Structures in FORSE (KP2021) and FORSE+ (this work). FORSE is designed to achieve a 12′ resolution from input large scales at 80′. We have three kinds of newly trained models here: FORSE+S12 and FORSE+S3 to produce stochastic maps at 12′ and 3′, and FORSE+D to generate a deterministic map at 3′. The output, $M_{{12}^{\prime }}^{Q/U,{20}^{°}}$, from FORSE and FORSE+S12 are the input to FORSE+D and FORSE+S3 to get deterministic and stochastic small scales at 3′, respectively.