208 Characterizing Spatial Variability in Precipitation and Throughfall in a Tropical Pre-Montane Cloud Forest

Monday, 7 January 2013
Exhibit Hall 3 (Austin Convention Center)
Nicole C. Shibley, Yale University, New Haven, CT; and N. G. Teale, E. R. Morris, S. M. Quiring, O. W. Frauenfeld, E. B. Roark, and A. D. Rapp

A lack of understanding about spatial variability of precipitation and throughfall in tropical forests motivates this research. Previous research has indicated that throughfall varies based on density of overlying vegetation. In the current study, three hyper-dense networks of wedge-type rain gauges (6 × 6, 2-m spacing) and one extensive network (5 × 5, approximate 10-m spacing) were set up in different regions of the Texas A&M Soltis Center for Research and Education, a pre-montane cloud forest in Costa Rica. One hyper-dense network was installed under no canopy, adjacent to a 10-m weather station. Gauges were measured and emptied approximately every twenty-four hours from June 17–July 18, 2012. Leaf area index and visible sky were estimated at each site using a hemispheric camera. The data indicate that spatial variability in precipitation and throughfall exists, but there are no readily discernible patterns. Over a series of fifteen precipitation events, a single network did not consistently receive more or less throughfall than the other networks. When compared to the control, each network showed statistically significant variance between gauges. The median percentage of throughfall averaged over all events at each hyper-dense site demonstrated a negative linear correlation (R2=76%) with leaf area index, and a positive exponential relationship with clear sky (R2=75%). The same canopy variables over each extensive gauge did not yield a strong correlation. Data were also divided into high and low cumulative amounts and high and low intensity events based on the data from the 10-m weather station. High cumulative events and high intensity produced a greater percentage of throughfall than low cumulative and low intensity events. Spatial variability on each day was compared to meteorological data from the tower data. Few conclusions can be drawn from meteorological conditions, although all precipitation events with a predominant northerly wind showed the maximum spatial variation. Finally, it was determined that 15 randomly-selected gauges from among all 164 can accurately represent the grand mean of all gauges to within 1%. These results demonstrate the complexities of vegetation and spatial variation in characterizing throughfall over a small spatial scale, and motivate more research questions.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner