Supervised classification or crop mapping

Hi,
Can someone please guide me about crop mapping or supervised classification ??
i want to classify cotton, maize, sugarcane, rice and wheat to detect the cropping pattern change during 2014 to 2020 i have training sites (field polygon) of 2020 but i don’t have training sites for past years so how can i classify the crops in past years using one year training sites ??

There is a RUS tutorial on this topic: Crop mapping with Sentinel-2
Maybe radar data is also an option to you, in certain cases it’s much easier: Time-series analysis with Sentinel-1