PARENT SESSION
Poster Session 2: Forest Ecology
Monday, August 8, 5:00 PM - 6:30 PM, Exhibit Hall 220 A-E, Level 2, Palais des congrès de Montréal

Scaling of eddy covariance measurements and plot-based surveys for validation of MODIS GPP in a topographically complex forested landscape.

Kim, Eunsook^*,1, Kang, Sinkyu², Kim, Youngil³, Hwang, Taehee⁴, Kim, Joon⁵, ¹ Forest Ecology, Seoul, South Korea² Forest Ecology, Chunchon, Kangwon-do, South Korea³ Forest ecology, Seoul, South Korea⁴ Forest ecology, Chapel Hill, North Carolina, U.S.A⁵ Micrometeorology, Seoul, South Korea

ABSTRACT- MODIS GPP algorithm (MOD17 GPP) provides a useful tool for monitoring seasonal variations of global vegetation primary production. Numerous validation efforts for MOD17 GPP were conducted for diverse geographic regions and biome types by using eddy covariance measurement at fluxtower sites. Our understanding on the reliability, however, has gaps for some geographic regions, especially including topographically complex areas. We tested reliability of MOD17 GPP algorithm to determine its applicability for rugged forested landscapes in Korea. Our study site is a steep rugged watershed (3 km x 3 km) with mixed hardwood and conifer forests. Eddy covariance measurements and plot-scale field survey were implemented and utilized to validate spatially explicit ecohydrological models, which scaled field measurements up to watershed-scale carbon process in this study. We applied two different types of ecohydrological models (RHESSys and BigFoot) to estimate fine-scale GPP distribution within the watershed and then, spatially and temporally aggregate the results comparable to MOD17 GPP for testing its reliability. We, first, validated RHESSys estimation of GPP indirectly by comparing between soil respiration, NEE, and relevant environmental variables estimated and measured, respectively. The RHESSys GPP was, then, utilized to validate BigFoot GPP, which provided final aggregated validation dataset for MOD17 GPP. For fine-scale spatial GPP estimation, Landsat ETM+ image and surface meteorological data with a spatial resolution of 30 m were used as inputs of the models. We found that our RHESSys simulation showed good agreements with the field measurements (e.g. soil respiation, r2=0.84; NEE, r2=0.82; soil water content, r2=0.67, p<0.05). BigFoot GPP also showed good agreements with RHESSys GPP and then, the reliability of MOD17 GPP was tested by using aggregated BigFoot GPP in the topographically complex watershed.

Key words: GPP, MODIS, BigFoot, Eco-hydrological model