Conference PaperPDF Available

Convective storms - the AVHRR channel 3 cloud top reflectivity as a consequence of internal processes



One of my first papers describing the topic of increased 3.7 micron cloud top reflectivity of convective storms and its possible link to storm internal processes and severity. My first paper in English defining the above anvil plumes (of high 3.7 micron reflectivity).
Report No. 12
Beijing, China
(8-12 May 1989)
Volume l
WMO/TD - No. 269
During the last ten years a number
of methods based on satellite data from
10 - 12.5 pm channels have been developed
for the purpose of estimating the intensity
of various meteorological phenomena. Let's
mention here for instance the criteria for
estimation of precipitation intensity
( e.g. Barrett, 1985, Yamashita et al.,1981),
criteria defining mesoscale convective
complexes (Maddox ,1980 ) , or the description
of the distinct patterns of cloud top
temperature fie]_d of severe storms ( Fujita,
198]_ ). Much less attention has been paid
to the AVHRR channel 3 (3.55 - J.91 pn)
utilization, particulary in daytime hours
when this channel is composed of emitted
thermal radiation as well as of reflected
solar component.
For c}ouds the channel } reflectivity
value depends most of all on the micro-
physical composition of the cloud top.
lalhi}e sma}ler raindrops exhibit a relatively
high channel ] ref]ectivity, }arge raindrops
( over 20pm ) and, especially, ice particles
have a very }ow value of channel J
reflectivity and their properties approach
those of a black body ( Scorer, L987 ).
At the Cz,ech Hydrometeorological
Institute in Prague, a method ( described
in the third part of this paper ) has been
developed for elimination of the thermal
component and for the subsequent ca}cula-
tion of the channel ] reflectivity. This
method is now being used in the convective
storms research and for the detection of
snob/ cover.
Fig.1 shows the spectral density of
therma1 radiation for different temperatures
of the Earth's surface or cloud tops
( tutl lines ) and the spectral density of
ref]ected solar radiation at different
values of the Sun's zenith angle p and at
different values of spectra1 reflectivitya4
( dashed lines ) assumíng a Lambertian
The spectra} density of the emitted
radíation from the Earth's surface or c]oud
jI (c )-4
tops is gíven by the Planck function 8(e,T).
Tn this case, for the sake of simplicity,
the surface is assumed to behave 1ike
a perfect b}ack body.
In order to determine the values of
the spectra1, denslty of the reflected solar
radiation, the spectra1 density of the
solar constant must be first expressed.
This function can be expressed as
s*(x) =rB(i,5800 K) (:)2 (1)
where 13(^,5800 K) is the Planck function
for the effective temperature of the Sun's
photosphere, R is the radius of the Sun
and r is the radius of the Earth's orbit.
Assuming the diffuse reflection mode1
( Lambertian surface ) and neglecting the
absorption by theatmosphere, the spectra1
density of the ref]ected solar radiation
can be expressed as
Ix = o<^B(x,5800K) (l)' ""rl (2)
Figure 1. Intensity of emítted radiation
( fu}l ]ines ) and of reflected
solar radiation ( dashed 1ines)
within the range of the AVHRR
channel J. For details see text.
M. Setvák
Czech Hydrometeorological Institute
Prague, Czechoslovakia
|nW / n2. sr. pm ]
- - -ÉA=
0,1 !=40'
275 K
^10:l_ l = ro'
3.5 3.6 3.1 ].8 3.9 ^
250 K
,225 K
200 K
[p, ]
the equation ( 4 ) can be rewritten as
Tt is obvious from Fig. 1 that at
tenperatures around 2]5 K and at spectral
ref}ectivity of c.^ = 0.1, the two components
of the channel } are approximately equal,
while at the same temperature but at
a spectra1 reflectivityc<,1= 0.01 the emitted
component becornes dominant. 0n the contrary,
at temperatures ranging from 200 to220 K,
the reflected component is severa1 times
greater for Oc^= 0.01 and more than by one
order higher for ua= 0.1 than the emitted
component. This fact wil1 plove to be
important especially when calculating the
channel ] cloud top reflectivlty of convec-
tive storms.
To calculate the channe} } reflectivity
we must first proceed from spectral densi-
ties to integra1 quantíties measured by
NOAA radiometers, i.e. radiances.
The relation between radiance N( T )
measured by the radiometer's sensor in the
given IR channel, temperature T of the
emitting black body, characteristics of the
given sensor described by normafized
fesponse functions ó(v, ), and the P]anck
function B(ý..T) is as Ťo]Iows :
N(T) =;B(ý,,T)ó(y.)aý (1)
i 1, , 1,
where the normalized response functions
ý(v,) are tabulated for dlscrete values
of wáve number /, for individual TR channels
of each of NOAS sate]lites in '|Appendix B
to NOAA Technica1 Memorandum 107". The
relation between the count vafues and the
measured radiances is a relatively simple
calibration procedure described in "NOAA
Technical Memorandum 107 - Data Extraction
and Calibration of T]ROS-N/NOAA Radiometers''.
Denoting €1 as the emissivity related
to AVHRR channe1 } and, similarly,ooz a5
the channe1 ] refl ectivity, the tďtat
radiance measured by the channel } during
daytime can be expressed as
Nr = N|"f * él,Nl(T) )(4)
N] = &l,Sr(r,q) *él.Nl(T) (7)
Note that the method assumes the indepen-
dence ofOC3 and §,. on ý within the range
of channe] 3. If no-w, for sufficiently dense
clouds or for the Earth's surface, a zero
transmissivity is assumed, Kirchhoff law
wil} hold
Ýr+Lr=I (B)
From ( 7 ) and ( 8 ), relations forcomputing
the emissivity and reflectivity ín channel 3
are obtained:
N. - S.(r,() \ ))
N](T) - Sr(r,f) \
N] - N](T) ( lo;
&3= Sr(r,f) - N](T)
The va]ue of N. is obtained from the
calibration relatiďn between counts and
radiances for each pixel, while S,(r,Ě )
is calculated from ( ] ) and ( 6 )' if'the
actual va]ues of r and Ě are known.
A certain simplifícation has to be
made in order to determine the value of
N.(T). To be able to determine it, weneed
td know the temperature T. This can be found
from channel 4 data provided that g, = 1.
Since rea1 values of Ei are always sfiat]er
than l, this step is*a source of certain
error ín the calculations.
what are the inaccuracies of this method ?
a) absorption and dispersion of the
incident, reflected and emitted
radiation by the atmosphere are
b) neglection of emissivity when
determining the temperature from
channel 4
c) effect of water vapour on data
measured by channel 4 is neglected
d) zero transmissivity is assumed
e) diffuse reflection is assumed
Applying the procedure to the cloud
tops of convective storms, and excluding
their fringes which might be partially
transparent, ble ínay disregard inaccuracies
listed under a), c) and d). Limiting our
study to convective storms which occur in
the middle part of the scanned zone and
within a 1irnited range of ]atitudes and
thanks to the heliosynchronous orbit of the
NOAA satellites, there is no need to take
into account the dependence of the reflec-
tivity on the zenith angle, the scan angle
and the azimuth between the vertica1 planes,
invo}ving incident and reflected rays,i.e.
we may admit e).
where N'"' stands f or the ref lected
componenťof the channel 3, and 6;..N,(T)
for the component emitted by a -'body at
temperature T and having the emissivity
of aj.
- The reflected component can be speci-
fied as
,i"' = oL] N](5B00 *) (:.} 2. "o=l ( 5 )
and, denoting
Sr(r,J) = N](5800-' (i)2 "o=Y ( 6 )
The processing of several situations
has revealed that at temperatures ranging
from 200 to 2?5 K, theref]ected component
is at least by one order higher than the
emitted component even ín the worst, cases.
Thus, the error incured by determination
of temperature T from channef 4 and, con*
sequentJ_y, the error of the va}ue of Nl(T) ,
affects the resulting values of €] aňd 0a]
only a litt}e.
Ice particles whích make up the cloud
tops of convective storms should behave in
IR channels almost as a black body, i.e.
thei.r channel J ref}ectivity should approach
Zero. Using a LUT where lower count values
( higher radiances ) are depicted as darker
in comparison with higher count values
(1ower radiances), convective storms should
appear in channel } images as nearly white,
homogeneous areas. However, when a suitable
LUT is used, channel } images sometimes
show convective storms containing darker
parts, or darker as a whole, when compared
with the surrounding convective storms.
By way of comparíson with enhanced
channel 4 images which depict only tempe-
ratures ranging from -40 to -70oC, higher
temperatures can immediately be eliminated
as a reason for observing higher channe} J
radiances in these dark parts. From ( 7 )
it then fo]lows that the explanation should
be sought in the increased channe] ] ref-
The greatest values of the channe] ]
cloud top reflectivity of convective storms,
as determined by our method, amount to 10
to 12% (oc. = 0.10 to 0.1,2), whereas coínmon
convectivď storms show channel ] reflec-
tivity va]ues of one to three per cent.
The cause for such great differences
in channel ] reflectivity must be looked
for in the microphysica} composition of the
cloud tops of convective storms. The
following two explanations are possible :
- The differences in the channel J reflec-
tivity are caused by differences in size
or shapes of the ice particles present
in the cloud tops of convective storms.
- The greater reflectivity might be ascribed
to the presence of supercooled water
droplets. However, since the increased
channel ] reflectivity occurs. also at
temperatures of about -60oC and lower,
this explanation_deesnot.seem to be too
The increased channel 3 reflectivity
has. been observed in the case of ear}y
stages of storm developement as well as in
mature or dissipating stages, therefore it
is not likely that the increased channel ]
reflectivity is related to the storm's
instantenuous activity, ( Liljas, 1987 ).
Neither it seems probable that the increased
channel ] reflectivity is caused by a hígher
concentration of ice partic}es as the
decreased concentratíon wou]d be compensated
by a thicker reflecting }ayer (if the whole
J_ayer composed of these particles is thick
enough). Several situations are archived
which confirme the 1ast statement.
If we reject the differences j.n particle
concentrations and the presence of super-
cooled water drop}ets, as an explanation
for increased channel ] reflectivity remains
only the size and shape of ice particles
present in cloud top. Both size and shape
may inf}uence the 3.7 pm reflectivity,
especially when the size of partlcles
approaches this wave]_ength. At the same
tlme, both the size and shape of the ice
particles present in the cloud top are
affected by the environment and the condi-
tions under which they are formed ( e.g.
the vertica} velocities inside the storm ).
From this point of view, we can therefore
regard channel ] as a source of information
on processes that take place inside storms.
There appears to be a connection
between the increased channel ] reflecti-
vity and the occurence of hai]storms. If
a significant hailstorm is registered
either by a public source or by a ground
meteorological station and íf sate]lite
data are avaj_lable for that period, the
relevant convective storm exhibits an
increased channe} ] ref]ectivity. However,
due to a ]ack of ground based data, this could
not be up to date prooved unambiguously.
In this investigation a certain
drawback stems from the fact that so far
no geostationary satellite images are
available which wouId depict the storm
developement in the 1.7 4Ln channel. This
shortage of data is expected to end l,n L995
when the first new generation Meteosat
should be launched which wiIl provide
irnages also in !h" 3.7 pm range ( at 15
minute intervals ).
0n the contrary, NOAA satellites will
cease transmitting }.74m images during
daytime in the year L992 for which the
rep}acement of this channe} is beíng planned
by a t.77.rm channet ( switched back to ].7|nl
channe}- during the night part of the orbit)
- an unfortunate step in the light of the
above mentioned research.
In the future, this research wi}l be
focused on two areas: modelling of the 3 .7 pm
reflectivity of ice particles havíng
different sizes and/or shapes, and a more
accurate search for the links between ground
and satellite observations of meteorologica1
phenomena. If some kind of connection
between increased channe1 } reflectivity
and occurence of hailstorms is proved ,
it will be possible in the -future to use
the ].7 3lm channe} of Meteosat (or possibly
of other future geostationary satellites )
for a timely detection of hailstorms.
Fig. 2a - NOAA 9 14 Jun l98] BZOW Channel 2
Fig. 2c - NOAA 9 9 JuI l9U Ď55UT Channel 2
Figure 2 Two examples of storms, exhibi-
ting an increased channel } reflectivity.
The storm over Rumania (at the top) shows
a "plumelike" type of an increased channel ]
ref]ectivity, while in the case of storm
Fig. 2b - NOAA 9 14 Jun 1987 Channe] ]
Fig. 2d - NOAA 9 9 Jul 1987 1]55UT Channe1 ]
at Fig. 2c and 2d ( located over centra]
Ita}y) th,ilincreased channe} } reflectivity
is limited to a tower. The size
of the area shown in the images is approx.
250 x250 kn.
Barrett,E.C., 1985: Monitoring the Hydro-
sphere from space, Proc. Remote sensing
Fujita,T.T., ],981: Mesoscale aspects of
convective storms, Proc.Nowcasting,June
1981. Hamburg. ESA SP-165, ] - 10
Liljas,E,, I9B]: Convective Rain as seen
by channel 3, Proc. Satellite and Radar
Imagerv InterpretatJo-ňl--Julv f987l
Reading, 509 - 5I9
Maddox,R.A., 1980:
Complexes, Bull. Mesoscale convective
ications in hydrolooy and water 1314 - 1381
IeSoUrceS Bratis]ava, 60 -
Amer. Meteor. Soc.,61,
Scorer,R.S., 1987: Convective Rain as Seen
by Channel }, Proc. SatelliteandRadar
Imagerv Interpretation. --JuTv- 19EJl
Reading, 509 - 5l9
Yamashita, H, , et a1 , 1981 : Infrared pafame-
ters for nowcasting severe rainstorms,
Proc. |,Iowcasting. June 1981 . Hamburg,
ESA SP-165 , 59 - 6J
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.