Time Series Data Missing Value Treatment

I have an hourly time series data for a solar plant which covers 3 years (2019, 2020, 2021). I have a categorical feature named WWCode which has 54 unique values. WWCode is actually a weather condition code. WWCode feature is fully missing for the 2019 except December and there is no missing value at all in other years. I am thinking about how to treat this missing values. I first thought about deleting the feature since it's correlation with the target is very low. But then I started to think about extracting some information from that feature as well. Then after a little research i realized that WWCode is a scale starting from 0 to 99 and the weather condition is worsening as the WWCode is increasing.

https://www.meteopool.org/en/encyclopedia-wmo-ww-wx-code-id2

Now I am working on treating the feature as a numeric one but if someone gives me an idea i would appreciate it indeed.