I got unreasonable first- and second-order O3 sensitivities to anthropogenic NOx emission (O3_ENO, O3_2EN) in CMAQ188.8.131.52 when the simulation reaches the second day. The simulation started from September 20th and turned on the online BVOC emission calculation by MEGAN. Apart from the second-order O3 sensitivities to BVOC (O3-2EB in SENGRID file), the others seemed to be reasonable on September 20. On September 21, O3_ENO and all second-order sensitivities in the SENGRID file showed abnormal values while O3_EVA(AVOC) and O3_EVB were reasonable. The daily anthropogenic emissions are constant, while only natural emissions are changing. No sharp emission changes happened.
Additionally, why does the sensitivity coefficient of O3 to BVOC appear to be negative?
Looking forward to any help.