The files contained in this directory are the results of the 2017/05/05 cloud simulation run from DAG for the purposes of 
creating a neural network training set for RSP observations during ORACLES. 

THESE FILES ARE INTENDED FOR ER-2 HIGH ALTITUDE OBSERVATIONS

The loop over parameters is as follows:

    sizeb=[0.01, 0.03, 0.05, 0.07, 0.1, 0.15]                        ;6
    n_sizeb=n_elements(sizeb)

    cld_top = [1000.]                                ;1
    n_cld_top = n_elements(cld_top)

    cod = [2.5, 5.0, 10.0, 15.0, 20.0, 30.0]    ;6
    n_cod=n_elements(cod)

    sizea=[5.0, 7.5, 10.0, 12.5, 15.0, 20.0]    ;6
    n_sizea=n_elements(sizea)

    sza=[30.0]                    ;5 to 75 in 5 increments,    sza and azi are now iterated with VEC_INTERP
    n_sza=n_elements(sza)

    azi=[0.0]                    ;0 to 180 in 2 increments,    sza and azi are now iterated with VEC_INTERP
˚    n_azi=n_elements(azi)
                                                    ;total 216


;setup variable to contain all the parameter values, then loop over them to fill ---------
props=make_array(n_sizeb*n_cld_top*n_cod*n_sizea*n_sza*n_azi, 6)

for aa=0,n_sizeb-1 do $
    for bb=0,n_cld_top-1 do $
        for cc=0,n_cod-1 do $
            for dd=0,n_sizea-1 do $
                for ee=0,n_sza-1 do $
                    for ff=0,n_azi-1 do $
                        props[ff+(ee*n_azi)+(dd*n_azi*n_sza)+(cc*n_azi*n_sza*n_sizea)+$
                            (bb*n_azi*n_sza*n_sizea*n_cod)+(aa*n_azi*n_sza*n_sizea*n_cod*n_cld_top),*]=$
                            [sizeb[aa],cld_top[bb],cod[cc],sizea[dd],sza[ee],azi[ff]


-rw-r--r--    1 kknobels kknobels 28822762606 May 12 12:33 NN_20170505_PP_withnoise.tar.gz           	-what I put up originally
-rw-r--r--    1 kknobels kknobels 354172160 Jun 20 12:32 NN_clouds_20170505_PP.nc                     	-no noise, rel azi 0-15, 165-180
-rw-r--r--    1 kknobels kknobels 177086720 Jun 20 12:32 NN_clouds_20170505_PP_cut.nc                  	-no noise, rel azi 0-15
-rw-r--r--    1 kknobels kknobels 752614400 Jun 20 12:32 NN_clouds_20170505_sampled.nc                  -no noise, rel azi 0-180 (not evenly sampled, more close to PP)
-rw-r--r--    1 kknobels kknobels 376307840 Jun 20 12:32 NN_clouds_20170505_sampled_cut.nc         	-no noise, rel azi 0-90 (not evenly sampled, more close to PP)
-rw-r--r--   1 kknobels  staff    569248104 Feb 14 12:08 NN_clouds_20170505_sampled_cut_std.nc		-same as above, but ‘standardized’ so that the mean across
													 cloud types (for each geometry and wavelength) is subtracted from the training set
													 values. The result is divided by the computed measurement uncertainty based on that
													 mean value.