I agree with Randy that a plausable explanation is increase in denitrifikation. When you have equa denitrifikation as nitrification it is still an decrease in Alk as it is produced 4 H to oxidice one ammonium molecule and only gets 2 OH to reduce same molar/amount nitrate. Thus, it must be att least twice so much denitrifikstion compared to nitrification to get a nettoincrease in alkalinity. As you dose organic carbon i think yoy maybe have an explanation here. In that case it Will stabilize itself because when nitrate go down the reduction of that will decrease over time and the balance between nitrification and denitrifikation Will be back and Alk be more stable.
Randy have to correct me if i am wrong on the exact stochiometri.
Concerning Hanna I believe there method is not sensitive to pH. The reason is (i believe) is that they adding a constant amount acid of that amount so the sample always go below the eqvivalentpoint for HCO3, thus a bit under pH 4.2. Then in that case the pH beyond this slope is not any more sensitive to CO2 as 100% of all carbonates is pushed to H2CO3 and CO2(aq) and as long as the pH is below 4.2 there will be no dissociation at all of H from H2CO3.
So, the point we Will reach on the curve beyond /below 4.2 is only dependent on the alkalinity in the water before we added the constant amount of acid.
I have tried this method of myself and experimentet with different acids and concentrations and it is working.
I tried measure same water with different CO2 levels and got down to excactly the same pH point with same alk. So, i think you can trust Hanna method. There is only one weakness: if the alk is so high so the amount of acid is not enough to reach a point below 4.2. But i guess Hanna has calculated at least that the metre will manage alk from dKH 14 and down maybe.
They by the way use bromocresolgreen with acid added. The light from the diod is 610 nm as the difference in asorbance between the different colorchange for bromocresolgreen in this wavelenght is maximal.