This paper presents a systematic optimization approach to improve the thermal scattering law (TSL) for light water by integrating experimental total cross-section data from the EXFOR database. The methodology is based on a chi-square minimization technique that adjusts the underlying phonon spectrum and discrete oscillator parameters in the LEAPR framework of the NJOY code. A detailed analysis of various perturbed phonon spectra was conducted, leading to the identification of optimized parameters that minimize discrepancies between the calculated and experimental cross section over the energy range of 0.1 to 500 meV.

The optimized TSL evaluation demonstrated improved agreement with experimental data, reducing the mean absolute deviation from approximately 2.99% to 2.77%. Additionally, a preliminary correlation matrix is introduced to assess the interdependencies between different energy bins, offering insights into uncertainty quantification. Furthermore, validation of the optimized TSL against selected International Criticality Safety Benchmark Evaluation Project benchmarks confirms its enhanced predictive capability for reactor physics and criticality safety applications.

The framework developed in this study provides a robust methodology for refining TSL evaluations for light water through a computationally efficient workflow leveraging parallelized GAIA processing and lays the groundwork for extending these optimization strategies to other moderator materials.