Leuning, R, Cleugh, HA, Zegelin, SJ, Hughes, D (2005). Carbon and water fluxes over a temperate Eucalyptus forest and a tropical wet/dry savanna in Australia: measurements and comparison with MODIS remote sensing estimates. AGRICULTURAL AND FOREST METEOROLOGY, 129(4-Mar), 151-173.

Abstract
Measurements over a period of 33 months are presented for the fluxes of carbon, sensible heat and water vapour over a tropical wet/dry savanna in northern Queensland (Virginia Park), and for 38 months over a cool temperate Eucalyptus forest in southeast Australia (Tumbarumba), as part of the OzFlux network. Fluxes were measured using micrometeorological methods and neural network analysis was used to fill gaps in the hourly flux time series. Productivity at Virginia Park is controlled by rainfall amount and timing during the wet season since this determines the leaf area index (L-ai) of the C-4 grass understorey which varies greatly compared to Lai of the trees. The savanna lost 99 g C m(-2) over the period July 2001-March 2004 due to the failure of the 2002-2003 wet season when rainfall was in the lowest 20th percentile. Net uptake of carbon at Tumbarumba was 2510 g C m(-2) from February 2001 to March 2004, with annual productivity ranging from 1060 g C m(-2) year 1 in a normal year to 360 g C m(-2) year(-1) during the 2002-2003 drought which affected all of eastern Australia. Annual rainfall at the savanna was matched by annual evapotranspiration, indicating there was no surface runoff or deep drainage during our study. Trees at both sites were able to extract water deep within the soil profile to maintain annual transpiration above rainfall during the drought of 2002-2003. Measurements of L-ai and gross primary production (P-G) at the two sites were compared to estimates from the Moderate Resolution Imaging Spectroradiometer (MODIS) remote sensing data products. The MOD15 algorithm overestimated L-ai by a factor of two at Tumbarumba but gave reasonable magnitudes and seasonal variation for the savanna. At Tumbarumba, MOD17 gave excellent estimates of the annual amplitude in PG but was less satisfactory in predicting the phase of the variations. MOD17 overestimated PG at Virginia Park during the dry season and the low-rainfall summer of 2002-2003 but gave satisfactory results during the other two wet seasons. Most of the variance in PG at both sites was explained by absorbed photosynthetically active radiation. Significantly improved predictions were obtained for the savanna by modifying the MOD17 algorithm to account for rainfall and potential evaporation in the antecedent 3 months, a surrogate for soil water availability. © 2005 Elsevier B.V. All rights reserved.

DOI:
10.1016/j.agrformet.2004.12.004

ISSN:
0168-1923