Exploring the use of astronomical seasons in municipal solid waste disposal rates modeling

Abstract

This study quantifies seasonal variations of municipal solid waste (MSW) disposal rates in a mid-sized Canadian city with respect to both meteorological and astronomical seasons using RNN – LSTM models with a single LSTM layer and 128 units. Meteorological seasons are related to the annual temperature cycle, whereas the astronomical seasons are based on the position of Earth in relation to the Sun. It is hypothesized that the number of hours of natural daylight could also affect MSW generation rates. The use of both meteorological and astronomical seasons in waste modeling is original. During the study period, considerably higher rates in summer (417.8-418.2 tonne/day) than in winter (289.0-290.0 tonne/day) were observed. Waste disposal behaviors in winter are however more consistent, with standard deviations about 138.5-139.8 tonnes/day. The astronomical models appear to better handle extreme situations in the winter and summer. The results indicate that the astronomical models may be more appropriate in some cases. Overall, the predictive accuracy of all models are acceptable, with R2 ranging from 0.70 to 0.86. In general, meteorological models outperformed astronomical models slightly, with RMSE ranged from 72-95 tonne/day. The RMSE of the winter sets are lower, probably due to the lower disposal rates in winter. The results generally support the use of astronomical data to supplement meteorological data in waste seasonal variation studies.