COUPLED MODES OF AIR-SEA INTERACTION AND THE MADDEN AND JULIAN
OSCILLA TION
Charles Jones and Catherine Gautier
Institute for Computational Earth System Science
University of California
Santa Barbara, CA 93106-3060
e-mail: cjones@icess.ucsb.edu
1. INTRODUCTION
The Madden and Julian Oscillation (MJO) (Madden and Julian, 1971, 1972)
is one of the primary modes of low-frequency variations in the tropical
atmosphere. In addition to manifesting itself in the lower and upper troposphere,
the MJO has been observationally detected near the surface and in the upper
layers of the tropical oceans. Variations of 30-60 days are clearly seen,
for example, in surface winds, surface heat fluxes, sea surface temperature,
and ocean currents (Madden and Julian, 1994).
Although the observational description of the MJO has progressed steadily
throughout the years, none of the present theories is able to fully explain
the basic characteristics of the MJO (Hayashi and Golder, 1993). Nonetheless,
two main mechanisms are generally accepted to play a role in the maintenance
of the MJO: wave-CISK (Conditional Instability of the Second Kind) (Lau
and Peng, 1987) and evaporation-wind feedback (Emanuel, 1987; Neelin et
al., 1987).
In two recent studies, Jones and Weare (1994a, 1994b) investigated the relationships
between variations in surface latent heat fluxes and tropical convective
activity associated with the MJO. The spatial and temporal patterns found
in those studies are such that anomalies of surface latent heat fluxes are
positive (negative) to the west (east) of the region of anomalous convection.
These patterns contrasts with the basic requirements of the evaporation-wind
feedback mechanism, which assumes that surface latent heat flux anomalies
are strongest to the east of the region of convection (Emanuel, 1987; Neelin
et al., 1987). The objective of this paper is to show further observational
evidence of the coupling between surface latent heat fluxes and tropical
precipitation in the frequency domain of the MJO. In this paper, new estimates
of surface latent heat fluxes and a different statistical technique are
used to determine the coupling between surface latent heat fluxes and precipitation.
The results shown here demonstrate that the results of Jones and Weare (1994a,
1994b) are robust, since the spatial and temporal patterns found are very
similar.
2. DATA
The tropical precipitation field (hereafter P) described in this study uses
data from the Global Precipitation Climatology Project (GPCP) described
by Janowiak and Arkin (1991). This data set provides pentad (5-day averages)
accumulations of estimated rainfall based on infrared data from geostationary
and polar orbiting satellites. The period of record analyzed consists of
pentads from January 1988 through June 1994 (73 pentads per year; total
of 474 pentads) on a 2.5° x 2.5° latitude/longitude global grid
extending from 40° S to 40° N.
In this study, the estimation of surface latent heat fluxes (hereafter E)
uses a combination of two data sets: European Centre for Medium-Range Weather
Forecast (ECMWF) surface analyses and Special Sensor Microwave/Imager (SSM/I)
data. The ECMWF surface analyses provide the following variables: surface
wind speeds (Vs) at 10 m, surface pressure (Ps), surface temperature (Ts),
air temperature (Ta) and air mixing ration (Qa) at 2 m, saturation mixing
ratio at the surface temperature (Qs). The SSM/I data set provides surface
wind speeds (Vs) at 10 m (Wentz, 1994). Pentads of the above variables are
computed for the same period and spatial domain used in the P field.
Mooring buoys data from the International Tropical Ocean Global Atmosphere-Tropical
Atmosphere Ocean (TOGA/TAO) Program are used to validate the estimation
of E. The TAO array in the tropical Pacific Ocean measures sea surface temperature
(SST) at -1 m, air temperature (Ta) and relative humidity (RH) at 3 m, and
surface wind speeds (Vs) at 4 m (Zhang and McPhaden, 1993). Based on these
variables, Qs and Qa are computed at 3 m. Since the measurement heights
from the TAO array are different than the heights used in the ECMWF and
SSM/I data sets, Vs, Ta, and Qa for the TAO data are also estimated at heights
of 10 m, 2 m and 2 m, respectively, using similarity theory relationships
(Liu et at., 1979). Pentads from January 1988 through June 1994 are computed
when at least three observations are available in a 5-day period. Since
the total number of pentads for these variables varies quite significantly
among the buoys, only 27 buoys with more than 100 pentads in the period
January 1988 - June 1994 are selected for validation purposes (average number
of pentads: 212).
3. ESTIMATION OF SURFACE LATENT HEAT FLUXES
In Jones and Weare (1994a, b), E is estimated using the bulk formula method
with constant exchange coefficient, and all variables used in the estimation
of E are derived from ECMWF surface analyses. Although the method used to
estimate E in the present study is still preliminary and is part of an ongoing
research, it shows improvements in estimating intraseasonal variations over
the tropical region.
Surface latent heat fluxes are estimated using the bulk parameterization
and the similarity theory model developed by Liu et al. (1979). Based on
a comparison between TAO-ECMWF, and TAO-SSM/I (not shown), the following
variables and correction factors are used: Vs - 0.08 m s-1 (SSM/I); Ta -
0.16° C (ECMWF); Qa + 1.84 g kg-1 (ECMWF); Ts (ECMWF); Qs (ECMWF);
Ps (ECMWF). The correction factors correspond to the magnitudes of the mean
biases found between TAO/SSM/I and TAO/ECMWF, averaged over the 27 TAO buoys.
In addition, Vs, Qa, and Ta are assigned to the following heights: Zvs =
10 m, Zqa = 3 m, and Zta = 3 m, respectively.
In order to estimate the uncertainties involved in the estimation of E as
described above, E is also estimated for the 27 TAO buoys, and mean biases,
root-mean square differences, and correlations between both estimates are
computed. The estimation of E using ECMWF and SSM/I pentads as described
above has a mean bias, root-mean square difference, and correlation coefficient,
averaged over the 27 buoys, on the order of -3.5 W m-2, 38.3 W m-2 and 0.51,
respectively.
4. RESULTS
In order to investigate the coupling between P and E, time filtering is
applied to the time series of pentads (474 pentads from January 1988 - June
1994). A band-pass Lanczos filter (Duchon, 1979) with 49 weights and cut-off
periods of 25 and 87 days is applied to the time series of P and E. The
resulting time series of anomalies start in May 1988 and end in February
1994 with a total of 426 pentads.
Single Value Decomposition (SVD) (Bretherton et al., 1992) is used to study
the relationships between intraseasonal variations in P and E. In summary,
the computation of SVD is done in the following way. First, the time series
of anomalies are spatially averaged to decrease the horizontal resolution
to 5.0° x 5.0° in latitude/longitude, so that the P and E fields
have 1152 and 779 points in space, respectively. Next, the temporal cross-covariance
matrix is constructed by computing covariances between time series of P
and E. Thus, the non-square cross-covariance matrix has the following dimensions:
1152 x 779. The method of SVD operates on the cross-covariance matrix, and
is an optimal technique to find spatial patterns that explain the largest
possible fraction of the mean squared covariance between the two fields.
The first singular vector of the P field is shown in Figure 1a, whereas
Figure 1b shows the first singular vector of the E field. The corresponding
expansion coefficients are shown in Figure 1c. This first coupled mode explains
18.2% of the cumulative squared covariance fraction (CSCF). A positive (negative)
fluctuation of the P expansion coefficient (Figure 1c) reveals positive
(negative) anomalies of P in the Indian Ocean, and negative (positive) anomalies
over the maritime continent in the western Pacific Ocean (Figure 1a). In
contrast, a positive (negative) fluctuation of the E expansion coefficient
(Figure 1c) indicates positive (negative) anomalies of E over the Indian
Ocean and central Pacific Ocean, and negative (positive) anomalies stretching
from the maritime continent towards the eastern coast of Asia.
5. DISCUSSION
In Jones and Weare (1994a, b), different estimates of tropical convection
and surface latent heat fluxes were used. In addition, a statistical technique
known as single field principal component analysis (SFPCA) (Bretherton et
al., 1992) was used to investigate the coupling between intraseasonal variations
in convection and surface latent heat fluxes. The spatial and temporal patterns
of the SVD analysis shown in the present study indicate that in the vicinities
of the region of enhanced convective activity, anomalies of E are positive
to the west and negative to the east of the region of convection. Indeed,
the spatial patterns of the second singular vectors of P and E, together
with lag homogeneous and heterogeneous correlation maps (not shown), further
confirm the pattern of P and E anomalies shown in the first singular vectors
(Figure 1). This results are consistent with the ones
discussed by Jones and Weare (1994a, b).
The pattern of P and E anomalies found in these observational studies contrasts
with the pattern assumed in the evaporation-wind feedback mechanism. That
theory assumes that, since the easterly wind anomalies to the east of the
convection are superimposed to mean easterly winds, evaporation anomalies
are enhanced to the east of the convection. Furthermore, the evaporation-wind
feedback theory claims that the positive anomalies of E to the east of the
region of anomalous convection plays a significant role in inducing unstable
eastward propagating waves (Emanuel, 1987; Neelin et al., 1987; Crum and
Dunkerton, 1994).
The observation of negative anomalies of E to the east of the region of
convection can be interpreted following the arguments discussed in Jones
and Weare (1994a, b). In the vicinities of the region of convection, low
level moisture convergence is observed, reaching a maximum to the east of
the region of convection. In addition, surface wind speeds to the east of
the region of convection reaches a minimum. The reduced surface wind speeds
to the east of the convection induce a reduction in surface latent heat
fluxes in that region. In contrast, since the region of convective anomalies
tend to be located in the region of westerly anomalies, surface wind speeds
reach a maximum to the west of the region of convection. The increase in
surface wind speeds to the west of the region of convection induces an increase
in surface latent heat fluxes in that region .
This conflict between observations and theory indicates that improved observational
studies and revised theories are needed to determine the role that air-sea
interaction processes play in maintaining the atmospheric MJO. Our future
research includes developing a method to estimate intraseasonal variations
in surface latent heat fluxes with better accuracy.
ACKNOWLEDGMENTS
This research is supported by NASA (NAGW-2460) and NSF/TOGA-COARE (ATM-9319483)
grants. The authors thank Dr. W. Berg (NOAA/ ERL) for kindly providing the
GPCP precipitation data, NCAR for providing access to the ECMWF analyses,
the TOGA International Project for making the TAO array data easily available
over the internet, and to Dr. Tim Liu (JPL) for providing the computer code
to estimate surface latent heat fluxes. Pete Peterson provided a great help
with graphics and data management.
REFERENCES
Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison
of methods for finding coupled patterns in climate data. J. Climate, (5),
541-560.
Crum, F. X., and T. J. Dunkerton, (1994): CISK and evaporation-wind feedback
with conditional heating on an equatorial beta-plane. J. Meteor. Soc. Japan,
(72), 11-18.
Duchon, C. E., 1979: Lanczos filter in one and two dimensions. J. Applied
Meteor., (18), 1016-1022.
Emanuel, K. A., 1987: An air-sea interaction model of intraseasonal oscillations
in the tropics. J. Atmos. Sci., (44), 2324-2340.
Hayashi, Y., and D. G. Golder, 1993: Tropical 40-50 and 25-30 day oscillations
appearing in realistic and idealized GFDL climate models and ECMWF dataset.
J. Atmos. Sci., (50), 464-494.
Janowiak, J. E., and P. A. Arkin, 1991: Rainfall variations in the tropics
during 1986--1989, as estimated from observations of cloud-top temperature.
J. Geophys. Res., (96), supplement, 3359-3373.
Jones, C., and B. C. Weare, 1994a: On the relationships between low-frequency
variations in latent heat fluxes over the tropical oceans and the Madden
and Julian Oscillation. Submitted to J. Climate.
_________________, 1994b: On the role of low-level moisture convergence
and ocean latent heat fluxes in the Madden and Julian Oscillation: an observational
analysis using ISCCP data and ECMWF analysis. Submitted to J. Climate.
Lau, K. M., and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations
in the tropical atmosphere. Part I: basic theory. J. Atmos. Sci., (44),
950-972.
Liu, W. T., K. B. Katsaros, and J. A. Businger, 1979: Bulk parameterization
of air-sea exchanges of heat and water vapor including the molecular constraints
at the interface. J. Atmos. Sci., (36), 1722-1735.
Madden, R. A., and P. R. Julian, 1971: Detection of a 40-50 day oscillation
in the zonal wind in the tropical Pacific. J. Atmos. Sci., (28), 702-708.
______________________,1972: Description of global-scale circulation cells
in the tropics with a 40-50 day period. J. Atmos. Sci., (29), 1109-1123.
______________________,1994: Observations of the 40-50 day tropical oscillation:
A review. Mon. Wea. Rev., (112), 814-837.
Neelin, J. D., I. M. Held, and K. H. Cook, 1987: Evaporation-wind feedback
and low-frequency variability in the tropical atmosphere. J. Atmos. Sci.,
(44), 2341-2348.
Wentz, F. J., 1994: User's manual to SSM/I-2 geophysical tapes. Remote Sensing
Systems, 20 pp.
Zhang, G. J., and M. J. McPhaden, 1994: On the relationships between sea
surface temperature and latent heat flux in the equatorial Pacific. Submitted
to J. Climate.
Figure 1.
Figure 1. Spatial and temporal variability of the first coupled mode of
precipitation (P) and surface latent heat flux anomalies (E). (a) First
singular vector of P. Thick solid line indicates the zero contour. (b) Same
as in (a), but for first singular vector of E. (c) First expansion coefficient
of P (mm) (solid line) and E (Wm-2) (dashed line).