]\gt ; Measurement of in Ctouds .
By L. F. RICHARDSON .
Communicated by Sir Napier Shaw , F.R.S.\mdash ; Received February 26 , 1919 .
Since the initial data on which a weather prediction is to be based must , to be adequate , include the water-content of the clouds ; and since observations of this are scanty , it will be well to survey what can be done in this respect .
Three types of cloud can be measured : I. Clouds into which an observer can enter .
Several observers , notably Conrad and independently Wagner , have measured the water in clouds on mountains by drawing a measured volume of atmosphere over absorbing substances .
The information is summarised in Hann 's 'Meteorology , ' 3rd edition , p. 306 .
II .
Cloud through which the sun 's outline can be seen and which also exhibit corona , as they often do .
III .
Uniform stratus , provided that some way can be found for measuring the size of the particles .
II .
Thin ( ilouds .
For clouds through which the sun can be seen shining , it is remarkable that the edge of the sun 's disc , if observed through dark spectacles , almost inyariably appears quite sharp , free from any blurring .
I have measured the diminution by clouds of the difference of brightness at the edge of the sun 's disc by a special photometer , arranged as follows and drawn diagrammatically in fig. 1 .
An image of the sun is formed on a piece of stout white paper by means of a lens fitted with an iris diaphragm .
The paper is illuminated from behind by an electric lamp , except for a small area which is rendered opaque and is brought into coincidence with the sun 's image .
Thus the lamp produces a constant difference of brightness , opposite in sign to that produced by the sun 's image .
By varying the iris diaphragm , or by the introduction of absorbing , the difference of brightness , due to the sun , is made to balance that due to the lamp .
The instrument is standardised on the unclouded sun at the same zenith distance .
To eliminate difficulties due to difference of colour between lamp and sun , the paper was observed through a piece of copper-ruby glass .
The instrument could be made convenient and portable .
Stray light scarcely disturbed the balance , provided the enclosure containing the lamp was black , except for its plane white translucent front .
For the interpretation of the readings we may reason as follows : difference of brightness between the sun and the adjoining sky must be due .
L. F. Richardson .
to the light which comes straight through the cloud , and is independent of the general glow of scattered light which is superposed on the disc and its FIG. l.\mdash ; Diagram of Contrast Photometer .
surroundings alike .
The fraction of the incident light which can come straight through must depend only on the size and number of the holes in the sieve , to which we may liken the cloud .
Thus , consider a portion of the beam of light 1 c in cross-section and of a length , , measured along the ray .
Let there be a number , , of particles per cubic centimetre , and let the particles be all of one size , and let A be the area of the diametral plane of a particle supposed spherical .
Then the particles in remove a fraction A of the incident energy from the direct beam , by reflexion , refraction , or absorption .
Diffraction may modify this result in the case of very small particles , but the ordinary cloud particle has a diameter some ten times longer than the wave-length of light , and for the present , diffraction has been neglected .
So if I be the difference of brightness at the sun 's edge in the direct beam at any point -IAndt .
1 ) Measurement of Water in Clouds .
If An be constant , we have on integrating , ( 2 ) where is the length of the ray passing through the cloud , and is the initial differenoe of brightness .
Now , by the geometry of the sphere , An .
( 3 ) The product of the right side into should therefore remain constant if remains fixed .
Now is some sort of measure of the mental sensation of contrast of the object with its surroundings , which will reach a small , more or less fixed , value whejl approaches the " " distance of visibility\ldquo ; of a terrestrial object seen through a mist .
Thus the above equation appears to be in harnJony with the statement of W. Trabert*that the distance of visibility is proportional to the diameter of the particle , for a fixed water-content presumably .
The equation harmonises with Aitken 's observation that the product of the distance of visibility into the number of dust particles per volume , is nearly constant at a fixed humidity , but increases if the particles grow in size by the condensation of vapour upon them .
Our equation finds further support in the observations of Conrad and Wagner , the amount of water in certain clouds , when the data are compared with the diameters of the cloud particles measured by the same observers .
S For the mean of their observations may be summarised in the following formula:\mdash ; of visibility)(volume oater pDiameter oloud particle)volume of cloud ) .
If we mayinsert this number into formulae ( 3 ) and ( 2 ) , it would mean that Conrad and Wagner considered an object just invisible when , the difference of brightness between it and its surroundings , fell to of its value in absence of mist .
This seems a not improbable ratio for a terrestrial object .
Now if the cloud is a horizontal layer it will be convenient to put where is the change of height corresponding to , and is the sun 's zenith distance .
With this substitution equation ( 1 ) becomes - .
( 4 ) * Hann 's ' Meteorology , ' 3rd edit .
, p. 260 .
Hann 's 'Meteorology , ' p. 16 .
HannHann ' , Mr. L. F. Richardson .
On substituting the value of An from ( 3 ) and integrating across the vertical thickness of the cloud , 4 ) yields of air ) .
( 5 ) Now if the particles are all of one size , their diameter may be taken outside the integral in ( 5 ) , which then comes out , no matter in what way the density of the cloud may vary with height , giving If the particles are of water , not ice , this is the same ( equivalent rainfall measured in diameters of a particle ) , which puts the result in its simplest form .
Here I is the contrast of brightness at the sun 's edge when seen through the cloud , and is the contrast for the unclouded sun .
If the cloud is a mixture of particles of different sizes , represented by their diametral areas , A2 , , , and if , , are the respective numbers of these particles per volume of atmosphere , then ( 1 ) becomes , ( 8 ) and the second member of ( 6 ) is made up of a series of terms , the total volume of each size of particles per horizontal area being divided by the diameter and the quotients being added .
It is conceivable that observations of coronae , if sufficiently thorough , might enable us to disentangle the terms due to the different iameters .
But in the present paper it is assumed that the droplets are all of one size .
My observational results are as follows:\mdash ; One noticed , by naked eye , a sensible dimming of the sunlight when , as it is for the thinnest cirrus .
At the other end of the scale , the sun 's disk was just discernible through stratus , when 1/ 10,000 .
Thus , for the clouds cirrus , cirro-stratus , and cirro-cumulus , the average value of the volume of particles per horizontal area appears to be of the order of the radius of the particle .
The of values for the diameter found by Pernter , from observations coronae , was 0 .
cm .
to cm .
* Inserting these , together with the density of ice , we get for the mass of ice per horizontal square centimetre to grm.-c * Hann 's ' Meteorology , ' p. 257 .
Measurement of Water in Clouds .
As a caution , it must again be noted that diffraction ought to be considered before this result is relied upon .
III .
Thick Stratus .
For uniform stratus it is customary to make a very rough eye estimate of the water-content expoessed in such words as " " thin , heavy , dark , ugly , lowering Can photometric measurements yield a corresponding numerical quantity ?
The apparatus which I have used for this purpose was made as follows : A wedge of blotting paper was fixed inside a blackened cigar box , where it could be observed through a small hole coyered with red gelatine .
The two sides of the wedge were illuminated by the light passing through two adjustable iris diaphragms .
The diaphragms were covered by pieces of white paper .
One piece of paper was illuminated by a small electric lamp , which served as an intermediate standard .
The other piece of paper was illuminated either by the unclouded sun , which served as the absolute standard , or alternatively by a small definite solid angle of stratus cloud .
The iris diaphragms were adjusted until the two sides .
the wedge appeared equally bright .
In this way the light from a small solid angle of sky was found .
In calculating the light from the whole sky , there is a trap of which it is well to beware .
The solid angle of a hemisphere is 2 , but the light falling on a horizontal area from a uniformly bright sky is only , not times the light from unit solid angle , on account of obliquity .
For if is the zenith distance , the solid angle between and is ; and the projection of the horizontal area normal to the same Mr. L. F. Richardson .
direction is proportional to .
So the light falling on the horizontal area is proportional to the bright1less of the sky at multiplied by which is equal to , the integral of which from to is This also shows that a correct way to average the obseryations is to plot them as a function of and to find the area under the curve , for this procedure gives proper weight to the different zoneS .
The light on an area of ground.must be equal to the light leaving an equal area of cloud , if the cloud is uniformly bright and if the scattering of the intervening air be neglected .
neglecting scattering , the light on the upper surface of the cloud is proportional to the cosine of the zenith distance , which can be found from the date and the time .
[ Scattering can be allowed for by reference to Tables of L. Kin based on the Smithsonian observations .
] In this way I found that a strato-nimbus cloud of ordinary appearance transmitted a quarter of the red light falling on its upper surface .
A preliminary interpretation of such a result may be developed from Schuster 's theory of " " Radiation through a Foggy Atmosphere .
Following Schuster , let us divide directions into two large groups , in our case by a horizontal plane .
Lt be the sum of all radiation which has a downward component .
the sum of all radiation which has an upward component .
Let , as usual , be an element of length drawn vertically upward .
As a rough allowance for the variety of directions of the rays , we may put the path of the ray as , corresponding to a slope of if is the fraction of the radiation which has its vertical component reyersed in sign by the scattering in unit length of ray , we have { E-E ' } , ( 9 ) { E-E whence ( 11 ) and .
( 12 ) Now , at the lowel ' surface of the cloud , the upward radiation bears to the downward radiation a ratio equal to the reflectivity of the earth , etation , ' Phil. Trans , 1913 .
' Astrophys .
Jour vol. 21 , No. 1 , Jan. , ) .
surement of in Ctouds .
or sea .
Putting for this reflectivity , and letting subscript distinguish values at lower surface of the cloud , we have .
( 13 ) ; .
But , by ( 11 ) , is independent of height , so that in eneral , ( 16 ) .
( 17 ) Consequently equation ( 12 ) becomes , since does not depend on .
( 18 ) Integrating upwards from the lower surface , const .
( 19 ) and the constant of ration must be .
( 20 ) So that at the top of the cloud , denoted by a subscript zero , .
( 21 ) Again , at the top of the cloud the downward component of direct sunlight is , ( 22 ) where the sun 's zenith distance and the intensity for a zenith sun .
Therefore ( 23 ) Now is the fraction of the incident light which is transmitted by the stratus cloud , and is a quantity which can be observed .
Another quantity required for equation ( 23 ) is , the reflectivity of the earth 's surface .
This has been observed by a simple photometer consisting of a white wedge in a black box , the sides of the wedge being illuminated by lights from diametrically opposite directions , admitted to the box through adjustable iris diaphragms , one opening to the earth , the other to uniform stratus .
In this way I found , for directions near the zenith and nadir , the following ratios of brightness of ground , to brightness of stratus , for red light .
Fresh snow , ; dry chalk in Eastern Champagne , France , 0.2 ; sea-sand , damp but not glistening , Seaview , Isle of Wight , rass , ; sea , grey green , between Portsmouth and the Isle of Wjght , ; stubble after hay harvest , ; green oats on chalk , ; loam , damp dark brown , ; dark Mr. L. F. Richardson .
green bushes , .
Except in the cases of snow and water these numbers may be taken as first approximations to B. For snow and water the reflectivity depends notably on direction , and the instrument which has just .
been described does not enable us to deal effectively with that variation , because it contains no reference standard .
Of the quantities in equation ( 23 ) it remains to consider where is the fraction of the radiation which has its vertical component reversed in sign by the scattering in unit length of ray , and where the integration is taken across the vertical thickness of the cloud .
To find we need in the first place the fraction of direct sunlight turned back by a single drop of water .
As previously we have only considered two roups of directions called " " upwards\ldquo ; and " " downwards\ldquo ; for short , so here we only need a division into " " forwards\ldquo ; and " " backwards the division being made by a plane through the drop at right angles to the incident sunlight .
However , it has not been found possible to obtain a convincing answer to the question : what fraction of the light is turned backwards ?
without exploring the distribution of light in detail .
This is not to be retted , for the detail will be required in any future thorough treatment of the problem , in which directions are divided into infinitesimal groups instead of into two hemispheres .
It will be convenient to express the intensity of the scattered radiation by a quantity which depends only on the angle between the incident and the scattered radiation , and which has unity for its mean valu .
Such a quantity is defined by ( 24 ) Now is independent of the unit chosen for candle-power , and therefore we may , if we like , interpret candle-power to mean times the energy per second falling normally on unit area placed at a great distance , , from the drop .
Then , since the area of a sphere of this radius is , it follows that the mean spherical candle-power is equal to the whole energy falling on this large sphere , divided by .
And since the drop is transparent , the total energy scattered on to the large sphere is equal to the total energy striking the drop .
Thus we reach a new expression for , more suited to calculation per second scattered odistance r ( 25 ) ( total energy per second on The distribution of is shown in figs. ( 2 ) and ( 3 ) .
The polar diagram , fig. ( 2 ) , represents a solid of revolution rather like a prolate spheroid with Measurement of Water in Clouds .
FIG. 2.\mdash ; Polar Diagram of Candle-power of a Sphere of Water Illuminated by Parallel Light .
Refractive index .
Diffraction neglected .
The unit is the mean-spherical-candle power .
The line represents the distribution for unpolarized light : the crosses for light polarized so that the magnetic vector is in the plane of incidence , the circles when the electric vector is in the plaIle of incidence .
The polarization is indicated by stating the vector , magnetic or electric , which lies in the plane of incidence .
Refractive index .
Diffraction is neglected .
Mr. L. F. Richardson .
two conical shells attached to one end .
It shows that the scattered light does indeed mainly go forward or backward rather than sideways , thus so far justifying the division of directions into two hemispheres .
The backwardly directed light , which is too small to show on this scale , except at the infinities , is exhibited in .
( 3 ) , in which the abscissa is proportional to area on the sphere at infinity .
If there were a uniform stratus cloud so thin that its droplets did not hide behind each other , then its ought to be proportional to , multiplied by the secant of the zenith distance of the portion of cloud observed .
In this connection it is interesting to note tha the polarisation at from the sun , for a single drop , is such that when the magnetic vector lies in the plane containing the sun , the cloud , and the observer ; and by when the electric vector is in this plane .
Here is proportional to the intensity of the light , and there is of course no relation between the phases of these polarisations , so that the light is not quite plane-polarised .
Now , as to the method by which these diagrams and numbers have been computed .
A ray of light striking a sphere of water at an angle of incidence , , is divided into two , one part reflected , the other entering the drop at an angle of refraction here called The entering ray is further subdivided by repeated impacts on the surface , at each of which the angle between the ray and the normal to the surface is the same , being for the external ray , for the internal .
Now it is known* that if incident light is polarised in the plane of incidence it remains so after reflection or refraction by water .
The same is true for light polarised at right angles to the plane of incidence , but not for polarisations in intermediate planes .
The reflectivity depends upon the polarisation , being , according to Fresnel 's formula respectively , according as the netic or the electric vector is in the plane of incidence .
If then either of these vectors lie in the diametral plane of the sphere at the point where the ray first strikes it , the reflectivity will remain the same at each successive impact , and the polarisation of the emergent portions of the ray will have the same relation to the diametral plane in which the ray lies .
this reason these two polarisations must be computed separately .
The desired result for unpolarised light is obtained at the end of the process by taking the mean .
To simplify the treatment of the infinities which represent the rainbows , it Drude 's 'Optics , ' Chap. II .
Measurement of Water in Ctouds .
is convenient to imagine the sun replaced by an ideal star , infinitely small and bright .
The angle of incidence , , may be used as a measure of position on the water-sphere .
The portion of the sphere included in catches a fraction , , of the whole energy which strikes the sphere .
This is distributed among various rays , deviated from the incident direction through angles .
The area of the portion lying within of a very large sphere of radius is .
Thus , if all the light falling on went to this single portion , then would be equal to .
( 27 ) is the sum of a number of such terms , one for every fraction of the ray , and each multiplied by for every reflection , and by for every refraction which this portion of the ray has undergone , where is the leflectivity corresponding to and to the polarisation .
The deviations , , of the successive fractions are easily shown to be Once reflected Refracted , reflected , refracted Refracted , ' refracted refracted . .
Refracted , twice reflected , refracted The light more than twice internally reflected has been neglected .
This is justifiable , because its intensity would be multiplied by , or by smaller quantities , none of which become appreciable except over a zone which catches very little incident light .
Now , when we attempt to count up areas under the curves in fig. 3 , in order to find the radiation going backwards , we are baffled by the presence of the infinities .
But the difficulty about infinite candle-powers may be entirely avoided by integrations over the sphere-at-infinity by corresponding integrations over the surface of the drop .
Thus , the light turned " " backwards\ldquo ; is made up of\mdash ; ( i ) Light once reflected at angles of incidence between and .
This , expressed as a fraction of the whole light striking the drop , amounts to , ( 29 ) because is proportional to the whole amount of light striking the drop for angles of incidence between and .
The reflectivity , , computed from formulae ( 25 ) and ( 26 ) , runs as follows , for a refractive index of Mr. L. F. Richardson .
On plotting these as a function of , or what is equivalent and more convenient , as a function of , since , it is found that the integral ( 29 ) comes to for the mean of the two polarisations .
That is to say , for unpolarised light ; for the reflectivity is not the ratio of the vectors after and before reflection , but the ratio of their , squares .
( ii ) The primary rainbow and other light once reflected inside the drop .
All this goes " " backwards The intensity is reduced in the ratio at the reflection , and in the ratio at each of the two refractions .
So this parG , as a fraction of the whole light incident on the drop , amounts to .
( 30 ) ( iii ) The secondary rainbow and most of the other light , which is twice reflected inside the drop , but not quite all of it , for one of the equations ( 28 ) shows that is less than when lies between and .
Accordingly , this backwardly directed part , as a fraction of the whole light incident on drop , amounts to .
( 31 ) ( iv ) The tertiary rainbow is directed forward .
The higher rainbows are negligible .
Now adding integrals ( 30 ) and ( 31 ) , we get .
( 32 ) The first integral in ( 32 ) has been evaluated as an area on a diagl.am for the two polarisations separately , with the result that the mean for Measurement of in Clouds .
unpolarised light is .
The second integral in ( 32 ) appears as a small correction , which reduces the total from to Adding to the integral ( 29 ) in amount , it follows that a fraction of the sunlight striking the drop is scattered into the hemisphere which we call the backward direction .
We are now in a position to evaluate , which is required for equation ( 28 ) .
Here is the fraction of the radiation turned backward by the scattering in unit length of ray .
But the whole radiation incident upon drops in a length of ray is equal to the intensity of radiation multiplied by Andl , where A is the area of the diametral plane of a drop , and the number of drops per volume .
Therefore An. Theu taken across a cloud , which , as before , is equal to , ( 33 ) and from ( 23 ) this is equal to , ( 34 ) which gives us what we require .
For instance , for a strato-nimbus from which rain fell on the afternoon of 24 May , 1918 , I found by photometer measurements with red light of wave-length to microns , that Scos was equal to , the sun 's elevation being such that was .
The reflectivity of the grass land was of the order of .
Inserting these numbers in the formulae , we find that the of per horizontal cm.2 of doud to 24 diameters of the doud-droplets .
This figure is greater , as it should be , than the largest number of diameters , namely , four , determined by the ' contrast\ldquo ; method for a stratus which just allowed the sun 's disk to be seen .
Possible developments are the improvement of the above treatment by taking a less crude , but still finite , division of directions in space ; or else by proceeding to infinite subdivision by meaus of Fredholm 's integral equation , * having a kernel involving the function , which is shown in figs. 2 and 3 .
A method for measuring the diameter of the cloud-droplets is also needed .
'Cambridge Tracts in Mathematics , ' No. 10 , by Maxim Bocher .

