Bug found in NOAA’s CarbonTracker System

“In preparing the CT2011 release, we discovered a bug in model sampling of observations at the South Pole during the 2002-2005 period.”

From NOAA:

Dear Colleagues,

We regret to announce that the 2011 release of CarbonTracker has been delayed.
In preparing the CT2011 release, we discovered a bug in model sampling of
observations at the South Pole during the 2002-2005 period. We believe this is
due to a change in how the model handles convergence of the lat-lon grid at
Earth’s poles, and that this sampling bug therefore is restricted to the South
Pole. Unfortunately, measurements from NOAA’s South Pole observatory are
fundamental to constraining large-scale fluxes and are given strong weighting in
CarbonTracker. As a result, we must reject the existing CT2011 simulations.

This bug does not affect previous CarbonTracker releases, and data available
from http://carbontracker.noaa.gov are still valid.

It will take us a month or two to recover from this problem. We intend to go
forward with the planned CarbonTracker release, but fluxes and CO2 fields for
2010 from this set of simulations will not be available before April. In the
mean time, we have decided to also move forward with a near-real time
CarbonTracker product. CT-NRT will use the climatology of optimized fluxes (and
projected fossil fuel emissions) with up-to-date transport. This product, while
not optimized against the most recent CO2 observations, simulates the majority
of seasonal and synoptic CO2 variability and agrees fairly well with in situ
observations. It may be suitable for some research purposes.

We recognize that researchers have come to count on regular releases of
CarbonTracker, and we are attempting to meet those needs in a timely fashion.
However, CarbonTracker remains a research project, and delays such as this one
are to be expected.

Best Regards,

Andy Jacobson, Ken Masarie, Mike Trudeau, John Miller, Pieter Tans, and the
CarbonTracker team

2 thoughts on “Bug found in NOAA’s CarbonTracker System”

  1. How do you ‘sample a model’? They nearly seem to be honest about the CarbonTracker project, saying “It may be suitable for some research purposes.” The really scary thing is that “researchers have come to count on regular releases of CarbonTracker.” Really? Relying on samples of a model that may be suitable, and calling that ‘research’? I bet some people even call this ‘studying the climate’.

    It’s amazing what you can come up with when you have too much money.

  2. You can guarantee when they’ve fixed it, they will discover that things are worse than they thought.

Comments are closed.