Dilemmas in Comparing Observations and Calculations

1
Dilemmas in Comparing Observations and
Calculations of Satellite Radiances (in the
MODIS CO2-slicing algorithm)
Rich Frey, Hong Zhang, Kathy Strabala, and Paul
Menzel January 5, 2005 MODIS Science Team Meeting
Aqua Collection 5 Cloud Top Pressure
(MOD06CT) Uses 8-day clear-sky radiance bias
adjustments - separate zonal means for
ocean, day and night surfaces Uses instrument
bias adjustment - constant in each of
bands 33-36
2
CO2-slicing equation
Rcld(?1) - Rclr(?1) N?(?1)
?pcld ?(?1)dB(?1) dp LHS ----------------------
- ------------------------------- RHS
Error Rcld(?2) - Rclr(?2)
N?(?2) ?pcld ?(?2)dB(?2) dp where
?1and ?2 are MODIS bands 34/33, 35/34, or 36/35
Traditionally, HIRS, GOES products use
observations on LHS RHS is calculated from NCEP
1-degree T(p), q(p), and forward radiance
model. Dilemma 1 MODIS processing restraints
allow only 1 pass through the data how to get
Rclr ? For MODIS, Rclr(?1), Rclr(?2) are
calculated, adding an error term on the LHS. To
mitigate this, we apply clear-sky radiance
biases, as functions of geographic region and
band (Collection 5). Work on the CHAPS
(Collocated HIRS and AVHRR Products) algorithm in
the mid-1990s indicated close agreement between
CTP results when calculated clear-sky radiances
were used with bias corrections.
3
Before Clear-sky Bias Adjustment (through
Collection 4) 1) Absolute amounts of high
clouds were comparable to HIRS values in the
tropics 2) MODIS found 5-15 less high, thin
clouds than HIRS in the mid-latitudes 3)
Inspection of L2 data showed thinner cirrus often
retrieved as middle or low cloud Terra LWIR
bands are noisy (especially band 34), so ability
to retrieve CTPs for high, thin clouds is
somewhat compromised. Clear-sky radiance biases
were attempted, but no positive results were
realized. Aqua LWIR bands are less noisy and
affords an opportunity for good characterization
of global cloud top pressures.
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
Clear-sky Radiance Bias Correction
Accumulate observed minus calculated clear-sky
radiances over eight days observations
based on cloud mask, calculations from forward
model using NCEP gridded data, 25-km
resolution, bands 33-36 Form 1-degree zonal means
of differences (biases) ocean, land day,
land night separately, 5-point running means
day and night land data combined south of 60?
south, ocean ice treated as land
Band 35 Clear-sky Radiances from November 23-30,
2004
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Dilemma 2 Clear-sky bias adjustment
alone yields non-physical
results
Clouds roll up at the edges where cloud is
thinner, some obvious cirrus not retrieved, too
few valid CO2-slicing retrievals Consistent
with clear-sky values being too warm While
correcting a known bias, we discovered a new one
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Evidence Hand-picked observed clear-sky
radiances lead to non-physical results
Bias-adjusted calculated clear-sky radiances
lead to similar non-physical results Even
in cloudy skies, many observed cloudy radiances gt
calculated clear MODIS LWIR (33-36)
observations are significantly warmer than AIRS,
SHIS (D. Tobin, C. Moeller) Solution
Subtract 0.75, 0.50, 0.25 mw/m2str?-1 from
LHS for bands 36-33, respectively
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Dilemma 3 What is the instrument bias
? Dilemma 4 How to correct Terra data?
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
Channel Pair Representing the Best CTP Retrieval
Pink, cyan, and green are 36/35, 35/34, 34/33,
respectively
Before subtracting Instrument Bias
After subtracting Instrument Bias