Kipp&Zonen BSRN Scientific Solar Monitoring System - Manual (Page 139)

127
the diffusom eter has been fitted to the pyrheliom eter. Table C 3.4 provides inform ation on the original
length of the diffusom eter arm and the optim al arm -length fitted for the pyrheliom eters used for these
calculations.
The m easures of fit are:
(1) the standard deviation
(2) and the m axim al absolute value of DE.
Diffusometer Fitted to HF Fitted to CH-1 Fitted to NIP Original
Austral CM 885 785 695 795
Germ an CM 790 690 595 687
SciTec CM 630 600 505 577
Canadian CM 940 910 730 750
Austral PSP 885 820 700 795
USA PSP 845 745 580 603
Table C 3.4. Original and “optim al” arm length values (m m ) for existing BSRN diffusom eters.
The second step was the variation of the radii of shading devices. T hese results are given in Table
C 3.5.
Diffusometer Fitted to HF Fitted to CH-1 Fitted to NIP Original
Austral CM 31 34 39 34.8
Germ an CM 27.3 29 34.5 30
SciTec CM 23.7 24 29 25.4
Canadian CM 29.5 31.7 37.5 37.5
Austral PSP 29 33 39 34.8
USA PSP 24.3 27 37.5 37.5
Table C 3.5. Values of original and “optim al” radii (m m ) of shading disk or sphere for the existing
BSRN diffusom eters.
C 3.2.4 Optim al system s
The third step was to look for the arm length and shader radius for each pyranometer – pyrheliometer
pair that gives the best possible fit. Since the full glass dom e m ust be shaded (this is the com m on
sense feeling contrary to the statem ent m ade by Kees van den Bos of Kipp & Zonen at the 1998 Budapest
BSRN W orkshop), therefore the shading disk/sphere m ust be a little larger than the glass dom e. For
three pyranom eters (the exception is the Eppley 8-48) the lower lim it value of the shading disk/sphere
is the optim al one (Table C 3.6).
(1) In the case of optim ized arm lengths or shading disks/spheres (let us say half optim ization)
the bias is zero, the m ean standard deviation is 0.5 W m , the m axim al difference is 1 W m .
-2 -2
(2) For the fully optim ized pairs the m ean (for the 12 pairs) of standard deviations is 0.2 W m , -2
the m axim al difference is 0.5 W m . -2
(3) The NIP fits best to each pyranom eter having one third of the values m entioned in the previous
conclusion. The weakest of the NIP’s fit is to the PSP, this is due to the large dom e size –
sensor size ratio at PSP.
(4) The 8-48 could be fitted best to any pyrheliom eter, since its dom e is only slightly larger than
its sensor.
(5) It is suggested to m ake individual optim ization for each station (solar elevation, aerosol, radiation
and sunshine conditions) taking into account the available instrum entation (pyranom eter,
pyrheliom eter, tracker). This way the best fit could be achieved within the available financial
possibilities and the actual differences can be determ ined for the system to be realized.