Notes about calibration

Most optical observations are calibrated using the Johnson-Cousins-Landolt standards.
The calibration transformation usually takes this form:

m=A+k*C-2.5*log(f)          (1)

where m is the magnitude in the Landolt system, A is the calibration zero
point, f is the flux in CCD counts and k*C is a color term(more than one are used sometimes),
which takes into account the differences between the total transmission of the system employed by the
observer (filter+CCD+optics+atmosphere) and the one used by Landolt. It should be kept
in mind therefore, that this expression gives the magnitude that the object will have
if observed with the Landolt transmission function.
Therefore, if we want to generate synthetic colors of the objects we will also need to use the
Landolt transmission function to convolve the corresponding templates. However,
the exact shape of this function is not known, and it may differ considerably from our filter
system, what could lead to big photometric errors for objects with rather extreme
colors, since the linear color term transformation is only an approximation. Thus it is usually
more reasonable to use the transmission functions from our own system, for which the expression (1)
is not valid.
 One solution is to follow the approach of Oke et al.1998, observing a set of spectrophotometric
standards and obtaining the absolute zeropoints of our filters in the AB system. This is a quite a
complex procedure, but luckily there is an approximate way of doing this which yields similar results.
As Oke et al. 1998 proposed we can define a 'natural' system based on our filter set. If we make
this system Vega-based, it will be defined for each filter as

m_n=2.5*log(f_Vega)-2.5*log(f)

Where f_Vega is the counts we will measure from Vega on our CCD through each filter.
Since by definition the magnitude and colors of Vega in the Landolt system is approx 0
(actually the magnitude of Vega on BVRI is approx 0.02), comparing the magnitude of Vega
in our natural system with its magnitude in Landolt system (equation 1) we have 
that 2.5*log(f_Vega)=A.

There our natural Vega system is defined as
m_n \approx A-2.5*log(f)

That is, it has the same zero point that the Landolt transformation but no color terms.
 If we want to make our magnitudes
AB calibrated, we have to add to them the magnitude of Vega on the AB system for that particular filter.

AB_n=A-2.5*log(f)+Vega_AB

Approximate expressions for Vega_AB can be found in Fukugita et al. 1995
or can be calculated using the function flux_sed in the bpz_tools package.
In this way it is straightforward to compute the adequate model colors by convolving the
templates with our filter set and then applying to them the adequate normalization, Vega or AB.