Identifying Modes of Temperature Variability Using AIRS Data Alexander Ruzmaikin, Hartmut H. Aumann and Yuk Yung Jet Propulsion Laboratory & California Institute of Technology
Jan 12, 2016
Identifying Modes of Temperature Variability
Using AIRS Data
Alexander Ruzmaikin, Hartmut H. Aumann and Yuk Yung
Jet Propulsion Laboratory & California Institute of Technology
Motivation The canonical global warming at the 100mK/decade rate is largely based on the rise in the ocean surface temperature.
One would expect that temperature trends in the mid-troposphere follow the surface due to convection.
Measuring such temperature trends is a challenge. It requires extremely stable radiometric performance over many years.
An additional challenge is the effect of natural interannual variability.
We use the first five years of Airs data and a new data analysis method to address this challenge.
Data
• Airs: daily zonal means at 2388 1/cm in the CO2 R-branch with weighting function peaking in the mid-troposphere (5 km), clear sky over tropical ocean • AMSU at 57 GHz Oxygen band independent of CO2 at roughly the same altitude for the same data
Empirical Mode Decomposition
j
j
t
i t
jj
i ( )d
jj
1. FFT :
x( t ) a e
2. HHT :
x( t ) a ( t ) e
( )
H( ,t )
ω
ω τ τ
φ ω
ω
⇒
= ⇒
=ℜ
∫ℜ
∑
∑
Huang et al., (1998)
Application of EMD to Airs Data: An Example
Annular Modes
CO2 at Mauna Loa
Linear trend1.695 ± 0.005 ppmv/year
2.001 ± 0.014 ppmv/year in 2002-2007
Sensitivity at 400 hPa is 40 mK/ppmv. Expect 80 mK/year in 2002-2007
CO2 Signal at Airs
The CO2 signal is calculated as
TB(Airs at 2388.2 1/cm) - TB(AMSU5 at 57 GHz)
in 0 - 20°N latitudinal band over tropical ocean
The EMD Modes of CO2 Signal at Airs
Linear trend0.046 ± 0.006 K/year day0.044 ± 0.012 K/year night
1 sigma confidence intervals found by Monte-Carlo simulation with white noise
Comparison of 2 Methods
- 45 ± 9 mK /year -- Airs 2388 1/cm using EMD
-43 ± 7 mK/year (day) -- AIRS 2388 1/cm using method Santer(2001), -50 ± 8 mK/year (night) corrected for autocovariance, ± one sigma
The EMD trend and the anomaly trend agree, but the EMD gives tighter error bars
Observed Trend with AIRS Data
- 45 ± 9 mK /year -- Airs 2388 1/cm data
+10 ± 1 mK /year -- frequency shift (instrumental)------------------------56 ± 10 mK/year -- spectral shift corrected
The observed cooling is due to the effect of increasing CO2, which causes the 2388 1/cm weighting function to shift to higher (colder) altitudes.
The radiometric stability of AIRS for 5 years is better than 8 mK/year.
Trend interpretation
We expected - 80 ± 2 mK/year based on the 2 ppmv/year trend in the CO2 column abundance.
+10 mK /year is expected from sea surface bulk measurements assuming that the mid troposphere follows the moist adiabat
The observed trend can be explained if the temperature at 5 km additionally increased by 24 ± 11 mK/year
------------------------------------------------------------------ 14 ± 11 mK/year discrepancy
Possible explanations:
1. The SST increased faster than 10 mK/year during 2002-2007
2. The mid-troposphere is warming faster than the surface
Conclusions
Five years of Airs data have climate quality and can be used to identify modes of natural variability and temperature trends in the mid-troposphere
There is a possible discrepancy between the expected trend in the mid-troposphere and the observed trend, which may be due to enhanced convection.
This is work in progress and is continuing as more AIRS data become available. The next step will be the extension of this work to lower and higher altitudes.