Tangent height offsets estimated by correlation analysis of ground-based data with O3 limb profiles
J.A.E. van GijselY.J. Meijer
Overview
• Introduction– Data description
• Methodology– Visual analysis versus statistical analysis– Example of statistical analysis
• Results– ESA OL 3.0 and IFE 1.63
• Limb ozone profiles• A-priori profiles
Introduction
• Validation of SCIAMACHY limb ozone profiles with collocated ground-based measurements
• Detection and analysis of altitude offset
• Need for objective method to deal with biases
Datasets
• ESA OL 3.0 O3 limb profiles
• IFE 1.63 O3 limb profiles– Both based on ESA level 1 products
• Collocated ground-based observations:– Lidar– Sonde– Microwave (used for IFE 1.63 only)
Datasets II
• ESA OL 3.0: 454 collocated profiles– 332 lidar ozone profiles– 112 sonde ozone profiles
• IFE 1.63: 2346 collocated profiles– 153 unique lidar ozone profiles– 313 unique sonde ozone profiles– 151 unique microwave ozone profiles
Methodology
• Splining of data points to obtain a common altitude grid
• Iterative shifting of the SCIA retrieved limb profiles for the correlation analysis (-5 to +5 km with steps of 200 m)
• Calculation of correlation coefficient between SCIA retrieval and collocated observation over altitude range 20 - 35 km
• Optimal altitude shift can be found at maximum correlation after all iterations
Uncertainties in methodology
• Chosen altitude range can influence results– Reliability of (collocated) instrument varies with
altitude:• ESA OL 3.0 lower ‘trust’ limit set to 20 km• ESA OL 3.0 has a reference height of 40 km where
relative error becomes very high• Sonde data becomes less reliable over 30 – 35 km
• Interpolation over large intervals– For instance microwave data
• Differences in time/space between collocated measurements
Results: ESA OL 3.0 A-priori
• Mean optimal shift is close to 0. • Spread increases towards the poles (as
expected).
ESA OL 3.0 Limb
• Optimal altitude shift = 1.04 km downwards• Standard deviation has decreased with
respect to a-priori
Mean optimal altitude shift using microwave data strongly deviates from lidar & sonde.
Microwave data have a low resolution and the registration of altitude is not very accurate, therefore they will not be further considered
IFE 1.63 A-priori
Lidar
Sonde
Microwave
160 m
-283 m
947 m
IFE 1.63 A-priori II
All West Central-West
Central-East
East
-178 m -79 m -161 m -212 m -273 m
• Dependency on state position due to differences in latitude
IFE 1.63 Limb
• Mean optimal shift (based on lidar+sonde) = 1.17 km (1.04 km for ESA OL 3.0)
• Difference between East and West for a-priori was ±200 m
All West
C.-West
C.-East
East
Lidar -966
-810
-941 -986 -113
2
Sonde -128
1
-113
3
-1174 -1338 -152
2
Microwave
-437
-265
-352 -478 -727
Distances in table in meters
4 profiles within state
-1600
-1400
-1200
-1000
-800
-600
-400
-200
0
Wes
tCen
tral-W
est
Centra
l-Eas
t
East
Opt
imal
alti
tude
shi
ft [m
] A-priori
Lidar
Sonde
Microwave
Linear (Sonde)
Linear (A-priori)
Linear (Lidar)
Linear (Microwave)
Summary
• The ESA OL 3.0 and the IFE 1.63 O3 limb profiles have been validated using ground-based O3 data– Visual analysis
• Not objective
– Statistical analysis• Offsets: ESA OL 3.0: 1.04 Km
IFE 1.63: 1.17 Km
• Ground-based data should be inter-compared to ensure quality