Modeling canopy conductance under contrasting seasonal conditions for a tropical savanna ecosystem of south central Mato Grosso, Brazil

Abstract

Temporal variations in land-atmosphere water vapor exchange are more pronounced in seasonal environments, especially grass-dominated savannas (known as campo sujo Cerrado) of the southern and eastern Amazon Basin. Recent work in campo sujo indicates that rates of canopy conductance (gc) were directly and indirectly affected by a variety of meteorological variables, which limited our understanding of how seasonal variation in meteorology affected rates of gc. Thus, our overall objective here is to determine how individual meteorological variables affect seasonal variations in gc. Estimates of gc were derived using the inverted Penman–Monteith equation from hourly eddy-covariance measurements of evapotranspiration made between March 2011 and December 2013. We employed the so-called Jarvis-type model to estimate gc, which allows us to specify the functional role of each variable on gc, to evaluate how variations in meteorology affect rates of Cerrado gc. The model was parameterized using eddy covariance data from 2012 and tested with data collected in 2011 and 2013. Using orthogonal regression and the Willmott index of agreement, we found that the model estimates compared well to those derived from eddy covariance, especially in 2011. Sensitivity analyses revealed that gc was sensitive to warming and drying, particularly during the dry season when drought stress from low soil moisture availability was already limiting gc. Warming and drying in response to climate change is expected to increase dry season duration, and dry season intensification may already be occurring in central and southern Mato Grosso. Given the high degree of surface-atmosphere coupling for this ecosystem, and the fact that gc is an important link between canopy-scale C and water cycling, our data suggest that dry season intensification may further limit CO2 and H2O vapor exchange during the dry season.