When analysing HII regions , a possible source of systematic error on empirically derived quantities , as the gas temperature and the chemical composition , is the limited size of the slit used for the observations .
In order to evaluate this kind of systematic error we use the photoionization code Aangaba to create a virtual photoionized region and mimic the effect of a slit observation .
A grid of models was built varying the ionizing radiation spectrum emitted by a central stellar cluster , as well as the gas abundance .
The calculated line surface brightness was then used to simulate slit observations and to derive empirical parameters using the usual methods described in the literature .
Depending on the fraction of the object covered by the slit , the empirically derived physical parameters and chemical composition can be different from those obtained from observations of the whole object .
This effect is mainly dependent on the age of the ionizing stellar cluster .
The low-ionization lines , which originate in the outer layers of the ionized gas , are more sensitive to the size of the area covered by the slit than the high-ionization forbidden lines or recombination lines , since these lines are mainly produced closer to the inner radius of the nebula .
For a slit covering 50 % or less of the total area , the measured [ O III ] , [ O II ] and [ O I ] line intensities are less than 78 % , 62 % and 58 % of the total intensity for young HII region ( t < 3 Myr ) ; for older objects the effect due to the slit is less significant .
Regarding the temperature indicator T _ { [ OIII ] } , the slit effects are small ( usually less than 5 % ) since this temperature is derived from [ OIII ] high-ionization lines .
On the other hand , for the abundance ( and temperature ) indicator R _ { 23 } , which depends also on the [ O II ] line , the slit effect is slightly higher .
Therefore , the systematic error due to slit observations on the O abundance is low , being usually less than 10 % , except for HII regions powered by stellar clusters with a relative low number of ionizing photons between 13.6 and 54.4 eV , which create a smaller O ^ { + + } emitting volume .
In this case , the systematic error on the empirical O abundance deduced from slit observations is more than 10 % when the covered area is less than 50 % .