A primary objective of the effort described here is to optimize the performance of a modeling environment for radiometric instruments capable of predicting their complete end-to-end behavior, integrating the optical, electrothermal, and electronic systems. The numerical environment consists of a Monte Carlo ray-trace (MCRT) model of the optical system coupled to a transient three-dimensional finite-difference electrothermal model of the detector assembly with an analytical model of the signal-conditioning circuitry. The resulting model provides a complete simulation of the dynamic optical and electrothermal behavior of the instrument. The modeling environment has been used to create an end-to-end model of the CERES scanning radiometer, and its performance compared to the calibration performance of an operational CERES total channel as a benchmark. To optimize the accuracy of the electrothermal model, the nominal properties of certain key parameters in that model are modified using an evolutionary search algorithm such that the model's simulated output exactly matches the actual instrument ground calibration data. Results indicate that varying the layer thickness, effective thermal conductivity, and effective thermal capacitance of the thermistor, kapton, and epoxy layers in the thermistor bolometer within reasonable uncertainty bounds provides an excellent match with the recorded instrument data.