Local cooling and warming effects of forests based on satellite observations

Abstract

The biophysical effects of forests on climate have been extensively studied with climate models. However, models cannot accurately reproduce local climate effects due to their coarse spatial resolution and uncertainties, and field observations are valuable but often insufficient due to their limited coverage. Here we present new evidence acquired from global satellite data to analyse the biophysical effects of forests on local climate. Results show that tropical forests have a strong cooling effect throughout the year; temperate forests show moderate cooling in summer and moderate warming in winter with net cooling annually; and boreal forests have strong warming in winter and moderate cooling in summer with net warming annually. The spatiotemporal cooling or warming effects are mainly driven by the two competing biophysical effects, evapotranspiration and albedo, which in turn are strongly influenced by rainfall and snow. Implications of our satellite-based study could be useful for informing local forestry policies.

Figure 1. Annual LST difference of forest minus open land.

Effect of forest on temperature is represented by ΔLST (forest minus open land), where positive (negative) values indicate a warming (cooling) effect of forests. (a–c) Spatial pattern (averaged on 1 × 1° grids) and (d–f) corresponding latitudinal dependence of ΔLST for daytime, nighttime, and daily averages (blue line denotes 95% confidence interval (CI) estimated by t-test). Latitude bars with CI out of display range are not drawn. Latitude statistics are in Supplementary Table 1.

Figure 2. Seasonal and latitudinal variations of LST differences of forest minus open land.

(a) Daytime ΔLST. (b) Nighttime ΔLST. (c) Daily average ΔLST. Grids with crosses indicate that the differences are insignificant at 95% by t-test. Latitude statistics are given in Supplementary Table 2.

Figure 3. Albedo and ET differences of forest minus open land.

Seasonal and latitudinal variations of Δalbedo (a,d), daytime ΔET, (b,e) and nighttime ΔET (c,f). Negative Δalbedo shows that forest has lower albedo than open land. Positive ΔET shows that forest has higher ET than open land. Latitude bars with CI out of display range are not drawn. See latitude statistics in Supplementary Tables 3 and 4.

Figure 4. Impacts of biophysical and climate variables on the LST difference of forest minus open land.

(a) Effects of Δalbedo and ΔET on daily ΔLST. (b) Effects of rainfall and snow frequency on daily ΔLST. The regression surfaces in a and b are computed by the least squares method. (c) Relationship between snow frequency and Δalbedo (R=−0.82). (d) Relationship between precipitation and ΔET (R=0.51). Black dots in (c,d) are data samples of Δalbedo or ΔET and red solid lines are the best fit lines.