A journal on how scientists who are accustomed to Unix environments can switch to OS X and feel comfortable there. See it here. Includes information on getting Ferret and Vis5D running on OS X.
A note of caution in interpreting this movie: early on in the history of CO2 measurements, most of the stations were situated in such a way as to sample "clean" air from the marine boundary layer. Lately, many continental stations have been established. These stations see very complex CO2 signals from the land biosphere and from anthropic sources. The measurement data from these stations are frequently characterized by large excursions (both positive and negative). So...the early part of record is biased towards nicely interpretable marine signals, probably representing carbon dioxide levels after much east-west mixing in the atmosphere, and the later part has some wild, difficult-to-interpret land signals. Nonetheless, I think this remains a nice way to see the increase in atmospheric concentrations and the seasonal cycle. Compare this to the flying carpet depiction of carbon dioxide concentrations.
Feel free to share this animation with others, but please remember to give credit to the scientists who have worked so hard to collect the carbon dioxide measurements.
Anthropogenic carbon flux in 1995 estimated using MOM3-LL
Once released into the atmosphere, carbon dioxide is practically inert. Since there is no signficant chemical sink for CO2 in the atmosphere, the main method by which it is removed is dissolution into the ocean. This process is "slow", in that it takes place on timescales of decades to centuries. In addition, atmospheric sources of carbon dioxide (principally oxidation of CO and CH4) are much smaller than the surface sources and sinks. Thus we say that carbon dioxide has a "long" atmospheric lifetime.
If we assume that atmospheric carbon dioxide is a passive tracer--that is, that the only sources and sinks of CO2 are at the land and ocean surfaces--then there is some hope of inverting CO2 concentration measurements for the magnitudes and geographic distributions of the surface sources and sinks. The general idea here is that some surface flux of CO2 (e.g., a major industrialized area) would have a persistent "footprint" in the atmosphere. You would hope, for instance, that measurements made at a site that is climatologically downwind of a CO2 source to be higher than measurements made just upwind. Furthermore, you might hope to figure out how much CO2 is being emitted by that source by analyzing the difference in concentrations between the upwind and downwind sites.
Using transport estimates from general circulation models (GCMs) in both the atmosphere and oceans, I am trying to establish the extent to which we can retrieve source and sink information from in situ measurements of carbon dioxide concentrations. The problem with this is that the amount of data available for this task is currently not sufficient to allow us to get surface flux estimates on continental scales. The sparseness of the data network, coupled with significant uncertainty about atmospheric transport (since we have to use coarse-resolution GCMs) means that we cannot quite resolve continents from one another. This means, for instance, that is currently not possible to make a robust distinction between Eurasian and North American CO2 emissions.
I am building on work conducted by many other colleagues, including the researchers involved with the TransCom model intercomparison project.
40-month mean SeaWiFS chlorophyll
One of the major uncertainties in carbon cycle research involves the response of ocean biology to climate change. Other members of our group are engaged in the development of a predictive ecosystem model to be integrated into an existing ocean general circulation model.
The compounded complexity of an ecosystem model with marginal data constraints built on top of a nonlinear circulation model with known deficiencies presents special challenges. I am trying to use simple statistical models to provide insights into the dynamics of SeaWiFS chlorophyll estimates. We hope that these simple statistical models will help us to identify important processes for guiding development of more mechanistic models.
This work is funded by a NASA grant I helped to write with Jorge Sarmiento, Yuan Gao, and John Dunne.
I am helping to develop a method for providing improved estimates of oceanic Redfield (remineralization) ratios. These are the average stochiometric ratios of elements as they are taken up by phytoplankton during photosynthesis, and the ratios at which elements are returned in oxidized form as nutrients during bacterial remineralization of detrital organic material.
The chemical composition of seawater at any given point is assumed to be the result of two processes: mixing among a small number of distinct water masses, and biological activity. Each of the endmember water masses is characterized by unique set of observables: temperature, salinity, and nutrient concentrations. As water parcels move from the surface where they form these characteristic properties, photsynthesis and remineralization act to change their nutrient concentrations. The changes in nitrogen, carbon, and oxygen, it is hoped, can be linked to the change in phosphorous via these fixed stoichiometric ratios. The problem is to disentangle the effects of mixing and biological processes on a given sample of seawater.
This is inherently a nonlinear problem, as we are simultaneously trying to guess the change in phosphorous concentration and the remineralization ratios which multiply it. I am currently developing methods to compare and contrast the ability of linear and nonlinear optimization techniques to retrieve robust estimates of these vital parameters.
This work is being carried out in collaboration with Ben McNeil and Bob Key.
My dissertation work involved the analysis and interpretation of long-range acoustic transmissions in the Pacific ocean. I built a Kalman filter inversion system capable of ingesting hydrographic observations and acoustic travel time signals to estimate the three-dimensional time-varying ocean sound speed field. The figure above shows the hydrographic observations (dots), the acoustic sections (sources S to receivers R), and a thousand-kilometer-square box in the center of the acoustic array, for which I computed volume-averaged heat content anomalies. Click on the map to see a larger version.
Sound speed variations in the ocean are principally due to changes in water temperature. My work so far has failed to detect significant heat content anomalies in the Northeast Pacific. However, the only acoustic transmissions we have used so far are for 5 months in 1987. Recent work by Levitus et al. (Science 287, pp. 2225-2229, March 2000) agrees that despite large interannual variations in heat content in this area of the ocean, 1987 was near the climatological mean.
This is work I conducted at Penn State University with John Spiesberger, Anatoly Fabrikant, Bruce Einfalt, and Mark Keller, among others.
Using historical hydrographic data, we show evidence for strong Rossby waves in the Northeast Pacific between 1968 and 1991 directly after significant El Nino events. In the latter part of the record, we compare observations with simulations made by the NRL layered ocean model. Evidence suggests that Rossby wave phase speeds exceed predictions of linear theory. This is work I conducted at Penn State University with John Spiesberger, with help from Harley Hurlburt at NRL Stennis.
