Research Project: ENSO Indices For a Changing Climate
Loading...
Date
Authors
Principal Investigators
Department
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract or Project Summary
Many years before a basic dynamical understanding of the El Niño-Southern
Oscillation (ENSO) was available, ENSO was monitored using the Southern
Oscillation Index, which was based on differences of air pressure across the offequatorial Pacific. Later, as the fundamental role of equatorial ocean temperatures was realized, ENSO monitoring and quantification focused on the ocean surface. At present, the most commonly used ENSO indices are based on sea surface temperature (SST) anomalies in fixed regions of the tropical ocean. NOAA’s operational definition of ENSO is based on the Optimal Niño Index (ONI), a running mean of anomalies in the NINO 3.4 region. Many other indices are in use, including those that attempt to quantify different variants of ENSO or those that combine atmospheric and oceanic information.
There are at least two problems with this situation. First, the research community presently does not have a standard definition for the intensity of ENSO. Even a researcher who chooses to use a common index, such as NINO 3.4, must decide on the base period from which anomalies are calculated. NOAA has attempted to keep up with climate change by updating its 30-year reference period every five years, but this is a patch rather than a cure. Researchers working with climate model output have chosen to take anomalies relative to fixed periods, to linearly detrended SSTs, to quadratically detrended SSTs, and so forth.
The second problem is a more fundamental one. ENSO’s atmospheric impact
involves a shift in the location of tropical convection across the tropical Pacific
basin. This shift is driven by a reduction or reversal of the zonal sea surface
temperature (SST) gradient. The atmospheric response is nonlinear, as a large
horizontal displacement of tropical convection is possible once the SST gradient becomes flat or reverses. In a changing climate, the mean zonal SST gradient is likely to change, and so is the magnitude of the NINO 3.4 SST anomaly needed to cause an SST gradient reversal. And in climate models with different tropical SST climatologies, the threshold for SST gradient reversal is model-dependent. A regional anomaly index is only weakly related to the SST variations that control the response of the atmosphere in different climates, be they model climates, paleoclimates, or future climates.
NOAA CPO FY17 COM 1 2 Nielsen-Gammon Statement of Work
This sort of problem is not unique to ENSO. The Atlantic Multidecadal Oscillation is also defined in terms of temperature departures, in this case from a long-term linear trend. While the AMO is strongly correlated to Atlantic hurricane activity, it has been noted that a more physical correlation might be to the difference between tropical Atlantic temperatures and global tropical temperatures. So, in the Atlantic, the scientific community is already moving in the direction of a difference index as a replacement for the AMO index for some applications. (Third problem: relative vs. absolute indices) Motivated by the problems described above, and building on the work of Chiodi and Harrison, we propose to develop an ENSO index based on SST differences in the two tropical regions where convective activity (as measured by outgoing long wave radiation, or OLR) exhibits the largest ENSO-related variability. These regions are the NINO 3.4 region and the Maritime Continent (MC) region. The difference between these two regionally-averaged SSTs (3.4 – MC) reflects the extent to which SST variations act to keep tropical convection confined to the MC or allow it to spread eastward along the equator. Preliminary testing shows that such an index shares the advantage of an OLR-based index in being able to distinguish the dichotomy between weak and strong El Niño events and more precisely indicate extratropical teleconnections, but our SST-based index can also be applied to the pre-satellite era or to any ocean model simulations. It takes advantage of community familiarity with NINO 3.4 and related indices but does not require establishing a reference period from which anomalies should be calculated. The value of such an index is affected by the background modeled climate state or climate change, but in a good way, because it still quantifies the most physically relevant aspect of oceanic drivers of atmospheric response.
Description
Grant