Reconciling trends from radiosondes and satellites (MSU4 and SSU15x) Data: LKS radiosondes (87 stations), updated to 2004 using IGRA data MSU4 and SSU15x.

Reconciling trends from radiosondes

and satellites (MSU4 and SSU15x)

Data:

• LKS radiosondes (87 stations), updated to 2004 using IGRA data

• MSU4 and SSU15x data (1979-2004)

vertical integrals of radiosonde data

Problem: radiosondeds give substantially larger cooling than satellites

Seidel et al., 2004

radiosonde vs. satellite trends

trends of differences(evaluated at each station)

Trend differences are associated with jumps

jump

Jumps occur at different times at different stations, suggesting problems originate with sondes (not satellites)

Simple solution: just omit stations with large biases (identified by MSU4 comparisons)

trend biases< 0.3 K/decade

this retains ~ 40 stations

Sondes vs. MSU4 for limited LKS stations

Time series: satellite vs. ensemble radiosondes

Time series: satellite vs. ensemble radiosondes

satellite vs. ensemble radiosondes: global means

Radiosonde trends

Radiosonde trends

Seasonality of radiosonde trends spring-summercooling in bothpolar regions

seasonal variation of MSU4 trends

Seasonality of radiosonde trends small seasonality

at all latitudes

vertical profile

of trends

Seasonal variation of polar trends

Solar cycle

Solar cycle

vertical profile of solar signal in tropicsfrom radiosondes

comparisons with SSU15x

note SSU data arenot co-located

with sonde station(10 degree zonal means)

SSU15x trends vs. sondes

SSU15x time series vs. sondes

note trends in differences(SSU15x cooling too much)

SSU15x time series vs. sondes

SSU15x time series vs. sondes

global mean