CO2 should really be a met variable but it generally has a lot of data gaps and cannot be gap-filled using ERAinterim. Need to find a solution