By merging both datasets, I hope you are referring to KAMIS and agriBORA. If thats the case, IMO the dataset of interest is the agriBORA but you can still use KAMIS for reference in imputing missing data for overlapping window periods. Alternatively,you could use it to expand your training window as KAMIS spans from around 2021 unlike agriBoRA which only goes as far back as 2023.