| 2. |
- Sun, Ge, et al.
(författare)
-
Environmental Controls of Ecosystem Evapotranspiration (ET): Why generalized ET models do not work for forests?
- 2012
-
Ingår i: 2012 AGU Meeting, December 3-7, 2012, San Francisco, CA Advances in the Theory, Modeling, Measurement and Remote Sensing of Evapotranspiration from Terrestrial Surfaces.
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Forests return large amount of fresh water back to the atmosphere through the evapotranspiration (ET) processes, and thus forests have enormous influences on global water, energy, and bigeochemical cycles. Accurately quantifying forest evapotranspiration (ET) is essential to understanding ecohydrological processes, developing regional-scale water and carbon balances, and projecting impacts of environmental changes on natural resources. However, measuring and modeling forest ET remain challenging. Traditional ET models are designed for reference crops (e.g. short green grasses) and not for forested conditions with much higher above and below ground biomass. The large spatial and temporal variability of forests, and complex interactions between physical (temperature and precipitation) and chemical (CO2, Ozone) climate and tree ecophysiological responses also contribute to ET measurement complexity. In this study, we examined environmental controls on ecosystem level ET including forests using multiple measurements (sapflow, watershed hydrologic records, eddy flux measurements, and controlled experiments) and statistical techniques (multivariate linear regressions, best subset regression, stepwise regression). In general, eddy flux data suggest that temperature-based potential ET (PET), measured precipitation (P), and remotely sensed Leaf Area Index (LAI) – a key parameter of ecosystem structure - explain most of the variability of observed monthly forest ET (R2=0.67-0.95) across a wide range of climatic and ecosystem types. VPD is a major driver of forest ET, but P is a rather weak predictor for forest ET. Solar radiation and LAI were highly correlated with ET in grasslands, croplands, and shrublands. Ozone effect was detected only in the mature forests in the Eastern USA where the O3 levels can reach relatively high values. The O3 effect varied with climatic conditions, but the increase in tree transpiration was always associated with reduced streamflow. As the measured ET in forests exceeded Hamon’s PET during the growing season, we conclude that the hydrological models based on these PET estimates are systematically biased low. A series of simple ET models in the form of ET=f(PET, LAI, P) were implemented in a water balance model, WaSSI (Water Supply Stress Index model), for mapping seasonal ET distributions across the continental United States at a scale of approximately 100 km2. We conclude that when eddy flux datasets were extremely useful for understanding ET processes at a fine scale when combined with long term gaged watershed streamflow and remote sensing data. Adding land cover properties into traditional ET models improved model accuracy for simulating seasonal ET. Future models should consider the influences of air pollution, such as CO2, ambient ozone and climate-ozone interactions, on forest ET. Models that include other soil processes such as hydraulic redistribution in the root zones may also improve model accuracy at a large scale.
|
|
