Clouds modulate terrestrial carbon uptake in a midlatitude hardwood forest

Abstract

[1] Thirteen years of turbulent exchange and radiation measurements in a midlatitude hardwood forest show that clouds enhance radiation use efficiency of carbon uptake (RUE), and that maximum carbon uptake occurs under moderate cloud cover. We find that both cyclic and secular variability of a simple observable metric of cloudiness (transmittance index) is the best statistical predictor of the interannual variability of both net ecosystem production (NEP) and gross ecosystem production (GEP) seen in our dataset. In contrast other factors analyzed show much weaker relationships with the terrestrial carbon uptake. This suggests that clouds play a pivotal role in driving the interannual variability of terrestrial carbon uptake by this forest and are an important mechanism of carbon cycle/ climate interaction. [2] The interannual variability in the growth rate of atmospheric CO2 is modulated significantly by terrestrial ecosystem processes [Houghton, 2000]. The interaction of climate with regional characteristics of ecosystems imposes complex controlling factors on carbon uptake in different vegetation and climate regimes around the world [Churkina and Running, 1998; Nemani et al., 2003; Barford et al., 2001]. Recent studies suggest that climate change, along with alteration of CO2 fertilization, nitrogen deposition and land-use characteristics, alters the global terrestrial ecosystem through several controlling factors in terrestrial carbon uptake (e.g., temperature [Braswell et al., 1997; Lucht et al., 2000], precipitation [Nemani et al., 2002], and radiation [Nemani et al., 2003; Gu et al., 2003]). However, the presence of clouds can both cause, and be the consequence of, changes in these controls, and subsequent impacts on stomatal dynamics through changes in leaf temperature and leaf-to-air water vapor pressure deficit (WVPD) [Min, 2005]. [3] Total cloud cover has increased about 2% over many midto high-latitude land areas since the beginning of the 20th Century [Intergovernmental Panel on Climate Change (IPCC), 2001]. The increase in total cloud amount, combined with secondary damping effects through soil moisture and precipitation, affects surface temperatures and is negatively correlated with diurnal temperature ranges [Dai et al., 1997]. Clouds also strongly affect the geographic patterns of diurnal temperature range [Dai et al., 1999]. Moreover, changes in precipitation in midand high-latitudes over land have a strong correlation with long-term changes in total cloud amount [IPCC, 2001]. Clouds also modulate solar radiation and photosynthetically-active radiation (PAR) to favour photosynthesis by changing the spectral distribution (light ‘‘spectral quality’’) and diffuse fraction [Min, 2005] and alter stomatal dynamics in fluctuating light environments [Fitzjarrald et al., 1995]. Observational evidence shows that carbon uptake by plants is enhanced on days when the diffuse component of PAR is augmented by clouds or aerosols [Gu et al., 2003; Min, 2005; IPCC, 2001; Hollinger et al., 1994; Price and Black, 1990; Fan et al., 1995; Gu et al., 1999; Freedman et al., 2001; Gu et al., 2002; Niyogi et al., 2004]. [4] Clouds are linked to key factors of climate variability, but their role in the interannual variation of terrestrial CO2 exchange on ecosystem or larger scales has received relatively little attention. Here we examine interannual variability of temperature, precipitation, and cloud distribution and their effects on carbon uptake by analyzing long term turbulent CO2 exchange and radiation measurements from 1992 to 2004 at a northern hardwood forest (Harvard Forest, 42.5N, 72.2W) [Wofsy et al., 1993] (see http://harvardforest. fas.harvard.edu/). The net ecosystem production (NEP) was computed based on the eddy correlation measurements of carbon uptake [Wofsy et al., 1993]. The gross ecosystem production (GEP) was calculated by subtracting respiration (RESP) from NEP, where respiration was measured directly at night and extrapolated for daytime on the basis of daynight changes in soil temperature [Barford et al., 2001; Goulden et al., 1996; Wofsy et al., 1993]. [5] We use the atmospheric transmittance index (TI) as a measure of cloudiness and aerosol loading, as derived from measured surface shortwave radiation after removing dependences of solar zenith angle and solar distance. Radiation impacts due to changes in aerosol and water vapor are relatively smaller than those due to changes in cloud fraction and optical properties. Thus, TI can be viewed as a measure of clouds with the combined effect of both cloud fraction and cloud optical depth, directly linking to the surface cloud forcing defined by Betts and Viterbo [2005]. A smaller TI is produced by a larger cloud/ aerosol optical depth or a greater cloud cover or a combined effect of cloud cover and cloud optical depth in partly cloudy conditions (see auxiliary material).