AMERICAN METEOROLOGICAL SOCIETY 2009, VOL. , NO. , PAGES 1–2, Diurnally VaryingWind Forcing and Upper Ocean Temperature: Implications for the Ocean Mixed Layer Sarah T. Gille Scripps Institution of Oceanography and Department of Mechanical and Aerospace Engineering, University of California, San Diego Abstract. Solar radiation varies on a diurnal cycle, and therefore so do all the climate variables that it forces, including sea surface temperature (SST), wind, and in turn mixed-layer depth and upper- ocean heat storage. Satellite scatterometer data from the QuikSCAT and ADEOS-2 tandem mission have been used to estimate the am- plitude and phasing of diurnal wind variations on a global basis. Statistically significant diurnal wind variations occur along coast- lines all over the world, where they are commonly thought of as the land/sea breeze. Open ocean winds also undergo substantial diurnal variability at latitudes equatorward of 30 ◦ latitude. The phasing of diurnal winds varies with distance from the shore. Up- per ocean temperatures measured from profiling Argo floats are compared with microwave SSTs from the Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) to estimate the amplitude and phasing of the diurnal cycle in up- per ocean temperature. Differences between Argo and AMSR-E measurements imply that the diurnal cycle has an amplitude that decreases with increasing latitude, from about 0.1 ◦ C near the equa- tor to 0.02 ◦ C near 60 ◦ N/S. Maximum upper ocean temperatures occur around 18:00 local time at most latitudes. If only tempera- ture or only wind underwent a diurnal cycle, then over the course of the day, the variations would average to zero, and we would expect no net impact on climate. Since the two processes both vary, with different phasings, they are expected to have a combined (rectified) effect on the mixed-layer, and this effect is evaluated. 1. Introduction Solar radiative forcing of the Earth varies on a 24-hour cycle, and this diurnal periodicity is a fundamental character of the Earth’s climate system. Diurnal solar forcing translates into diurnal varia- tions in winds, particularly equatorward of 30 ◦ latitude [e.g. Walsh, 1974; Niino, 1987; Gille et al., 2005] and diurnal variations in upper ocean temperature [e.g. Donlon et al., 2007]. Diurnal variability is challenging to measure from satellite, because many Earth-orbiting satellites are launched on sun syn- chronous orbits, meaning that they sample at roughly the same lo- cal time on each ascending (northward) and descending (southward) satellite pass. Sun synchronous orbits offer engineering design ad- vantages, because no action is required in order to maintain the same orientation of solar panels relative to the sun for each satellite pass, and sun angle corrections are the same for each satellite pass. However sun synchronous orbits have the distinct disadvantage of sampling at the Nyquist frequency of the diurnal cycle. This study explores two strategies for evaluating diurnal variabil- ity using data from sun synchronous satellites. For scatterometer wind data, the diurnal cycle can be studied for a six month period in 2003 corresponding to the QuikSCAT and ADEOS-2 tandem Copyright 2009 by Sarah T. Gille mission. For microwave sea surface temperatures, the diurnal cycle can be evaluated by comparing sea surface temperatures measured by the Advanced Microwave Scanning Radiometer (AMSR-E) with upper ocean temperatures measured by profiling Argo floats. Diurnal variability in wind and sea surface temperature do not appear to have consistent timing everywhere. While sea surface temperature follows a fairly consistent diurnal cycle, with maxi- mum temperatures in mid to late afternoon, diurnal winds vary with distance from the coast. The final part of this study explores the impact of this effect on ocean mixed-layer depth. 2. Diurnal Winds The SeaWinds scatterometers aboard QuikSCAT and ADEOS-2 measure wind speed and direction with roughly 25 km resolution. QuikSCAT, launched in 1999, has equatorial overpass times of 6 am and 6 pm. The ADEOS-2 satellite, which flew for six months from April through October 2003, crossed the equator at 10:30 am and 10:30 pm. Gille et al. [2003] made use of the QuikSCAT morning–evening wind differences to assess the basic characteris- tics of diurnal wind variability. Gille et al. [2005] used the four measurements per data collected from the QuikSCAT and ADEOS- 2 tandem mission to consider how winds rotate along an elliptical hodograph through the course of the diurnal cycle. Figure 1 shows that the amplitude of the diurnal wind stress varies substantially depending on location, with strong diurnal winds near coastlines and elevated diurnal amplitudes typically near large orographic fea- tures. Statistically significant diurnal variability occurs throughout the tropics [Gille et al., 2005]. As reported by Gille et al. [2005], these global findings are generally consistent with linear theory [Alpert et al., 1984]. The timing of the maximum wind, shown in Figure 2, changes with location. Near coastlines there is evidence that maximum winds propagate offshore progressively through the course of the day. Within the tropics, the time of maximum wind varies progres- sively across the Pacific Ocean [Gille et al., 2005]. 3. Diurnal Upper Ocean Temperatures Compared with diurnal wind variability, diurnal cycles in sea sur- face temperature are harder to extract from satellite data, because the radiative properties of the lower atmosphere could change on a diurnal cycle, thus providing different biases for daytime and night- time sea surface temperature calibration data. In order to evaluate the diurnal variations in upper ocean temperatures, AMSR-E SSTs were compared with co-located upper ocean temperatures measured at 5 m depth (T5m) by profiling Argo floats. AMSR-E (launched in 2002) measures temperature at 1:30 am and 1:30 pm local time. In contrast, Argo floats (which achived global coverage roughly start- ing in 2002) rise to the surface throughout the course of a 24-hour cycle, and are therefore able to measure T5m at all times of day. For these comparisons, daytime and nighttime AMSR-E data were treated separately. Co-located AMSR-E and Argo data pairs were sorted based on the time separation between measurements and bin averaged based on time separation. Since AMSR-E measurements occur within a narrow window in local time, the time variability 1