Context : Oxygen is the third most abundant element in the universe , but its chemistry in the interstellar medium is still not well understood .

Aims : In order to critically examine the entire oxygen budget , we attempt here initially to estimate the abundance of atomic oxygen , O , in the only one region , where molecular oxygen , O _ { 2 } , has been detected to date .

Methods : We analyse ISOCAM-CVF spectral image data toward \rho { Oph A } Â to derive the temperatures and column densities of H _ { 2 } Â at the locations of ISO-LWS observations of two [ O I ] ^ { 3 } P _ { J } lines .
The intensity ratios of the ( J =1-2 ) 63 \mu m to ( J =0-1 ) 145 \mu m lines largely exceed ten , attesting to the fact that these lines are optically thin .
This is confirmed by radiative transfer calculations , making these lines suitable for abundance determinations .
For that purpose , we calculate line strengths and compare them to the LWS observations .

Results : Excess [ O I ] emission is observed to be associated with the molecular outflow from VLA 1623 .
For this region , we determine the physical parameters , T and N ( H _ { 2 } ) , from the CAM observations and the gas density , n ( H _ { 2 } ) , is determined from the flux ratio of the [ O i ] 63 \mu m and [ O i ] 145 \mu m lines .
For the oxygen abundance , our analysis leads to essentially three possibilities : ( 1 ) Extended low density gas with standard ISM O-abundance , ( 2 ) Compact high density gas with standard ISM O-abundance and ( 3 ) Extended high density gas with reduced oxygen abundance , { [ O / H ] } \sim 2 \times 10 ^ { -5 } .

Conclusions : As option ( 1 ) disregards valid [ O i ] 145 \mu m data , we do not find it very compelling ; we favour option ( 3 ) , as lower abundances are expected as a result of chemical cloud evolution , but we are not able to dismiss option ( 2 ) entirely .
Observations at higher angular resolution than offered by the LWS are required to decide between these possibilities .