Green Roof Hydrological Modelling With GRU and LSTM Networks

Green Roofs (GRs) are increasing in popularity due to their ability manage roof runoff while providing a number of additional ecosystem services. Improvement hydrological models for the simulation GRs will aid design individual roofs as well city scale planning that relies on predicted impacts widespread GR implementation. Machine learning (ML) has exploded recent years, however there no studies focusing use ML GRs. We focus two types ML-based model: long short-term memory (LSTM) and gated recurrent unit (GRU), modelling performance, with sequence input andsingle output (SISO), synced (SSIO) architectures. Results this paper indicate both LSTM GRU useful tools modelling. As time window length (memory length, step data) increases, SISO appears have higher overall forecast accuracy. SSIO delivers best when is close to, or even exceeds, maximum size.