Tropics

Randel and Wu (2005) pointed out that large scale temperature variances associated with Kelvin wave activity were two to three times smaller than those associated with regional or mesoscale phenomena. COSMIC provides dense enough data to be able to study these regional differences in Ep due to both the differences in source (convection) and subsequent wave-mean flow interaction. A discussion of the Madden-Julian Oscillation (MJO) is presented here. Full results, including latitude-longitude Ep maps and an analysis of equatorially trapped waves during 2007 are discussed by Alexander et al. (2008b).

The 20° x 5° x 7 day resolution of COSMIC Ep allows direct mesoscale study of the effects of deep convection on the Upper Troposphere and Lower Stratosphere (UTLS) and mid-stratosphere. Deep convection is responsible for the emission of a broad spectrum of equatorially trapped and 3-D gravity waves, some of which are resolvable using COSMIC. Tropical mesoscale convection is associated with the intraseasonal MJO (Madden and Julian 1994; Wheeler and Kiladis 1999) which has a periodicity of ~30-90 days and is especially dominant between the Indian Ocean and Western Pacific regions.

A time-height contour plot of equatorial Ep (2.5°S to 2.5°N) centered on 80°E is shown in Fig. 2. This region is above the Eastern Indian Ocean, close to Sumatra, Indonesia. Data are plotted where at least three profiles are available during each seven day interval in this cell. The background cold-point tropopause height is fairly constant at 17 km. Because the Ep are calculated over 7 km vertically, the value at e.g., 25 km is the average of the region 22-28 km inclusive. OLR data are also averaged over the region 2.5°S to 2.5°N and are plotted directly below the main Ep panel in Fig. 2. The 30-90 day bandpass filtered OLR are also shown (MJO intra-seasonally filtered).

sondj fmamjjaso

Fig. 2 August 2006-0ctober 2007 Ep at [0°S, 80° E] marked as the color contours in the upper panel, with u overplotted (units of m s- 1, solid eastward and dashed westward). The cold-point tropopause is marked by the black line. The OLR is shown in the bottom panel (black, scale to left) with the MJO bandpass filtered overplotted (red, scale to right). Missing Ep data are marked white sondj fmamjjaso

Fig. 2 August 2006-0ctober 2007 Ep at [0°S, 80° E] marked as the color contours in the upper panel, with u overplotted (units of m s- 1, solid eastward and dashed westward). The cold-point tropopause is marked by the black line. The OLR is shown in the bottom panel (black, scale to left) with the MJO bandpass filtered overplotted (red, scale to right). Missing Ep data are marked white

In most cases, high cloud tops (low OLR) correspond directly to large tropopause Ep, with the bandpass filtered perturbations showing this even more clearly. Deep convection produces gravity waves which increase the temperature variance from the mean as they propagate upwards, thus increased Ep is observed at these times. It should be noted that the Ep calculated in the tropical UTLS region over-estimates the true value because the very sharp cold-point tropopause in the tropics increases the temperature variance.

Increases in UTLS Ep are usually accompanied by enhanced stratospheric Ep, extending up to about 30 km, emphasizing that some of these convectively generated waves propagate to high altitudes.

Stratospheric QBO background mean flow filtering occurs, with changes in Ep dependent upon the QBO phase. During 2006 and 2007, the winds move from eastward to westward. As the QBO phase descends, increasing amounts of waves are filtered. This results in less wave energy reaching higher altitudes and can be noted in Fig. 2 by considering the higher 30 km Ep in October 2006 than in April 2007. The 30 km Ep is again large in October 2007 because of the QBO phase structure, consistent with the CHAMP results of de la Torre et al. (2006). Upward propagating gravity waves after June 2007 are encountering increasingly weak westward winds as they approach the 0 m s- 1 phase line. Given a typical gravity wave source spectrum centered around 0 ms- 1, temperature variance increases as the critical level is approached (Randel and Wu 2005). Furthermore, the decreasing vertical wave velocities towards the waves' critical levels increases the likelihood of their observability (Alexander and Barnet 2007). The latitude-longitude variability of gravity wave propagation, filtering, and mean-flow interaction observed by COSMIC is discussed fully by Alexander et al. (2008b).

2.3 Northern Hemisphere Winter

A detailed study of Northern Hemisphere lower stratospheric Ep during the boreal winter of 2006/07 reveals large values above mountainous regions such as the Himalayas, Canadian Rockies, and Japan (Alexander et al. 2008a). It is not possible from the COSMIC data to separate the jet-stream geostrophic adjustment contribution to total Ep from the mountain wave Ep contribution, but it is likely that some of the enhanced Ep above these mountains is due to mountain waves. Results from an Atmospheric General Circulation Model (AGCM) show that most of the large Ep observed by COSMIC is due to sub-tropical jet processes, with energy propagating upward.

Ground-based radar and radiosonde observations in Japan showed large wintertime stratospheric Ep (Murayama et al. 1994; Ogino et al. 1999). COSMIC results from the winter of 2006/07 independently confirmed this to be the case. The large Ep are a combination of the strong winter sub-tropical jet-stream above Japan and the islands' moderate topography (Alexander et al. 2008a). Previous generation GPS-RO satellite mean Ep results from the Northern Hemisphere winter did not reveal the large Japanese peak (Tsuda et al. 2000; Ratnam et al. 2004) because their data were not of sufficiently high temporal resolution. The enhancements in Ep above Japan occur on time scales on the order of a week or two, thus could not be captured previously. These short-term processes do, however, have a significant impact on the mean winter Ep and demonstrate the relevance of using the higher resolution COSMIC data for constructing seasonal means.

3 Champ

3.1 Tropics

CHAMP data have been compared with intensive radiosonde campaigns in equatorial Indonesia to study the effects on the tropopause of Kelvin wave activity (Tsuda et al. 2006). Zonal mean time-height profiles of equatorial CHAMP results showed the interaction between gravity waves and the stratospheric QBO but were unable to resolve longitudinal variability. Temperature variance increases as the 0 m s- 1 phase speed line is encountered (Randel and Wu 2005). de la Torre et al. (2006) showed that large Ep of waves with vertical wavelength between 4 km and 10 km occurred only during the eastward wind shear phase of the QBO. Potential energy calculated from CHAMP and from intensive ground based radiosonde campaigns in Northern Australia show favorable agreement (Tsuda et al. 2004).

Fig. 3 (Leftpanel) Cloud top height in Kelvin from OLR data during JJA 2004 in units of W m (Right panel) CHAMP Ep at 19-26 km during JJA 2004 with units of J kg- 1

Fig. 3 (Leftpanel) Cloud top height in Kelvin from OLR data during JJA 2004 in units of W m (Right panel) CHAMP Ep at 19-26 km during JJA 2004 with units of J kg- 1

The zonal propagation of Kelvin waves was studied previously (Tsai et al. 2004; Ratnam et al. 2006). Here we show the relationship between large scale convection and CHAMP derived 19-26 km Ep (Fig. 3) for the Northern Hemisphere summer months of JJA 2004, when the QBO is in its eastward phase. The CHAMP Ep are calculated using a similar procedure to that for COSMIC detailed above, except that the background T is calculated over a three month interval and smoothed by applying a low-pass filter with cut-off at 7 km. The T' are extracted for waves shorter than 7 km vertical wavelength. The Ep are then determined using Eq. (1) over three height intervals, including 19-26 km.

The seasonal OLR are shown as cloud top temperature in the left hand panel of Fig. 3, with colder temperatures indicating deeper convection. Cloud convection is centered over the Bay of Bengal and is part of the Asian monsoon. Another region of deep convection is above and to the east of the Philippines. The large convection above the Bay of Bengal corresponds to a seasonal mean 19-26 km Ep of 6.5 J kg- 1. Large Ep are also observed over the Indian Ocean at 70°E and east of Papua New Guinea.

Large equatorial Ep are observed east of 120°E, increasing from 4.0 J kg- 1 to 6.0 J kg- 1 in the Central Pacific Ocean. This region does not correspond to deep convective activity but what may be being observed here are (equatorially trapped) Kelvin wave-like disturbances.

0 0

Post a comment