Impact of biases in gridded weather datasets on biomass estimates of short rotation woody cropping systems

Results suggest that weather variables in all gridded datasets are characterized by some degree of bias leading to considerable bias in biomass estimates, in some cases up to 45%. PRISM and Daymet were shown to have lower uncertainty in most of the weather variables, likely due to their higher spatial resolution and higher dependency on station weather. Site level simulations indicate that relative to the reference biomass estimates based on actual weather measurements, NARR data yielded 4.1 Mg ha−1 y−1 higher biomass while NLDAS, Daymet, and PRISM resulted in 3.3, 1.2 and 0.3 Mg ha−1 y−1 lower biomass. Regional simulations suggest that total biomass varied substantially with gridded data sources ranging between 47.4 and 58.3 Tg on the croplands and rangelands in the region (Columbia Plateau), which subsequently led up to 23% variation in the estimate of poplar based jet fuel production from the SRWC resource. Therefore, findings of this study reinforce the need to account for uncertainties in biomass estimates introduced by biases in gridded weather when modeling bioenergy production.