To the Editor

In relation to the use of radar backscatter to estimate forest biomass, Woodhouse et al.1 state that our study2 provides an extreme example of a common analytical error, namely fitting a log function instead of a sigmoidal function. In fact, there is almost no difference between these functions, because the slope of the log function is very small in higher biomass ranges.

Furthermore, they criticize us for using the fitted function and not the data to calculate the saturation. The authors make the incorrect statement that the fitted function is used to project sensitivity to aboveground biomass (AGB) values higher than 600 tonnes per hectare (t ha−1). Most studies on radar and AGB estimate the saturation through visual interpretation. In contrast, we calculated the saturation level on the basis of the radiometric accuracy of the radar data and a chosen accuracy level. We used two accuracy levels (50 and 100 t ha−1) and clearly state in the paper that “within the accuracy interval of 50 t ha−1 the estimations are supposed to be accurate whereas estimations within the 100 t ha−1 accuracy interval are only indicators for the spatial AGB distribution.” We found a maximum saturation of 300 t ha−1 at the accuracy level of 50 t ha−1 and emphasized that the 100 t ha−1 saturation level is not accurate enough for a reliable AGB estimation.

Woodhouse et al. contest the wording 'direct measurement'. However, we stated prominently that “no remote sensing can directly measure biomass”. If the whole abstract is read, it is clear that we speak of AGB estimations based on radar, even if we use the term 'direct measurement'.

By using the term 'direct AGB estimation/measurement', we refer to the 'direct remote sensing approach' to estimate AGB, introduced by Goetz et al.3, for which radiometric satellite measurements are calibrated to field-based AGB values. In contrast, the 'indirect AGB estimation' refers to the 'stratify and multiply approach' which links a biomass value determined for a specific vegetation type to a remote-sensing-based land-cover map.

In our opinion, it makes sense to emphasize the difference between these two completely different biomass estimation approaches as they lead to differing results, which we also showed in our paper2.