Test-separator units, as traditional methods of well surveillance, mainly suffer from their inherent constraints including the expensive instrumental, mechanical, electrical, piping and safety devices along with technical and protective inspections, repair and operation services, facilities and infrastructures. Other problems are time and cost consuming, uncertainty of well isolation in test separator and need to close the co-line wells, which are diminished using multivariate thermal well testing. A novel approach was proposed and tested to classify the oil samples taken from individual wells by source and type. The novelties of this work were the use of the applied aspects of (1)H NMR spectroscopy in petroleum upstream engineering, the replacement of traditional test methods, the improvement of the confidence of tests and the recognition of multisource streams. The weighed sum method was used to correlate the spectra information, taken from the samples of Iranian offshore oil wells. The experimental results and the field data revealed that the present approach was appropriate for precocious, quick and reliable surveillance of individual oil wells located in an oil field. The model was supported by field experiments and has predicted the accurate productivity of oil wells with respect to the current expensive techniques since 2010.