Warning: This writeup was done to help organize my thoughts prior to the presentation.
The actual presentation may have been quite different!

Fall AGU 1998  San Francisco CA

Presented at the Salinity from Space Session  of the Ocean Sciences Section. (20 minute slot).

Diurnal Cycle of Rainfall and Surface Salinity in the
Western Equatorial Pacific
Meghan F. Cronin and Michael J. McPhaden
Acknowledgments: Roger Lukas (UH)

    I.   Motivation for study
    II.  Data
               COARE Enhance Monitoring Array
    III.  Diurnal Cycle Analysis
                   Temporal Variability
                   Spatial Variability
    IV.  Summary
Before getting started, I would like to acknowledge Roger Lukas for providing some of the 3m sea surface salinity (SSS) data that I'll be discussing here.

The outline of the presentation is as follows:
I'll begin with a brief discussion of the motivation for studying the diurnal cycle of SSS and rainfall.
I'll then discuss the data and results from the analysis and will end with a summary.

Motivation for Analysis of Rainfall and SSS Diurnal Cycles

Cronin, M. F., and M. J. McPhaden (1998): Upper ocean salinity balance in the western equatorial Pacific.  J. Geophys. Res., 103, 27567-27587.

Why are we doing this study:
    1) From a technical perspective, undersampling the diurnal cycle can produce aliasing. Also, depending upon which part of the dc is undersampled, the fields can be either biased high or low. This is a particularly important issue for polar orbiting satellite measurements.

    2) In a recent study which has just come out in the November issue of JGR, Mike and I found that on time scales of a month to the record length (1-2 years), variability in western equatorial Pacific SSS was due primarily to zonal advection of the fresh pool, and only weakly correlated with local rainfall.  The correlation between local rainfall and SSS increased though for shorter time scales.
Here we ask,  on the diurnal cycle, can we relate SSS variability to rainfall variability?

    3) One might ask Why diurnal? Why not 2day or 3day cycles? The diurnal cycle in heat flux is one of the strongest clocks in nature.  Every morning, the sun rises, and every evening it sets. This clock has a profound effect on the ocean surface, surface heating, and convection.  To understand its effect on the full hydrological cycle, we must also ask, how the diurnal cycle in rainfall effects the sea surface salinity, and the buoyancy of the ocean surface.

4) Finally, to fully understand this hydrological cycle, we must understand the spatial and temporal variabiltiy. How does the dc depend upon its location relative to the SPCZ, ITCZ, equator,...? How does the MJO affect the dc?

 In this analysis, I use hourly data from Tropical Ocean Atmosphere (TAO) PROTEUS and ATLAS  buoys which were enhanced for the Coupled Ocean Atmosphere Response Experiment. 2 sites (0,156E and 0,165E) had PROTEUS buoys, such as shown here.

Hourly rainfall is measured with an Optical Rain Gauge (ORG), which computes rain rates from scintillations caused by raindrops falling through a near-infrared light.  ORGs rainrates are high relative to radar and satellite rainfall rates, although the correlations with various satellite rainrates is as good as the satellite to satellite correlations.  Factory calibration are the largest source of error and are typically 20-25%, and occasionally larger.  Because of this, ORGs are not used at present in the TAO array.

SSS and SST  were measured by SEACAT T&C sensors mounted either on the toroidal bridal at 1m or on the wire at 3m. After postprocessing, SSS errors are approximately 0.02 psu.

These 2 sites also measured shortwave radiation.  Thus, using the COARE v2.5b bulk algorith to compute the latent and sensible heat fluxes, and a bulk algorithm to estimate the net longwave radiation, we have hourly timeseries of the net surface heat flux at these 2 sites.

The COARE Enhanced Monitoring Array is shown here superimposed upon the mean SSMI rain rate and mean GOES precipitation index. "X" indicate sites with Optical Rain Gauges; "O" indicate sites with either 1m and/or a 3m Sea Surface Salinity measurements.

Note that while all the sites are within the convection region of the western equatorial Pacific, there is spatial variability in the mean., with stronger convection off equator in the South Pacific Convergence Zone and the Intertropical Convergence Zone in the northern hemisphere.  On the equator, convection has a maximum near the center of the array. Later, I'll be discussing this in relation to the spatial variability of the diurnal cycle.

This figure shows the 1m SSS  time series at 0156e, 2s156e, and 0165e. To produce the composite., the 0,165E time series was shifted by 1 hour to put it in the same time zone as the other records, and then all sites were average together.

Notice that the low frequency dynamic range is about 1-2 psu. These downward 1-2 psu fresh spikes are caused by rain. The fact that they are spikes, rather than steps, is because of other processes such as mixing. and advection. Although smaller, these spikes are still evident in the composite.

These are the 3m SSS. Again, fresh spikes in individual sites are about 0.5 psu and about half that in the composite.  And just to be complete,... these are the sites that went into the rain rate composite.
   In the left column, I have the diurnal cycles of the net surface heat flux, 1m and 3m SST anomaly.   The right column shows the rainfall, 1m and 3m SSS anomaly diurnal cycles. The time axis goes from midnight to midnight and is repeated in the right panel.  The thin lines are the 95% confidence intervals. These cannot be seen on the thermal diurnal cycles because they are narrower than the thickness of the main lines.  This is Nature's  Clock. During daylight periods when Q0 > 0, SST warms. When Q0 is negative, SST cools.

Consistent with other investigators' results, we see peak rainfall  near 4 am.

SSS also has a clear diurnal cycle, although it is weak -- peak to peak amplitude fo rthe composite is about 0.01 psu at 1m and smaller at 3m.  Surprisingly,  SSS is saltiest during the period of highest rainfall. What's happening is that nighttime mixing is injecting that freshwater below the surface and bringing up saltier water to the surface.

Although the freshwater from rainfall tends to stabilize the ocean surface,  looking at the buoyancy scale on the right axes, we see that the destabilizing thermal buoyancy flux is 2-3 times larger.  Daytime rainfall, although substantially weaker than the nighttime peak, is sufficiently large to produce a 0.01 psu freshening.

Now lets look at the rainfall and SSS diurnal cycles during convectively active and suppressed periods.  This figure shows the time longitude plot of the OLR along the equator in the Pacfic from Feb 92 through April 1994. Cold cloud tops have low (< 210 W/m2) OLR values.  During suppressed periods, OLR is high (> 220 W/m2).  This 30-60 day oscillation is a manifestation of MJO. Left column shows the dc of rainfall and 1m and 3m SSS during convectively active periods; the
right column shows the diurnal cycles during convectively suppressed peridos.

Not surprisingly, rainfall is more than twice as high during the convective periods. In both periods, rainfall peaks at nighttime. However, during the suppressed periods, there is also a secondary maximum in the afternoon.

Although the 3m SSS anomaly is essentially flat during the suppressed periods, the 1m SSS is as large as during the convective period. This is consistent with the formation of very shallow rain puddles  due to afternoon showers during a period of light winds.

Now lets look at the spatial variability of the rainfall and 3m SSS diurnal cycles.  Each grouping shows the diurnal cycle of rainfall and 3m SSS at individual sites. These are laid out in rough geographic coordinates: 0,154E;  0,156E; Rain at 0157.5E/SSS at 0161E; and 0165E;   and 2N and 2S156E.

While most sites have a primary peak rainfall at 4 AM local, at 0156E the primary peak is in the afternoon, and at 0157.5E no clear diurnal cycle is present, -- perhaps because this is a shorter record, or  because of its location near the center of the convection max on the equator (I don't know why this would be).

The 3m SSS diurnal cycle also has interesting spatial variability. The amplitude tends to increase to the west, perhaps because mixing is reduced.  At 0154e, the peak to peak amplitude is about 0.015 psu at 3m. 0 156E 1m SSS anomaly has an amplitude of roughly 0.02 psu.

Also, quite interestingly,  the dc phasing changes zonally.  In the east at 0165E, the anomalous low salinity peaks at about 9AM; at 0161E and 156E, it peaks at about 3PM; in west at 154E, it peaks in the evening at about 8PM.   As discussed earlier, the phasing of the SSS diurnal cycle is due to the relative phasing of the rainfall  and diurnal mixing, ultimately between the thermal and rainfall buoyancy fluxes.

This phase shift in the SSS diurnal cycle causes the composite diurnal cycle to be much weaker than at individual sites.


In the western equatorial Pacific, rainfall occurs preferentially during nighttime at 4am.   On the equator, and during suppressed periods there is also an afternoon rainfall maximum.  At 0,156E, this afternoon peak was larger than the nighttime peak.

SSS at 1m and 3m have weak diurnal cycles, with 0.005 - 0.02 psu peak to peak amplitudes.

The SSS diurnal cycle is controlled by daytime rain and nighttime mixing. The relative strengths and timing of the daytime rain and diurnal mixing, cause large spatial variability in the phasing of the SSS diurnal cycle. 

SST and SSS are not coupled on the diurnal time scale. Although SSS is affected by the SST diurnal cycle, on this time scale, buoyancy associated with the SSS anomaly is too weak to affect the SST diurnal cycle.

Return to Meghan Cronin's Home Page

Meghan F. Cronin
Pacific Marine Environmental Laboratory
7600 Sand Point Way NE
Seattle, WA 98115 USA
Return to:
Meghan Cronin's Home Page
Privacy Policy | Disclaimer