Nicolas Gaussiat, Anthony Illingworthand Robin Hogan

Beeskow, 12 Oct 2005

Liquid Water Path from radiometers and lidar.

m sky

m b

T - Tτ = log

T - T

• Radiometers measure brightness temperatures. Tb, that are converted into optical depths, .

• Optical depths are linearly related LWP and VWP :

• kl and kv are path averaged coefficients. d is the ‘dry’ optical depth

• Two frequencies, two equations, two unknowns – find LWP and VWP.

l v dτ = k LWP + k VWP + τ

PROBLEM: Calibration errors, uncertainty over ‘k’ coefficients Cause errors in lwp – it can even go negative.

In clear sky conditions non-zero values of LWP are retrieved.

Some values negative.

l v dτ = k LWP + k VWP + τ

SOLUTION: Add a calibration error, ‘C’ tothe equations.When lidar identifies no water cloud, set LWP = 0,use this to constrain ‘C.

Assuming calibrations errors :

28,

2828,28

22,

2222,22

v

d

v

d

kC

kC

2828,28,28,28

2222,22,22,22

CVWPkLWPkCVWPkLWPk

dvl

dvl

228

222 CCJ

2822,

28,22min C

kk

CJv

v

In clear sky conditions LWP = 0:

Radiometers have same perf : Radiometers have different perf :

where 22 and 28 are the expected standard deviations of respectively C22 and C28.

28

228

22

222

CCJ

282822,

2228,22min C

kk

CJv

v

Principe of the lidar+radiometer technique:

Example :‘C’ factors reset each time no water cloud.

LWP forcedto zero whenno water cloud.

Another Another example:

LWP OFFSET +200 g m-2

- 60g m-2

Sensitivity to drift in T:old technique

Add 5K to Tb (28GHz) and thento Tb(22GHz)

(a) old technique (b) new method

One month’s data: apply 1 to 5K offsets.

Robustness of the new technique :

NEW METHOD:Tb error 5K:introduces only 2% error in LWP

1

5

1

5

LWP error as function of time betweenclear sky events

1hr6min 10hr

Error about 5-10 g m-2

Comparison of three methods

old remove mean lwp new before and after cloud