Fig. 5: Mechanisms of the contrasting runoff responses in warm regions versus cold regions due to global potential forestation.
From: Latitudinal divergence in runoff responses to global forestation due to forest-atmosphere feedbacks

The schematic illustrations and numbers are based on results from IPSL-CM simulations. a In low-latitude warm regions, runoff increases because the positive forest-atmosphere feedbacks (increased precipitation and reduced PET) fully compensates for the negative land surface effects (increased water retention capacity). The substantial precipitation increase mainly results from enhanced vertical moisture transport, amplified by horizontal moisture advection, intensified thermodynamic effects and increased ET. The PET decline primarily results from the reduced VPD, whose effect dominates over the increased surface net radiation. b In high-latitude cold regions, runoff decreases due to the larger negative land surface effects that override the minor positive effects from forest-atmosphere feedbacks (slightly increased precipitation and largely increased PET). The minimal precipitation increase arises from ET-driven additional moisture supply counterbalanced by diminished thermodynamic effects and potentially weakened vertical and horizontal moisture transport. The PET rise primarily stem from the dominant role of increased surface net radiation over the reduced VPD. The direct land surface effects of forestation, which are negative in both warm (a) and cold (b) regions, stems from the increased water retention capacity. This enhanced water retention allows vegetation to allocate precipitation more to ET and less to runoff. This response is most pronounced in areas where PET/P is approximately around 1. The residual term indicates the interactions between forest-atmosphere feedbacks and direct land surface effects. +, positive response to forestation. –, negative response to forestation. The number of + or – denotes the magnitude of the response.?, insignificant response to forestation using Student’s t-test at the 5% significance level.