Dear Chen,
Thank you for forwarding this article, it is an interesting study. However, most of the conclusion is misleading: PVsyst doesn't use the Module temperature in the simulation but rather the cell temperature. The cell temperature is not measured directly in this study (it would require placing temperature sensor inside a module during fabrication).
The difference between module temperature and cell temperature is corrected for in section 3.4 and the coefficients Uc and Uv are recomputed there. These values (table 6) correspond physically to what should be used in PVsyst. The range found is ~[29-32] which is reasonably close to our default recommendation. Note that the authors also mention that the model used to compute cell temperature has not been validated for their specific technology, so the results should be taken with a grain of salt.