For an ideal multi spectral remote sensor

there is a completely linear relat ionship between the output of the sensor and the incident radiation due to the even in band spectral response; but for an actual sensor

the output of the sen sor not only depends on the incident radiation

but also on the spectral respon se of the sensor

this will result in the nonlinearity between the output of th e sensor and the incident radiation. Correcting the output data of the sensor b y the use of the spectral shapes of the calibration source and the measured objec t will effectively reduce the measuring error due to the nonlinear response of t he sensor. We have adopted the measured spectral response of Ocean Color CCD Imager （OCCI）

and calculated the response of OCCI to blackbodies at different temperatures which have been normalized to have the same in band average spectr al radiance. It can be seen from the calculated results that the spectral shap es have great influence on the output of OCCI. To reduce the measuring error due t o the nonlinear response

the nonlinear correction factor has been computed acco r ding to the spectral shape of the integrating sphere and the on orbit spectral shape for OCCI ocean measurement

which can fit a 12000 K blackbody.