time. The R2 for these two periods are higher than those for the whole period of 2000e2012. The R2 of spatial SUR with spatially autocorrelated dependent variables is even over 0.5. All these imply the spatial-temporal patterns of biomass consumption are heterogeneous
over time. As expected, the biomass price has significant negative impact on biomass consumption and the total disposable personal income has significant positive impact on biomass consumption
(Tables 1e6), regardless of the time periods, which is consistent with classical economic theory. Such results prove the correctness of applying the spatial SUR models to investigate the spatialtemporal patterns of biomass consumption. The magnitude of the coefficient of biomass price is decreasing over time during the recession period of 2009e2012 and the period of 2000e2008, indicating the fading impact of biomass price on biomass consumption. This also happens to the total disposable personal income. However, there is relatively small variation in the magnitude of the coefficient of total disposable personal income, and therefore the impact of total disposable personal income on biomass consumption is relatively stable.