the same spatial domain considered by Walsh et al. (1982, hereafter WRA) but with a much higher spatial resolution dataset that permits the documentation of the relative contributions of smaller-scale regions to the dominant modes of precipitation variability (e.g., enables distinction between the California coastal regions and inland valleys). In general, the higher spatial resolution of the present dataset with respect to that employed in WRA can be expected to yield a more definitive characterization of the precipitation variability. Both the loading vectors (spatial patterns) and the associated rotated principal components (time series) that are outputs of RPC analysis will be employed in this study.

Previous investigations of the seasonal-mean timescale mechanisms that influence California wintertime precipitation focused on the role of ENSO-related ocean–atmosphere interactions centered in the tropical Pacific and on ocean–atmosphere interactions in the North Pacific that may or may not owe their origin to ENSO. It is envisioned in these studies that the interaction between the ocean and atmosphere in these two regions of the Pacific Ocean can influence the planetary-scale atmospheric circulation and affect the climate over the adjoining land areas through modifications of the storm tracks.

Studies of ENSO variability in the tropical Pacific have suggested that ENSO-related changes in the tropical heating can influence the position and intensity of the major extratropical upper-level jets and hence the storm track positions (e.g., Horel and Wallace 1981). In observational studies based on this hypothesis by Ropelewski and Halpert (1986, 1989), Yarnal and Diaz (1986), and Schonher and Nicholson (1989), California precipitation amounts were composited using various indicators of the polarity and strength of the ENSO phenomenon for their compositing criterion.

Namias (1978a,b, 1979, 1981, 1982) introduced the idea that ocean–atmosphere interactions in the North Pacific were responsible for several extreme wet and dry California winters in the period 1976–80. Studies by Wallace et al. (1992) and others cited therein documented the strength of the ocean–atmosphere coupling in the North Pacific and are consistent with the hypothesis that coupled ocean–atmosphere variability is an important contributor to climate variability in the North Pacific and neighboring land regions. Studies of seasonal-mean atmospheric conditions in the region bordering the North Pacific by Blasing and Lofgren (1980), Rogers (1981), and Cayan and Peterson (1989) indicated that there are distinct seasonal-mean planetary-scale atmospheric circulations associated with significant variations in California wintertime precipitation.

In the present study, the RPC time series for California wintertime seasonal-mean precipitation are correlated with contemporaneous sea surface temperature anomalies for the global oceans. This analysis extends the ENSO composite analyses cited above by the inclusion of both ENSO warm and cold episodes. Also, it is not obvious a priori that the patterns of SST anomalies associated with ENSO fluctuations or other possible forms of variability in the North Pacific are the patterns most closely related to California precipitation variability. The SST correlation calculation will therefore document the patterns of SST that are best related to significant precipitation variations.

Other studies have emphasized the importance of fluctuations of monthly-mean atmospheric conditions in understanding the variability in wintertime California precipitation. In particular, statistically significant correlations have been documented between monthly-mean precipitation and anomalies of sea level pressure (SLP), and 700- and 500-hPa geopotential height in the immediate vicinity of California (Stidd 1954; Sellers 1968; Namias 1979, 1981; WRA; Cayan and Roads 1984; Chang 1986; Klein and Bloom 1987). In general, these investigations emphasized the importance of translations of the storm tracks in the immediate vicinity of California and/or changes in the flow relative to orography in producing variations in California precipitation.

In the present study, the large-scale 500-hPa geopotential height and SLP fields characteristic of the more extreme wet and dry months are constructed through a linear regression analysis. Qualitative comparison with conditions for individual extreme wet and dry months will suggest whether or not a single form of atmospheric variability is responsible for the observed precipitation variability (as the linear regression analysis employed is most effective if a single phenomenon contributes to fluctuations in the precipitation time series).

The organization of the paper is as follows. The data and analysis methods employed are described in section 2. Documentation of the seasonal-mean wintertime precipitation variability that influences California, and the relationship between this variability and the global distribution of SST anomalies, is presented in section 3. The association between wintertime California precipitation variability and the monthly-mean atmospheric circulation anomalies is examined in section 4. The results are discussed in section 5.

2. Data and analysis methods

a. Data

The National Climatic Data Center (NCDC) Climatic Division data employed in this study comprise monthly precipitation totals for the 344 climatic divisions of the contiguous United States for the period January 1931–July 1988 (Karl et al. 1983). Additional precipitation values through February 1995 for selected California climatic divisions, presented in section 3, were provided by the Research Customer Service Group of NCDC. The climatic divisions were defined with the objective of partitioning each state into homogeneous climatic regions; the divisional mean for a given month is defined as the arithmetic average of all available station data for that month (Karl et al. 1983). The climatic divisions for California are shown in Fig. 1, while those for the entire domain can be seen in Fig. 1 of Fovell and Fovell (1993). The climatic divisions vary significantly in size and are distributed such that the greatest density of divisions is found in the eastern United States. The subjective choice of the climatic divisions for California by Karl et al. (1983) is supported by an RPC analysis of monthly-mean precipitation totals for 90 California stations by Willmott (1977), with the regions of coherent (predominantly seasonal) precipitation variability identified in the RPC analysis exhibiting fidelity with the climatic division partitionment of California.

Monthly means of 500-hPa geopotential height and SLP for 1946–88 were taken from the National Oceanic and Atmospheric Administration/National Centers for Environmental Prediction, formerly the National Meteorological Center (NOAA/NCEP), Northern Hemisphere octagonal grid dataset. The data, for the latitudes north of 20°N, were interpolated onto a 2° latitude–longitude resolution grid for graphical display (i.e., to enable use of existing graphics software). The correlation analyses presented in section 4 were repeated for subsets of the record to confirm that the spurious temporal inhomogeneities in the 500-hPa geopotential height dataset (Lambert 1990; Yin 1996) did not unduly influence the results.

Monthly mean SST and monthly total number of observations for the period 1931–88 were obtained from the Comprehensive Ocean–Atmosphere Data Set (COADS) (Woodruff et al. 1987; 1993), a compilation of ship-of-opportunity and buoy measurements for the global oceans, with a spatial resolution of 2° latitude–longitude. Changes in the sampling density of SST observations for the period of this study are documented in Woodruff et al. (1987).

b. Calculation of anomalies

Climatological means were calculated for each calendar month for precipitation, 500-hPa geopotential height, SLP, and SST for the periods 1931–88, 1946–88, 1946–88, and 1946–88, respectively. The years used for the SST climatology were chosen to take advantage of the superior sampling of the post–World War II record. For the SST climatology calculation, the contribution of each SST value was weighted by the number of observations in that year and month. The resultant SST climatology was subjected to a three-point running mean along latitude circles to reduce the noisiness of the field. The climatologies were then subtracted from each of the fields to produce monthly anomalies.

Seasonal-mean anomalies of precipitation and SST were calculated as the average of the anomalies for December, January, and February, beginning with January of 1931 and ending with February of 1988. The noisiness of the resulting SST anomalies was reduced by five applications of a two-dimensional binomial smoother. The SST correlation map presented in section 3 was compared with an analysis based on unsmoothed anomalies to ensure that whatever significant correlations might have existed in near-coastal SST were not removed by the spatial smoothing.

c. Analysis methods

1) Rotated principal component analysis

The dominant patterns of wintertime monthly- and seasonal-mean precipitation variability are documented in this study with RPC analysis subject to the varimax criterion. Horel (1981) and Richman (1981, 1986) provided extensive discussions of this technique and its application to meteorological analysis. The present study employs the nomenclature and RPC analysis methodology described in Horel (1981). In particular, the output of the RPC analysis consists of loading vectors (spatial maps) and the associated rotated principal components (RPCs) (time series). The term RPC can also refer to an individual loading vector and its associated time series, with the particular meaning indicated by the context.

Information on the areal extent of each climatic division was incorporated into both the principal component (PC) and the RPC calculations. The inclusion of area information in the PC analysis follows Buell (1971) and North et al. (1982). The mathematical conditions necessary to fulfill the varimax criterion in RPC analysis were described in Harman (1976) and are outlined here only to identify the modifications to the rotation procedure that are necessary to incorporate area information. Davis (1986) provided a FORTRAN source code for RPC analysis subject to the varimax criterion, and this code may be easily modified for this purpose. First, for each grid point, the loadings (values) for the input PC loading vectors (scaled by the square root of the eigenvalues) are squared and summed. Called the communality, the sum of the squares is used to normalize the loadings for each grid point and to rescale the loadings at the end of the calculation. Next, the vectors are rotated two at a time by an angle determined to maximize the fourth power (varimax) statistic for that vector pair. This step acts to simplify the loadings under the constraint of an orthogonal transformation matrix. After all pairs of vectors have been rotated, the fourth power statistic is calculated for the entire set of vectors. The pairwise rotation of the vectors is repeated until the value of the fourth power statistic for the entire set of vectors converges. In the present study, the area of each climatic division is used to linearly weight each climatic division’s contribution to the calculation of both the pairwise vector rotation angles and of the varimax statistic for the entire set of vectors. The code for the orthogonal rotation including area information was developed and graciously provided by Y. Kushnir of the Lamont-Doherty Earth Observatory.

Eigenvalues and the percentages of area-averaged variance explained by each of the first 15 PCs of the temporal covariance matrices of monthly and seasonal-mean precipitation are given in Table 1. The number of PC loading vectors to rotate in the RPC analysis was determined with the methods of O’Lenic and Livezey (1988, hereafter OL) and Kushnir and Wallace (1989) (hereafter KW). O’Lenic and Livezey (1988) employ a scree test, with the number of PCs rotated adjusted to include possibly degenerate PCs and the sensitivity of the RPC solution checked with respect to the inclusion or omission of the last PC. Kushnir and Wallace (1989) rotated increasingly larger numbers of PCs until the RPC solution exhibited stability with respect to the inclusion of an additional PC. The OL and KW methods suggested the retention of 11 and 8 PCs for the seasonal- and monthly mean analyses, respectively, and these choices were employed in our analyses.

For the interested reader, two tests of PC signal and sampling characteristics are presented in graphical form in Fig. 2 and as digital values in Table 1. The rule-N test (Preisendorfer and Barnett 1977; Preisendorfer 1988), given by the dashed curves in Fig. 2, show that the leading 7 seasonal- and 10 monthly mean PCs capture as much or more precipitation variability signal than PCs generated in 95 of 100 Monte Carlo simulations. The North et al. (1982) criterion, depicted by the error bars in Fig. 2, indicate a numerically distinct eigenvalue break between the first five seasonal- and eight monthly mean PCs and the higher-order PCs, with the higher-order PCs being poorly sampled. An RPC analysis of the first five seasonal-mean PC loading vectors obtained RPC solutions that were simple distortions of the input PC loading vectors. As this outcome indicated underrotation (Richman 1981), it was decided to retain 11 PCs for the seasonal-mean calculation. Our choice to include seasonal-mean PCs associated with poorly separated eigenvalues can be justified by the results of Richman (1986), Barnston and Livezey (1987), and Cheng et al. (1995), which indicate the greater stability with respect to sampling variability for RPC analysis than for PC analysis.

The RPC loading vectors for the wintertime monthly mean data (not shown) captured precipitation fluctuations in many of the regions that have been identified in RPC analysis of normalized anomalous precipitation amounts by WRA and anomalous precipitation frequency by Englehart and Douglas (1985, hereafter ED). Chang (1986) also applied RPC analysis to monthly mean climatic division data but ignored variations in climatic division area with the result that the RPCs emphasized variability in the eastern part of the domain, where the highest density of climatic divisions is found [see also Karl et al. (1982)]. The RPCs in the present study were ordered on the basis of the percentage of area-averaged precipitation variance of the contiguous United States that is explained by each RPC. The percentage of variance explained by each of the first five RPCs of wintertime monthly and seasonal-mean precipitation variability are given in Table 2, with values for the relevant WRA RPCs also provided. We are not aware of any published RPC analyses of the seasonal-mean precipitation data. Only the subset of RPCs that explain significant amounts of California precipitation will be presented and discussed in this study.

RPC analysis is sometimes employed to identify nonoverlapping regions of coherent variability of the field of interest (e.g., Willmott 1977; ED). In such a “regionalization,” each grid point is identified with the single RPC that captures the largest percentage of variance at that location. In contrast, in this study [as in Horel (1981), Kushnir and Wallace (1989), and others] the spatial structure of the dominant RPC loading vectors are shown for the entire spatial domain. This form of presentation permits the identification of unique spatial signatures for the physical processes that contribute to that RPC’s variability, and it is consistent with the domain-average nature of RPC analysis.

2) Linear regression and construction of total fields

Maps of correlation coefficients are presented that document the normalized monthly mean atmospheric circulation anomalies and seasonal-mean SST anomalies that are associated with fluctuations in California precipitation. The linear correlation (and subsequent regression) analyses emphasize features in the anomalous meteorological and SST fields that are of similar magnitude but of opposite sign during extreme wet and dry conditions in California. Alternatively, the anomalies in extreme wet and dry episodes could have been composited, as in the studies of Klein and Bloom (1987) and Cayan (1991). Although the results of such a composite analysis can be quite similar to those from correlation–regression analysis, the latter approach is advantageous in that it is based on all temporal realizations in a dataset; it is thus less susceptible to sampling fluctuations and has a smaller degree of arbitrariness (Stidd 1954).

The large-scale atmospheric circulation associated with extreme wet and dry months is documented by adding (and subtracting) typical circulation anomalies, as determined by regression analysis, to (and from) the climatological wintertime circulation. The regression maps are multiplied by averages of the RPC values for the 10 wettest and driest months of the record. The choice of the number of extreme months is arbitrary, but the resultant “total fields” give a representation of the “average” circulation associated with the extreme wet and dry California periods. This technique has been employed previously by Pazan and Meyers (1982), Deser and Wallace (1990), and Mitchell and Wallace (1996).

3. Wintertime precipitation variability on the seasonal-mean timescale

California December–February (DJF) seasonal-mean precipitation variability is primarily captured by two RPCs. The loading vectors (spatial patterns) for these RPCs are presented as maps of correlation coefficients (Fig. 3), with the value at each grid point equal to the correlation coefficient between the precipitation time series at that grid point and the RPC.

The first RPC documents coherent precipitation variations in northern California, Oregon, eastern Washington, and western Idaho (Fig. 3a). The coherent fluctuations explain over half of the precipitation variability for the individual climatic divisions in this region. The largest amounts of variance explained are for the Oregon south-central division (83%) and California north coast division (81%). The region of coherent precipitation anomalies captured by this first RPC (henceforth referred to as the “Oregon” pattern) is bounded on the east by the Rocky Mountains and to the west by the Cascade Mountains in Washington State.2 This RPC is also characterized by a weak tendency for precipitation anomalies of opposite polarity to occur in southern Arizona.

The time series of this RPC, and the actual wintertime mean precipitation anomalies for the California north coast climatic division, are shown in Fig. 4a. The north coast division is characterized by, on average, large precipitation amounts (18 cm month−1) during December, January, and February, and large fluctuations in DJF-average precipitation totals (std dev = 7 cm month−1) in both individual and consecutive winters (Fig. 4a). The RPC exhibits notable fidelity in capturing this precipitation variability. Chen et al. (1996) examined the 1950–94 precipitation record for a region that includes the Oregon pattern and suggested that a long-term drying trend contributes to the observed fluctuations.

Fluctuations in the Oregon pattern explain 18% of the area-averaged precipitation variability for the contiguous United States, the largest percentage captured by any loading vector in the decomposition (Table 2).3 The determination that the Oregon mode is the leading pattern of wintertime precipitation variability for the contiguous United States, in terms of percentage of variance explained, is an important result of this study.

A second RPC captures coherent precipitation variations over southern California, northern Arizona, southern Nevada, southern Utah, and northwestern New Mexico (Fig. 3b). Coherent fluctuations in this RPC, henceforth called the “southern California” pattern, explain over half of the seasonal-mean precipitation variance for the individual climatic divisions in this region, with the largest amount of variance explained for the California south coast division (86%). As with the Oregon pattern, the southern California precipitation variability is seen to be bounded to the east by the Rocky Mountains. There is a weak tendency for precipitation anomalies of the opposite sign to occur in western Washington. Figure 3b also shows a secondary maximum centered in South Dakota and Nebraska; oblique rotation of the PCs (Harman 1976) may or may not have yielded a simpler pattern.

The time series of the southern California RPC along with the associated precipitation time series of the south coast climatic division are shown in Fig. 4b. The south coast climatic division is characterized by smaller climatological DJF precipitation amounts (mean = 8 cm month−1) and year-to-year variance (std dev = 4 cm month−1) than the north coast division. Nonetheless, fluctuations in the southern California RPC explain 10% of the area-averaged precipitation variability for the contiguous United States, the fifth largest of the loading vectors in the decomposition (Table 2). The prominence of this RPC results primarily from the coherence of the precipitation variability over a broad area. Englehart and Douglas (1985) reached a similar conclusion for an RPC of wintertime monthly precipitation frequencies that captured coherent variability in the same area. The temporal record for this RPC (Fig. 4b) exhibits a more distinct multidecadal-scale variability than does the Oregon RPC (Fig. 4a), with persistent positive anomalies in 1932–44, negative anomalies in the late 1940s through the early 1970s, and two multiple-year wet periods between the late 1970s and the present. The prominent multidecadal wet and dry epochs during these years have previously been inferred from tree-ring records for southern California (Meko et al. 1980) and central California (Michaelson et al. 1987), and Arizona and New Mexico tree-burn areas (Swetnam and Betancourt 1990). The wet (dry) periods also correspond to multidecadal periods of ENSO warm (cold) conditions documented in global SST and tropical SLP records by Trenberth (1990), Parker and Folland (1991), Parker et al. (1994), and Zhang et al. (1997).

The spatial distribution of correlation coefficients between the southern California RPC and simultaneous values of Pacific Ocean SST is presented in Fig. 5. Statistics were calculated for the period 1931–85 for the global oceans but are only shown for the subregion of the Pacific Ocean in which the strongest relationships were found. Wet southern California winters are associated with a modest tendency (correlation magnitudes typically 0.35) for positive SST anomalies over a broad region of the eastern equatorial and eastern subtropical Pacific, and negative SST anomalies extending from the western subtropical Pacific northeastward into the midlatitudes. The largest magnitude correlation, with a value of −0.60 (19°N, 151°E), is based on 57 yr of data.4 The pattern of SST anomalies in the equatorial and subtropical portions of the Pacific is broadly similar to that associated with the ENSO variability observed during the Northern Hemisphere winter season (see Rasmusson and Carpenter 1982; Deser and Wallace 1990). The broad latitudinal scale of the SST correlations in the eastern tropical Pacific is consistent with the multidecadal ENSO SST signature documented in Zhang et al. (1997). Schonher and Nicholson (1989), on the basis of a composite study, suggested the importance of the distribution of warm equatorial SST anomalies during ENSO warm events in explaining some of the extreme wet southern California winters. Based on a larger number of years of data, we found a similar result but with a fairly modest linear relationship between California seasonal mean precipitation and equatorial Pacific Ocean SST (typical correlations of 0.35).

The negative correlations shown in Fig. 5 in the western subtropical Pacific, although modest in magnitude, occupy a large portion of this region. The existence of this relationship has not previously been noted. Magaña (1991) related intraseasonal (30–60 day) variations in tropical and subtropical convective activity in this area to the amplitude of the subtropical circulations (e.g., the mid-Pacific trough) and to the intensity of the midlatitude flow.

Smaller correlations (correlation magnitudes < 0.25) were found between precipitation in the Oregon–northern California region and SSTs (not shown). The general weakness of the precipitation–SST relationship for the Oregon pattern is notable considering the dominance of this RPC in the decomposition of precipitation variance for the winter season.

The above analysis suggests that fluctuations in basinwide Pacific Ocean surface conditions may be associated with some small portion of the seasonal wintertime precipitation variability experienced in California. A cursory examination of the daily precipitation record for a number of stations in California, though, indicates that much of the precipitation is received during intervals that in sum represent only a small fraction of the winter season. As such, it seems appropriate to try to characterize the precipitation variability on a shorter timescale than the seasonal mean. We thus proceed to consider the relation between atmospheric circulation and fluctuations in California precipitation amounts on the monthly mean timescale.

4. The association between monthly mean atmospheric circulation anomalies and California wintertime precipitation variability

As described in the introduction, previous studies have documented the average anomalies of the atmospheric circulation associated with fluctuations of December, January, and February monthly mean precipitation anomalies. Here we extend these analyses through examination of the total fields associated with extreme wet and dry months, with the objective of identifying the large-scale flow regimes that are responsible for significantly anomalous precipitation on a monthly time scale.

Charts of the total 500-hPa geopotential height and SLP fields that are representative of the 10 driest north coast California winter months for the period 1946–88 are shown in Fig. 6. Charts of the total fields that are representative of the 10 wettest north coast California winter months are presented in Fig. 7. The methodology used to construct these maps is described in section 2c(2). The California north coast precipitation variability is characterized by the leading RPC of wintertime monthly mean precipitation variability for the contiguous United States, fluctuations of which explain 93% of north coast division precipitation variability. [The loading vectors for this RPC (termed the “north coast” RPC) and a second RPC that captures significant southern California precipitation variability (termed the “south coast” RPC) are presented in the appendix (Fig. A1), along with information on the months of extreme values of these time series.] Maps of correlation coefficients between the north coast RPC and both 500-hPa geopotential height and SLP are presented in Fig. 8.5

Extreme dry months for the north coast of California (Fig. 6) are characterized by an enhancement of the climatological wintertime flow pattern (Lau et al. 1981), with a strong ridge aloft along the west coast of the United States and an enhanced subtropical high with centers located off the central California coast and over the intermountain region. The associated low-level flow is weakly offshore. In contrast, extreme wet months for the north coast of California (Fig. 7) are associated with stronger and more zonal upper-level flow over the eastern Pacific Ocean; the surface subtropical high is displaced southward, resulting in onshore flow into northern California. These circulation changes are consistent with the finding of Cayan and Roads (1984) that months of above-normal wintertime precipitation in the region of the central California coast have enhanced 700-hPa zonal winds. The 500-hPa geopotential height and SLP correlations in Fig. 8 document several more subtle features of the precipitation-related circulation. The SLP correlation map (Fig. 8b) indicates troughing to the lee of the Sierra Nevada range,6 a feature not evident in the correlation analysis of lower-resolution SLP data (5° latitude–longitude) by WRA. The westward tilt with height evident in Fig. 8 indicates the baroclinic nature of the precipitation-bearing circulation. Finally, the correlations for both 500-hPa geopotential height and SLP document an offshore extension of the features southwestward into the subtropics, a region previously identified by Magaña (1991) as an area of significant tropical–extratropical interaction.

Even the typical extreme wet winter month for the dominant north coast RPC, as given by Fig. 7, is associated with a weak upper-level ridge; this contrasts with the upper-level trough pattern usually associated with precipitation-producing synoptic-scale wave disturbances (e.g., Holton 1992). The monthly mean 500-hPa geopotential height field thus fails to show the flow pattern of the synoptic-scale phenomenon primarily responsible for the production of significant precipitation. This is consistent with the comparatively short (3–6 day) timescale of the precipitation-bearing synoptic-scale waves (e.g., Weaver 1962) and implies at least a comparably long timescale to the intervening periods of little or no precipitation.

We next consider how representative the total field maps (Figs. 6 and 7) are of the flow patterns for the individual extreme wet and dry months for the north coast of California. The 500-hPa geopotential height and SLP analyses for December 1955 and February 1986, the first and third wettest months for the north coast RPC (Table A1), are shown in Figs. 9 and 10, respectively. In the vicinity of California, the 500-hPa analysis for December 1955 (Fig. 9a) bears a strong similarity to the reconstructed 500-hPa map (Fig. 7a), with only weak ridging over the western United States and strong upper-level flow. A significant planetary-scale ridge appears over the North Pacific near the date line. The SLP pattern (Fig. 9b) is also similar to that of the wet month total field (Fig. 7b).

The third wettest month, February 1986 (Fig. 10), in contrast, is characterized by a very different large-scale flow pattern. Here there is enhanced upper-level anticyclonic flow over the west coast of the United States, while a large area of deep low pressure, centered to the south of the Aleutian Islands, dominates the North Pacific basin (Fig. 10). Other extreme wet months (not shown) exhibit still further variations in flow regimes. Little variation in this regard was seen in the extreme dry months; the features evident in Fig. 6 recurred in each of the individual dry months examined.

It appears, then, that the wettest months for the north coast RPC are associated with a wide range of monthly mean large-scale flow patterns. A similar examination of the extreme wet months for the south coast of California, as characterized by a second RPC (Fig. A1b), yielded a similar variety of flow patterns. The implications of these results are discussed in the following section.

5. Discussion

The variability of seasonal- and monthly mean wintertime precipitation for the contiguous United States has been documented with RPC analysis, and the RPCs that capture the predominant portion of California precipitation have been presented. This study employed a higher spatial resolution dataset than previous studies and as such yielded more detailed characterizations of the regions of coherent precipitation variability. An additional aspect of the present work was the inclusion of area averaging in the RPC methodology; this extension to the standard calculation permitted the large variations of climatic division area to be properly taken into account.

On the seasonal-mean timescale, two RPCs captured much of the precipitation variability in the vicinity of California, and the associated time series of these RPCs were related to simultaneous large-scale SSTs. One RPC captured precipitation variations in northern California and portions of neighboring states, and exhibited precipitation variations on a wide range of timescales. Fluctuations in this RPC, which explain the largest percentage of any RPC of area-averaged seasonal-mean wintertime precipitation variance for the contiguous United States (18%), are poorly correlated with SST anomalies in the Pacific and other oceans. A second RPC captured coherent precipitation variations in southern California and adjoining states, and exhibited a distinct multidecadal character to its variability. Studies of tree-ring and tree burn area records for this region corroborate the dominance and polarity of the multidecadal signal. The extended wet and dry periods indicated by this RPC are consistent with respective decadal timescale warm and cool ENSO fluctuations documented by Zhang et al. (1997) and others. Further study of the few precipitation station records and secondary data sources (e.g., tree rings) for the years before 1930 will be helpful to verify the existence of a predominantly multidecadal signal in the earlier record.

Two RPCs also captured significant wintertime monthly mean California precipitation variability; they are similar to those derived from the seasonal-mean precipitation anomalies. As in previous studies, statistically significant regional relationships were found between the time series of the precipitation patterns and those of anomalous SLP and 500-hPa geopotential height fields. As an extension to the earlier studies, characteristic maps of the total circulation fields (climatology plus anomaly) for the extreme wet and dry months were then constructed using a linear regression approach. These maps were compared with the maps for individual extreme wet and dry months. It was found that several distinct large-scale monthly mean patterns were associated with the extreme wet months, with some of these patterns bearing little resemblance to the linear regression-based maps. Dry winter months, however, were all characterized by an enhancement of the amplitude of both the climatological upper-level ridge in the vicinity of the United States west coast and the sea level subtropical high.

In some cases, there was little difference between the 500-hPa geopotential height and SLP fields for a particular extreme wet month and a particular extreme dry month. As an example, consider the atmospheric circulation patterns for December 1951 (Fig. 11) and January 1972 (Fig. 12), the 20th and 169th wettest months (out of 173) for the RPC capturing California south coast precipitation variability (Fig. A1b), respectively. The 500-hPa geopotential height and SLP patterns for these months appear remarkably similar despite yielding 1.4 std dev more (December 1951) and 1.4 std dev less (January 1972) precipitation than normal in the south coast region. Apparently the precipitation-producing synoptic-scale wave disturbances in December 1951 were sufficiently few in number and/or sufficiently fast to have had little impact on the mean 500-hPa geopotential height and SLP statistics for this month.

The results presented here indicate the difficulty of developing a simple paradigm to explain monthly and seasonal timescale precipitation variations in the region of California. They also imply that it will be difficult for the present generation of atmospheric and coupled ocean–atmosphere climate models, which focus on predicting the conditions on these timescales, to produce useful forecasts of this important component of California climate variability.

Finally, our results suggest that an understanding of the circulation patterns associated with California precipitation variability will require analysis on the submonthly timescale. A goal of future work is thus to document and explain the shorter timescale transition between wet and dry conditions in months with anomalously large precipitation but characterized by monthly mean atmospheric upper-level and sea level circulation patterns that are more representative of dry conditions.

Acknowledgments

The authors wish to thank Robert Livezey, Hisashi Nakamura, and Walter Robinson for discussions of aspects of this work. We also appreciate the carefully considered anonymous reviews of the manuscript, which contributed significantly to its improvement. Yochanan Kushnir has been particularly generous with his ideas. TPM wishes to thank William K.-M. Lau for his support of this project. This research was supported in part by the California Space Institute Grant CS-88-92.

REFERENCES

  • Barnston, A. G., and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev.,115, 1083–1126.

  • Blasing, T. J., and G. R. Lofgren, 1980: Seasonal climatic anomaly types for the North Pacific sector and western North America. Mon. Wea. Rev.,108, 700–719.

  • Buell, C. E., 1971: Integral equation representation for “factor analysis.” J. Atmos. Sci.,28, 1502–1505.

  • Cayan, D. R., 1991: “Cool” versus “warm” winter precipitation and its effect on streamflow in California. Proc. Seventh Annual Pacific Climate (PACLIM) Workshop, Asilomar, CA, California Dept. of Water Resources, 91–99.

  • ——, and J. O. Roads, 1984: Local relationships between United States west coast precipitation and monthly mean circulation parameters. Mon. Wea. Rev.,112, 1276–1282.

  • ——, and D. H. Peterson, 1989: The influence of North Pacific atmospheric circulation on streamflow in the west. Aspects of Climate Variability in the Pacific and the Western Americas, Geophys. Monogr., No. 55, Amer. Geophys. Union, 375–397.

  • Chang, F.-C., 1986: On the seasonality of climate fluctuations over the contiguous United States. Ph.D. dissertation, University of Washington, 161 pp. [Available from University Microfilms, 300 N. Zeeb Rd., Ann Arbor, MI 48106.].

  • Chen, T.-C., J.-M. Chen, and C. K. Wikle, 1996: Interdecadal variation in U.S. Pacific coast precipitation over the past four decades. Bull. Amer. Meteor. Soc.,77, 1197–1205.

  • Cheng, X., G. Nitsche, and J. M. Wallace, 1995: Robustness of low-frequency circulation patterns derived from EOF and rotated EOF analyses. J. Climate,8, 1709–1713.

  • Davis, J. C., 1986: Statistics and Data Analysis in Geology. John Wiley and Sons, 550 pp.

  • Deser, C., and J. M. Wallace, 1990: Large-scale atmospheric circulation features of warm and cold episodes in the tropical Pacific. J. Climate,3, 1254–1281.

  • Englehart, P. J., and A. V. Douglas, 1985: A statistical analysis of precipitation frequency in the conterminous United States, including comparisons with precipitation totals. J. Climate Appl. Meteor.,24, 350–362.

  • Fovell, R. G., and M.-Y. Fovell, 1993: Climate zones of the conterminous United States defined using cluster analysis. J. Climate,6, 2103–2135.

  • Harman, H. H., 1976: Modern Factor Analysis. 3d ed. University of Chicago Press, 487 pp.

  • Holton, J. R., 1992: An Introduction to Dynamic Meteorology. 3d ed. Academic Press, 511 pp.

  • Horel, J. D., 1981: A rotated principal component analysis of the interannual variability of the Northern Hemisphere 500-mb height field. Mon. Wea. Rev.,109, 2080–2092.

  • ——, and J. M. Wallace, 1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Karl, T. R., A. J. Koscielny, and H. F. Diaz, 1982: Potential errors in the application of principal component (eigenvector) analysis to geophysical data. J. Appl. Meteor.,21, 1183–1187.

  • ——, L. K. Metcalf, M. L. Nicodemus, and R. G. Quayle, 1983: Statewide average climatic history. Historical Climatology Series 6-1, NOAA National Climatic Data Center, Asheville, NC, vi–vii. [Available from National Climatic Data Center, Federal Building, 151 Patton Ave., Asheville, NC 28801.].

  • Klein, W. H., and H. J. Bloom, 1987: Specification of monthly precipitation over the United States from the surrounding 700-mb height field. Mon. Wea. Rev.,115, 2118–2132.

  • Kushnir, Y., and J. M. Wallace, 1989: Low-frequency variability in the Northern Hemisphere winter: Geographical distribution, structure and timescale dependence. J. Atmos. Sci.,46, 3122–3142.

  • Lambert, S. J., 1990: Discontinuities in the long-term Northern Hemisphere 500-millibar heights dataset. J. Climate,3, 1479–1484.

  • Lau, N.-C., G. H. White, and R. L. Jenne, 1981: Circulation statistics for the extratropical Northern Hemisphere based on NMC analyses. National Center for Atmospheric Research Tech. Note NCAR/TN-171+STR, Boulder, CO, 138 pp. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307.].

  • Leith, C. E., 1973: The standard error of time-average estimates of climatic means. J. Appl. Meteor.,12, 1066–1069.

  • Magaña, V. O., 1991: Tropical–extratropical atmospheric interactions. Ph.D. dissertation, University of California, Los Angeles, 194 pp. [Available from University Microfilms, 300 N. Zeeb Rd., Ann Arbor, MI 48106.].

  • Meko, D. M., C. W. Stockton, and W. R. Boggess, 1980: A tree-ring reconstruction of drought in southern California. Water Res. Bull.,16, 594–600.

  • Michaelson, J., L. Haston, and F. W. Davis, 1987: 400 years of central California precipitation variability reconstructed from tree-rings. Water Res. Bull.,23, 809–818.

  • Mitchell, T. P., and J. M. Wallace, 1996: ENSO seasonality: 1950–78 versus 1979–92. J. Climate,9, 3149–3161.

  • Namias, J., 1978a: Recent drought in California and western Europe. Rev. Geophys. Space Phys.,16, 435–458.

  • ——, 1978b: Multiple causes of the North American abnormal winter 1976–1977. Mon. Wea. Rev.,106, 279–295.

  • ——, 1979: Premonitory signs of the 1978 break in the west coast drought. Mon. Wea. Rev.,107, 1675–1681.

  • ——, 1981: The heavy California winter rains of 1979–80 as a manifestation of macroscale air/sea coupling. Proc. Fifth Annual Climate Diagnostics Workshop, Seattle, WA, U.S. Dept. Commerce, NOAA, 35–50.

  • ——, 1982: Meteorological and oceanographic conditions for the enhancement or suppression of winter rains over California. Storms, Floods, and Debris Flows in Southern California and Arizona 1978 and 1980, National Academy Press, 25–41.

  • North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev.,110, 699–706.

  • O’Lenic, E. A., and R. E. Livezey, 1988: Practical considerations in the use of rotated principal component analysis (RPCA) in diagnostic studies of upper-air height fields. Mon. Wea. Rev.,116, 1682–1689.

  • Parker, D. E., and C. K. Folland, 1991: Worldwide surface air temperature trends since the mid-19th century. Greenhouse-Gas Induced Climatic Change: A Critical Appraisal of Simulations and Observations, M. E. Schlesinger, Ed., Elsevier, 173–193.

  • ——, P. D. Jones, C. K. Folland, and A. Bevan, 1994: Interdecadal changes of surface temperature since the late nineteenth century. J. Geophys. Res.,99(D7), 14373–14399.

  • Pazan, S. E., and G. Meyers, 1982: Interannual fluctuations of the tropical Pacific wind field and the Southern Oscillation. Mon. Wea. Rev.,110, 587–600.

  • Preisendorfer, R. W., 1988: Principal Component Analysis in Meteorology and Oceanography. Elsevier, 425 pp.

  • ——, and T. P. Barnett, 1977: Significance tests for empirical orthogonal functions. Preprints, Fifth Conf. on Probability and Statistics in Atmospheric Sciences, Las Vegas, NV, Amer. Meteor. Soc., 169–172.

  • Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev.,110, 354–384.

  • Richman, M. B., 1981: Obliquely rotated principal components: An improved meteorological map typing technique? J. Appl. Meteor.,20, 1145–1159.

  • ——, 1986: Rotation of principal components. J. Climatol.,6, 293–335.

  • Rogers, J. C., 1981: The North Pacific oscillation. J. Climatol.,1, 39–57.

  • Ropelewski, C. F., and M. S. Halpert, 1986: North American precipitation and temperature patterns associated with the ENSO. Mon. Wea. Rev.,114, 2352–2362.

  • ——, and ——, 1989: Precipitation patterns associated with the high index phase of the Southern Oscillation. J. Climate,2, 268–284.

  • Schonher, T., and S. E. Nicholson, 1989: The relationship between California rainfall and ENSO events. J. Climate,2, 1258–1269.

  • Sellers, W. D., 1968: Climatology of monthly precipitation patterns in the western United States, 1931–1966. Mon. Wea. Rev.,96, 585–595.

  • Stidd, C. K., 1954: The use of correlation fields in relating precipitation to circulation. J. Meteor.,11, 202–213.

  • Swetnam, T. W., and J. L. Betancourt, 1990: Fire–Southern Oscillation relations in the southwestern United States. Science,249, 1017–1020.

  • Trenberth, K. E., 1990: Recent observed interdecadal climate changes in the Northern Hemisphere. Bull. Amer. Meteor. Soc.,71, 988–993.

  • Wallace, J. M., C. Smith, and C. S. Bretherton, 1992: Singular value decomposition of wintertime sea surface temperature and 500-mb height anomalies. J. Climate,5, 561–576.

  • Walsh, J. E., M. B. Richman, and D. W. Allen, 1982: Spatial coherence of monthly precipitation in the United States. Mon. Wea. Rev.,110, 272–286.

  • Weaver, R. C., 1962: Meteorology of hydrologically critical storms in California. Hydrometeorological Rep. 3, Dept. of Commerce, Weather Bureau, Washington, DC, 207 pp.

  • Willmott, C. J., 1977: A component analytic approach to precipitation regionalization in California. Arch. Meteor. Geophys. Bioklimatol. Ser. B,24, 269–281.

  • Woodruff, S. D., R. J. Slutz, R. L. Jenne, and P. M. Steurer, 1987: A comprehensive ocean–atmosphere data set. Bull. Amer. Meteor. Soc.,68, 1239–1250.

  • ——, S. J. Lubker, K. Wolter, S. J. Worley, and J. D. Elms, 1993: Comprehensive Ocean Atmosphere Data Set (COADS) Release 1a: 1980–92. Earth Sys. Monitor,4, 1–8.

  • Yarnal, B., and H. F. Diaz, 1986: Relationships between extremes of the Southern Oscillation and the winter climate of the Anglo–American Pacific coast. J. Climatol.,6, 197–219.

  • Yin, Z.-Y., 1996: Discontinuities in the NMC winter 500-mb and 700-mb geopotential height data. J. Climate,9, 786–802.

  • Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal-to-century scale variability: 1900–93. J. Climate, 1004–1020.

APPENDIX

Monthly Mean Precipitation RPCs

RPC analysis was applied to the monthly mean December, January, and February precipitation anomalies to provide the optimal characterization of the precipitation variability on the monthly timescale and is not intended to suggest that a different physical mechanism is responsible for the precipitation variability at this frequency. The loading vectors for the two RPCs that explain the largest amounts of California precipitation variability are shown in Fig. A1. The analysis was performed for the contiguous United States, but only the region with significant correlations is plotted. The RPCs associated with Figs. A1a and A1b are termed the north coast and south coast RPCs, respectively, in recognition of the California regions of largest variance explained by each of the RPCs. The north coast and south coast RPCs explain 18% and 7% of area-averaged precipitation variance for the contiguous United States, respectively, and the time series values for the 10 extreme wet and dry months of these RPCs are given in Table A1.

Fig. 1.
Fig. 1.

Climatic divisions of California.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 2.
Fig. 2.

Eigenvalues (cm2 month−2) (circles), range of eigenvalues due to sampling errors (cm2 month−2) (North et al. 1982) (shaded bars), and 95th percentile eigenvalues from Monte Carlo simulations (cm2 month−2) (Preisendorfer and Barnett 1977) (dashed line) for (a) seasonal- and (b) monthly mean PCs.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 3.
Fig. 3.

Correlation maps of December–February seasonal-mean precipitation anomalies derived for (a) RPC 1 (the Oregon pattern) and (b) RPC 5 (the southern California pattern). The contour interval is 0.1, with negative (zero) contours dashed (thickened). Digital values for the three California coastal climatic divisions are plotted to supplement the contour analysis. Light, medium, and dark shading indicates elevations >1, 2, and 3 km, respectively.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 4.
Fig. 4.

Wintertime seasonal-mean (a) Oregon RPC (line) and north coast climatic division precipitation anomalies (bars), and (b) southern California RPC (line) and south coast climatic division precipitation anomalies (bars). The climatic division time series are for the California climatic divisions that are closest to the core of each loading vector, and the RPCs are scaled to be the regressed values for those grid points.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 5.
Fig. 5.

Correlation coefficients between the southern California seasonal-mean precipitation RPC and Pacific SST anomalies for the period 1931–88. The contour interval is 0.2, with negative, zero, and positive contours indicated by dashed, thick solid, and thin solid lines, respectively. Light (dark) shading indicates correlation magnitudes between 0.2 and 0.4 (0.4 and 0.6).

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 6.
Fig. 6.

Reconstructed (a) 500-hPa geopotential height and (b) SLP fields for an extreme dry month of the north coast RPC. See text for details on the construction of the maps. Isopleth intervals are 6 dam in (a) and 2 hPa in (b). The 564-dam contour in (a) and the 1020-hPa isobar in (b) are thickened, and the SLP isobars are labeled with the last two digits of the pressure in hPa (e.g., “1020” as “20”).

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 7.
Fig. 7.

As in Fig. 6 but for an extreme wet month of the north coast RPC.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 8.
Fig. 8.

Correlation coefficients between the north coast RPC and monthly values of (a) 500-hPa geopotential height and (b) SLP. Contour interval is 0.1, with negative, zero, and positive contours indicated by dashed, thick solid, and thin solid lines, respectively.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 9.
Fig. 9.

(a) 500 hPa and (b) SLP in December 1955, the wettest “north coast” month. Isopleth intervals of 6 dam in (a) and 2 hPa in (b).

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 10.
Fig. 10.

As in Fig. 9 but for February 1986, the third wettest north coast month.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 11.
Fig. 11.

As in Fig. 9 but for December 1951, the 20th wettest south coast month.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Fig. 12.
Fig. 12.

As in Fig. 9 but for January 1972, the 169th wettest south coast month (out of 173).

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

i1520-0442-10-9-2261-fa1

Fig. A1. As in Fig. 3 but for RPC analysis of December–February monthly mean precipitation anomalies for the contiguous United States, with (a) RPC 1 (the north coast pattern) and (b) RPC 8 (the south coast pattern) capturing the predominant portion of California precipitation variability. Only the part of the domain with significant correlations is plotted.

Citation: Journal of Climate 10, 9; 10.1175/1520-0442(1997)010<2261:TVOWPI>2.0.CO;2

Table 1.

Percentage of the area-averaged wintertime precipitation variance explained and eigenvalues (cm2 month−2) of the first 15 PCs of seasonal- and monthly mean precipitation anomalies. Also given are the range of eigenvalues due to sampling error (cm2 month−2) (North et al. 1982), and the 95th percentile eigenvalue from the Rule-N test (cm2 month−2) (Preisendorfer and Barnett 1977).

Table 1.
Table 2.

The percentage of precipitation variance explained by the first five loading vectors of monthly and seasonal-mean precipitation for the present calculation and for the corresponding loading vectors based on normalized monthly mean data in WRA. The loading vector numbers for the latter calculation refer to Fig. 2a of WRA.

Table 2.

Table A1. Ranked values of the normalized monthly mean precipitation RPCs that explain the largest amounts of California precipitation variability.

i1520-0442-10-9-2261-ta1

1

National Climatic Data Center Climatic Division precipitation records.

2

The topographical data, which is 5-min latitude–longitude resolution, were produced by the United States Defense Mapping Agency and obtained from the United States Geological Survey.

3

With ranking based on area-averaged variance explained, the number two RPC in the decomposition (not shown), which captures precipitation variability in the region of Appalachia, explains only two-thirds as much variance as the Oregon pattern (12%).

4

Confidence levels for the correlation coefficients are determined as follows. The 1-yr lag correlation for the southern California RPC is very low (0.06), indicating that the RPC values for each year are essentially independent (Leith 1973). The a priori 95% and 99% confidence limits for correlation coefficients with 58 degrees of freedom are 0.27 and 0.36, respectively. The true a posteriori confidence limits, which are relevant to the present case, are more stringent.

5

Confidence levels for the correlation coefficients were determined as follows. The 1-month lag correlation for the precipitation time series is 0.14 so that 85 of the 128 months in the analysis are essentially independent (Leith 1973). The a priori 95% and 99% confidence limits for correlation coefficients with this number of degrees of freedom are 0.22 and 0.28, respectively. The true a posteriori confidence limits, which are relevant to the present case, are more stringent.

6

The Sierra Nevada Range is oriented north-south along the eastern boundary of the Sacramento and San Joaquin Valley climatic divisions shown in Fig. 1.

Save