Cloud parameters

Task

task	description
cloud	3D cloud fields

Older tasks 'cloud-old' and 'cloud-olda' might be still available.

Content versions

The content version is an attribute in hdf files.

content version	description
10	initial number (march 2003)
11	bug corrected: cc was not truncated to [0,1] after matching zoom grids or collecting levels, which lead to strange cco/ccu values (oct 2003)

Fields and methode

file	param	description	unit	levels	from	time res. (hr)	grib code	hor. comb.	vert. comb	Comments
cld	clwc	cloud liquid water content	kg/kg	1:lm	gg	[21,03], ..	246	mass average	mass average	old file 'a'
	ciwc	cloud ice water content	kg/kg	1:lm	gg	[21,03], ..	247	mass average	mass average	old file 'a'
	cc	cloud cover	0-1	1:lm	gg	[21,03], ..	248	mass average	mass average	old file 'a'
	cco	overhead cloud cover	0-1	1:lm	gg	[21,03], ..	249	-	max random overlap; bottom	not a ECMWF field; computed from cc
	ccu	underfeet cloud cover	0-1	1:lm	gg	[21,03], ..	250	-	max random overlap; top	not a ECMWF field; computed from cc

Cloud fields are valid for 6 hour intervals, computed from ECMWF fields valid for the mid of the interval.

Cloud parameters are loaded as a Gaussian grid. Fractions of each gg cell are asigned to TM cells following section 4.9.4; the saved clwc and ciwc correspond to the $\chi$ variables.

The incloud mixing ratio's might be computed from $\kappa=\chi/c$. For small cloud covers, e.g. $c<0.001$, the incloud mixing ratio should be set to zero; the user is encouraged to do deceide on the treshold for himself ...

The overhead and underfeet cloud covers are the total cloud covers above or below a level interface.

output level	implict output level	cco	ccu
	lm+1	above model top; implicit 0.0
lm		above floor of highest layer	below ceiling of highest layer (model top)
:
k		above floor of layer k	below ceiling of layer k
:
1		above floor of lowest layer (surface)	below ceiling of lowest layer
	0		below surface; implicit 0.0

How to interpolate cloud cover

(Peter van Velthoven, 16/11/2002)

How to interpolate cloud cover $c$ (0-1) and cloud liquid water mixing ratio $\chi$ (kg water/kg air) to the TM grid ?

Horizontal interpolation

Consider 2 similar cells 1 and 2 of the base grid (figure 4.9.4). Suppose we want to calculate the cloud cover and liquid water mixing ratio in cell 3. Cell 3 has fractional overlap $f_1$ with cell 1 and $f_2$ with cell 2.

The areas of cell 1 and 2 are $A_1$ and $A_2$, presure gradients are $\Delta p_1$ and $\Delta p_2$, masses are $m_1=A_1\Delta p_1/g$ and $m_2=A_2\Delta p_2/g$.

The amount of liquid water (kg) in cell 1 is $L_1=\chi_1m_1$ (and in cell 2 $L_2=\chi_2m_2$ ).

We want to know the values of cloud cover $c_3$ and liquid water content $\chi_3$ in cell 3 which has a different size.

The area of cell 3 is (indicated by the red lines):

$$A_3 =  f_1 A_1 +  f_2 A_2$$ (4.25)

with mass:

$$m_3 =  f_1 m_1 +  f_2 m_2$$ (4.26)

Fractions $c_1$ of cell 1 and $c_2$ of cell 2 are cloudy (lightblue in figure). The cloudy air mass of cell 1 is thus $c_1m_1$ and that of cell 2 $c_2m_2$. The cloudy air mass of cell 3 is $f_1c_1m_1+f_2c_2m_2$ which by definition equals $c_3m_3$. Combining this with (4.26) gives :

$$c_3 =  \frac{f_1c_1m_1+f_2c_2m_2}{f_1 m_1 +  f_2 m_2}$$ (4.27)

The amount of liquid water in cell 3 is $L_3=f_1L_1+f_2L_2=f_1\chi_1m_1+f_2\chi_2m_2$ which by definition equals $\chi_3m_3$. Combining this with (4.26) gives :

$$\chi_3 =  \frac{f_1\chi_1m_1+f_2\chi_2m_2}{f_1m_1+f_2m_2}$$ (4.28)

All the liquid water is in the cloudy part of the cells. We might also have defined the in-cloud mixing ratio $\kappa=\chi/c$. Since c is less than 1, $\kappa$ is greater than $\chi$. We would expect $\kappa$ to be less variable from cell to cell than $\chi$.

Vertical combination of layers

The situation with regard to cloud cover interpolation in the vertical is quite complicated.

ECMWF assumes maximum random overlap to calculate overhead cloud cover (total cloud cover) from the cloud covers in each layer. See IFS documentation, Chapter 6. Clouds and large-scale precipitation, 6.2.5 Precipation fractions.

When joining 2 adjacent layers in the vertical it depends on what application one wants to use the joined cloud cover for:

1. For photolysis a reasonable approach is to assume maximum overlap for the 2 adjacent layers:
   

   $$c_{joined} =  \max( c_k, c_{k+1} )$$ (4.29)

   
   
   The overhead cloud cover at the base of each layer will be stored.
2. For use in microphysical/chemical calculations the total amount of liquid water in the cell should be conserved, so that it seems better to use mass weighted (pressure weighted) interpolation:
   

   $$c_{joined} =  \frac{m_k c_k + m_{k+1} c_{k+1}}{m_k + m_{k+1}}$$ (4.30)

   
   
   This will also allow us to conserve the relation $\kappa=\chi/c$.

It is not yet clear to me how to join more than 2 adjacent layers according to method 1. When more layers are joined, rather random overlap than maximum overlap should be imposed. Also the assumption of maximum random overlap will be violated when 2 or more layers have been joined, so that it is not possible anymore to calculate the same overhead cloud cover number from the new (joined) layer cloud covers above. That would be my main reason to rather archive the overhead cloud cover itself rather than the interpolated cloud covers per layer.

Currently implemented

1. mass weighted cc, clwc, ciwc
2. compute for each level overhead cloud cover cco from cc using ecmwf scheme (maximum random overlap assumption)
3. compute for each level underfeet cloud cover ccu from cc using maximum random overlap assumption APPLIED TO REVERSED CC COLUMN.

When combining levels, say from 60 to 25:

1. mass weighted cc, clwc, ciwc
2. save bottom cco from combined levels
3. save top ccu from combined levels

Figure 4.1: Example of cloud cover and overhead cloud cover.