Timely detection of changes in the meridional overturning circulation at 26° N in the Atlantic

It is investigated how changes in the North Atlantic meridional overturning circulation (MOC) might be reliably detected within a few decades, using the observations provided by the RAPID-MOC 26 degrees N array. Previously, detectability of MOC changes had been investigated with a univariate MOC time series exhibiting strong internal variability, which would prohibit the detection of MOC changes within a few decades. Here, a modification of K. Hasselmann's fingerprint technique is used: ( simulated) observations are projected onto a time-independent spatial pattern of natural variability to derive a time-dependent detection variable. The fixed spatial pattern of natural variability is derived by regressing the zonal density gradient along 26 degrees N against the strength of the MOC at 26 degrees N within the coupled ECHAM5/Max Planck Institute Ocean Model's (MPI-OM) control climate simulation. This pattern is confirmed against the observed anomalies found between the 1957 and the 2004 hydrographic occupations of the section. Onto this fixed spatial pattern of natural variability, both the existing hydrographic observations and simulated observations mimicking the RAPID-MOC 26 degrees N array in three realizations of the Intergovernmental Panel on Climate Change (IPCC) scenario A1B are projected. For a random observation error of 0.01 kg m(-3), and only using zonal density gradients between 1700- and 3100-m depth, statistically significant detection occurs with 95% reliability after about 30 yr, in the model and climate change scenario analyzed here. Compared to using a single MOC time series as the detection variable, continuous observations of zonal density gradients reduce the detection time by 50%. For the five hydrographic occupations of the 26 degrees N transect, none of the analyzed depth ranges shows a significant trend between 1957 and 2004, implying that there was no MOC trend over the past 50 yr