Multi-scale covariates as a tool in improving model validity: a case study of a long-term forest regeneration experiment in northern Finland