Towards a Global Tropical Ocean Moored Buoy Array for Climate TAO Project Office Dongxiao Zhang Mike McPhaden NOAA/PMEL and JISAO/UW I am Dongxiao Zhang, will give you a brief about the progress we made and implementation plans towards building a global tropical ocean moored buoy array for climate research and prediction. The team of TAO project is formed by PMEL and JISAO scientists. My–self is a JISAO Research Scientist, and I am proud of being one of the team. JISAO scientists participate every aspect of the TAO day-to-day operation, from hardware to software design and maintainance, from buoy deployments to data delievery and public outreach, and of course our own research. Outline: 1) Background on planning for the development of an Indian Ocean Moored Buoy Array; 2) Recent progress towards implementation; 3) Description of first data from Indian Ocean ATLAS; 4) Implementation challenges and immediate plans for 2005. JISAO 5-Year Review Seattle 19 April 2005

TAO System Overview The TAO buoy measures ocean and surface meteological conditions. The operation of TAO system consists of two parts: maintain observational array and data dissemination. This includes Buoy construction/design/update, sensor/database calibration, buoy deployment and recovery. The satellit sends the data back to our data management system in real time. Data quality controls are conducted daily, weekly, and monthly in different levels, to ensure highest standard of data quality. Quality controlled data are then sent to NODC/NCDC, meteorology and oceanographic communities.

Major achievements and milestones 1533 sea days on 8 ships (including Ka’immimoana and 5 foreign ships). 500th NextGen ATLAS deployed on March 4, 2005. TAO data return rate constantly over 92%. collaborative researches (EPIC, ARM, TRMM, CO2 time series …) In the past 5 years, we spent 1533 days at sea on 8 ships, including Kaimimoana the designated ship for servicing TAO array in the tropcal Pacific, and 5 foreign ships. In the last five years, next generation atlas mooring began to be used exclusively in TAO. 500th buoy was just deployed in March 4th, 2005, at 2s and 140w. Data return rate from TAO buoys has been constantly over 92%. The presence of TAO array and ship time in the tropical Pacific, provide excellent opportunities for other investigators. Over the past 5 years we had a number of collaborative researches, just name a few: EPIC in the eastern equatorial Pacific, the DOE ARM radiation measurements in the western Pacific, the TRMM related rainfall studies, Dick Feely and Chris Sabine at PMEL have put CO2 sensors on 4 of the TAO buoys, we now have CO2 time series in real time. Data display

Major achievements and milestones TAO web site “highly recommended to any interested in ENSO science.” EOS, 10 September, 2002. Over 1.9 million hits each month. Data display and delivery launched on 15 August 2000. 51,534 data deliveries through our own web site. 596,434 files delivered. 219 scientific papers in refereed journals used TAO data over 5-years. We maintain a TAO web site about ENSO with other related links to data and prediction centers. A Sept.10th 2002 eos article reviews our web site. In its conclusion, “This site is highly recommended to any interested in ENSO science. Bookmark this site and visit it frequently.” Indeed we got hit frequently. There were average of 1.9 million hits each month over the last 12 months. To make the data easily available to everyone, we designed and maintain the data display and delivery web site. Like the one shown here for TAO/TRITON buoys. You can down load data files, make plots of time series, time sections, lon-lat maps, and depth sections for all the variables in real time. Since the launch of this web site on Aug. 15, 2000. There were over 51 thousand data deliveries, with 596 thousand data files delivered from our web site alone. This does not include the files tranfered to NODC and prediction centers in real time. Over the past 5 years, there were 219 scientific papers that we know in refereed journals used TAO data. These are the papers that the authors informed us when published. There could be many more we do not know. We are sea going oceanographers, we know how difficult it is to get the data back from the tough seas, and we know oceanographic observational programs are expensive. our biggest dream is to have the data used widely.

Major achievements and milestones MOU between France, Brazil, and US extended for 2 years (to 2006). Southwest and Southeast extensions approved by PIRATA SSG in Dec 2004; Southwest extension is funded. Use of PIRATA data by researchers and operational centers (e.g. ECMWF, Meteo-France, UKMO, etc) is increasing. CLIVAR/OOPC review of PIRATA planned for Oct 2005 in Toulouse; results will provide basis for continuing array as permanent component of GOOS. In the tropical Atlantic, we are invloved in deploying and maintaining 10 PITRATA buoys, shown by the yellow squares. This is a collaboration between France, Brazil, and US. The multi-lateral agreement was just extended for 2 years to 2006.

Major achievements and milestones Draft strategy developed at first IOP meeting at Pune, India in Feb 2004. Sustained multi-national effort envisioned in support of monsoon research and prediction. Moorings already deployed by Japan (JAMSTEC), US (PMEL), India (NIO, NIOT). Already deployed It is just one component of a comprehensive IO observing system. Fast sampling (daily) is key to capturing intraseasonal variability which is a major component of variability in the Indian Ocean. No other observing component will provide the necessary information on this time scale. Array is intended to provide multi-variate data ocean-atmosphere variations in key regions of equatorial waveguide, thermocline upwelling ridge between 5-10S, Bay of Bengal, Arabian Sea. The arrray does not provide good coverage of eastern and western boundaries. Fast sampling (daily) is key to capturing intraseasonal variability which is a major component of variability in the Indian Ocean. No other observing component will provide the necessary information on this time scale.

First Data from Indian Ocean ATLAS Moorings Deployed 22 October 2004 This panel shows the real time wind, current and temperature in the upper 500 m of the Indian Ocean at 0, 80E. Plots like this are made available from the beta version of our data delivery page, which will go public in May after JAMSTEC installed their system. I will not go into the details of this plot. It shows weekly mixed layer variability, current reversal and spreading of the thermocline associated with the monsoon transition. A surprising mixed layer warming under a upwelling favorable NE monsoon. But would like to point out an interesting process. While the winds change from sw to ne monsoon, the mixed layer in white shoals, the thermocline spreads, the sst and mxed layer temperature under upwelling favorable eastliers gets warmer, instead cooler. All fluxes can be estimated at this flux reference site, and is found not to be the cause of the warming. Likely processes would be vertical mixing and horizontal advections across SST gradient. October to December is the so called Monsoon Transition season (between SW and NE Monsoons). Winds are highly variable and mostly westerly, unlike the more steady trades in the Atlantic and Pacific. Surface currents (20 m) are eastward rather than westward as in the Pacific and Atlantic until late January because of this westerly wind forcing. They reverse when the zonal winds reverse during the development of the Northeast Monsoon. Mixed layer depth is in white. It shoals and the thermocline spreads when the zonal winds reverse. Note the warm mixed layer temperatures usually over 28C and the pronounced week to week variability in thermocline temperatures asnd mixed layer depths associated with the variable winds and currents. Interestingly, when easterlies develop and the mixed layer shoals, we get the WARMEST temperatures.

The President’s FY06 Budget for NOAA Climate Observations and Services “$3.2 million to expand the Tropical Atmosphere Ocean array and the Pilot Research Moored Array in the Tropical Atlantic into the Indian Ocean. This expansion will enhance NOAA's capability to accurately document the state of ocean climactic conditions and improve seasonal forecasting capability.” (http://www.noaanews.noaa.gov/stories2005/s2386.htm) Other activities covered by this funding: Support the technological development of the next generation of moored buoys. Add salinity sensors to the TAO array to improve seasonal-interannual forecasting. Upgrades for 4 TAO and 3 PIRATA moorings to ocean reference station quality for satellite and model research Providing 4 additional buoys for the PIRATA array in the hurricane-genesis region of the Atlantic Ocean for improved understanding of ocean-atmosphere interactions on hurricane development. Challenges: Funding

Awards Grace Hopper Government Technology Leadership Award, 2003. “Leadership in the innovative application of information technology that contributes to the advancement of scientific knowledge and its application.” NOAA Outstanding Paper Awards (4 from TAO over the past 5 years). In 2003, TAO won the Gracie award for the leadership in the …. Over the past five years, we had 4 NOAA outstanding paper awards from our own research. In the following I will talk a little bit of my research on decadal changes of ocean circulation, since the study heavily relied on the data collected by TAO project over the years. It highlights the need of a sustained observation net-work, like TAO, for better understanding of long term climate variability.

PDO Basin Scale Fluctuation of the Ocean- Atmosphere System Affects the climate of North America (Latif & Barnett, 1994; Cayan et al, 2001) Affects Pacific marine ecosystems and the global carbon cycle (Mantua et al, 1997; Hare & Mantua, 2000; Chavez et al, 2003; Peterson & Schwing, 2003) Linked to decadal modulation of ENSO (Trenberth & Hurrell, 1997; Latif et al, 1997) Our motivation is to better understand the Pacific decadal oscillation, characterized organized patterns of SST, slp, and surface winds. Upper panel shows the positive phase of the PDO, lower panel shows the pdo index. Updated index can be found in this jisao pdo web site. By and large, it looks like a 20 year enso, with a major shift occurred around mid-70s. But anomalies in the equatorial SST is broader than those in ENSO in meridional scale, so is the weakening of the trades in western central Pacific. Its far-reaching influence on mid and high latitudes is also similar to ENSO. The PDO is shown to affect …,… the later two biological studies suggest a possible phaser reversal of the PDO. Finally, the PDO is linked to …. The mechanism of the PDO is not well understood. The understanding of the PDO is complicated by the lack of understanding of what the ocean has been doing on decadal time scales below the surface. In the following I will show you a relationship between the changes in ocean currents in the pycnocline and the SST anomalies in the central eastern Pacific. These pycnocline currents transport cold waters from subtropics to tropics. http://jisao.washington.edu/pdo PDO

Mean Circulation in Pycnocline 14 Sv (1 Sverdrup=106 m3 s-1) 7 Sv (Integrated over 22.5-26.5 kg m-3) 9°N 9°S X Here shows the mean circulation and PV on isopycnal surface 25.0, the center of upper ocean pycnocline. Arrows show the pathways of subducted waters making their ways into the equatorial Pacific, through interior pathways and western boundary pathways. The convoluted pathway in the north is due to the blocking of high PV ridge under the ITCZ, which can be viewed as a choke point. The 50 year mean transports, across the interior sections at 9N and 9S, integrated over 22.5 and 26.5 sigma_theta, are 7Sv from the north and 14 Sv from the south, respectively, resulting in 21 Sv equatorward convergence in the interior pycnocline . We published a paper in 2002 showing a slow-down of this equatorward convergence, and related it to the reduction of tropical upwelling, and a warming of SST. The paper generated a lot interests among modelers, using numerical models to assess the changing circulation. A current consensus is that, the variability of total pycnocline convergence is dominated by the changes in the interior ocean, with limited partial compensation from the western boundary. (µf∂r/∂z) McPhaden & Zhang, Nature(2002)

Pycnocline Convergence SST anomaly (9°N-9°S, 90°W-180°) Here shows the reduction of the interior pycnocline convergence from the mid-70s to the 90s, by the horizontal line, the vertical line are the error bars of our cauculation. The time series is the SST anomaly averaged over central eastern tropical Pacific, the cold tongue area. 5-yr running mean was applied twice to emphasize the decadal variability. The rising of SST can be interpreted as less cold water making its way to the equator. We now update the convergence for the last several years between 1998-2003, showing a dramatic increase of the transport. While the equatorward pycnocline convergence in the last several years has increase to the level of 70s, the eastern central tropical Pacific SST anomaly had a sudden drop to a negative value. Pycnocline Convergence

Interior Convergence and SST Anomaly Interior Convergence and Cold Tongue SST Anomaly A surprising rebound of the circulation after 1998, lead us to develop a convergence transport time series, to better characterize the changing circulation. The black curve shows this transport time series, it uses the data for every 7-year centered from 1958-2000. So the time series is equivalent to 7-year low passed. The SST in the cold tongue area (9n-9s, dateline-90w) is plotted together. The two are highly correlated at –0.92, notice that the transport is plotted from high to low values. A slow down of the circulation corresponding to a warming SST. You can see both the recent changes after 1998, and anomalies in late 80s, in the two time series. The amplitudes of decadal variability of the two time series are as large as their trend over the past several decades. The blue line here is the trend of the SST. SST anomaly (9°N-9°S, 90°W-180°) Interior Convergence and SST Anomaly

Multi-Taper Spectrum of SST (9°N-9°S, 90°W-180°) 500 year PICTRL What are the implications of the correlation between transport and sst. Here am show the spertrum of the eastern tropicla sst index in the 500-year preindustrial controll runs fromtwo models. The cnrm with low correlation of 0.3, the miroc-medium with high correlation of 0.9. The different between them are the spectrum peaks on 40 year, 20 year and 10 year band, above 99% significant level against the red spectrum, in miroc, and no energy peaks in the decadal multidecal bands in cnrm. The cnrm also has a strong narrow band of ENSO type of variability. This indicates that the communication between the tropics and subtropics through ocean curents greatly increase the energy in decadal and multidecal periods. CNRM MIROC-ME Spectrum

Multi-Taper Spectrum of SST (9°N-9°S, 90°W-180°) ERSST: 1900-2004 MIROC-Hi: 1900-2000 CNRM: 1860-1999 MIROC-Me: 1850-2000 Spectrum

Multi-Taper Spectrum of SST (9°N-9°S, 90°W-180°) ERSST: 1900-2004 MIROC-Hi: 1900-2000 CNRM: 500 year MIROC-Me: 500 year Spectrum

Interior Convergence and Temperature Variability Interior Convergence and Temperature Anomaly in the Pycnocline Further investigations show the SST changes were linked by the temperature changes in the pycnocline, in the eastern equatorial Pacific, shown by the red curve. It is caused by the upwelling changes associated with the transport convergence. Transport values are replotted here, in black. The tight correlation between the transport index and the eastern tropical sst index provide a way to evaluate the coupled climate models in simulating the tropical climate. We have been looking at this relationship in 11 climate models, that will be used to project the climate change in the next IPCC report, and found that about half of the models show significant correlation, although the variance of the transport are still underestimated. (sq = 25 kg m-3) (2°N-2°S, 80°-120°W) Interior Convergence and Temperature Variability

Interior Ocean Pycnocline Transport Changes The transport values for the two recent periods are updated here. Owing to the large over lap, the transports in 92-98 are similar to our previous estimates in 90-99, however, the transport values in the latest period is significantly increased to almost the level before mid-70s across both 9n and 9s.

Correlation between STC and SST in MIROC-High Resolution Here is the correlation map between the interior STC transport time series and the global SST in the high resolution model. Correlations range from –1 to +1. The well organized patterns is consistent with the SST patterns assoicated with the Pacific Decadal Oscillation in observation. STC and Global SST

Interior STC Transports in 9 Coupled Climate Models The interior pycnocline convergence time series in 9 coupled models. The time axis is meaningful because they are all forced by the history of green house gases in the 20th century. These are transport anomalies. The axis is ranged from 8 Sv at the bottom to –6 Sv at the top. Not all of them have a downward trend or a slowing-down trend like the miroc-high resolution model and the observation (note that upward trend plotted in the figure is actually a slowing-down trend, because I want to compare with the temperature). The UKMO-Hadley center model has the largest variability but has an upward trend or speeding-up trend because of this big bump. STC in all models

Variability of Tropical Pacific SST (EOF1) in 9 Coupled Climate Models The time series of the 1st EOF of tropical Pacific SST in all the models. There are good correspondence between the ups and downs in the PC1 and the STC shown in the last slide. Medium resolution: lats: 0.56 deg in the equatorial ocean, 1 deg. In mid lat. 1.7 at the poles, lons: 1.5 deg. High resolution: Lats: 0.56 deg. Everywhere Lon: 1.1 deg every where PC1s in all models

Correlation between Variability of STC and Tropical Pacific SST (EOF1) This chart surmarizes the correlation between the STC transport and the PC1. In observation, on the left, and in the models. The blue bars are the correlations with trends included, the green bars are the correlations with linear trends removed. The last four models give more significant correlations at about 0.8, the IAP and CSIRO model also have high correlation. Observation CNRM CSIRO GISS-AOM GISS-ER IAP MIROC-High MIROC-Med. MRI UKMO STCs and PC1s

Annual Mean STCs (Total and Interior) The plot shows the annual mean stc transport in observation and models. The blue bars represent the total STC transports, the green ones represent the interior transport. The models that simulate more realistic mean STC strength, tend to have more significant correlation between the STC and tropicla SST variability, except this GISS-ER model which has a strong STC but low correlation, this model has very little decadal variability but a steady trend on its SST field, and it is a very coarse resolution model, with euc like current shifted to the southern hemisphere. Observation CNRM CSIRO GISS-AOM GISS-ER IAP MIROC-High MIROC-Med. MRI UKMO Mean STCs

PV and Data Distribution Potential Vorticity (sq = 25 kg m-3) CTD Casts to 900 m July 92-June 98 11,585 July 98-June 03 Potential vorticity on 25 sigma_theta surface and ctd casts deeper than 900m in the two periods are shown here. Data in the later period are mainly obtained from TAO/TRITON sections and some argo float profiles. Relative to the prior period, the PV ridge along about 9N during 98-2003 is weakened and retreat further to the east, so that the restriction for the northern water to reach the equator is reduced. 6,729 PV and Data Distribution