Geog801 Earth System Sciences/Physical Geography SeminarFall, 2007Project DescriptionIntegrated Modeling of Terrestrial Ecosystem Carbon and Water FluxesIn this semester long project, we will model the carbon and water fluxes for the Eno River watershed from 2001 to 2003. The landscape components can be broadly classified as hardwood, conifer, agricultural/pasture, urban and water. We will use the classification map by Ross Lunetta for this lab. The gross primary production for the vegetated areas will be estimated based on the light-use-efficiency model asAPARGPP , (1)where ε is the actual light-use-efficiency. We will estimate ε from the data measured fromthe eddy-covariance flux tower in the Duke Forest for conifers, hardwoods, and pastures. APAR is the absorbed photosynthetically active radiation. APAR will be estimated based on Beer’s law as))*exp(1(* LKIPARAPAR , (2)where IPAR in the incident PAR, which we will assume constant through the entire watershed, and equal to the measured IPAR at the Duke Forest flux tower. K is light extinction coefficient, which will take the value of 0.46 in this study. L is the leaf area index (LAI). We will use the LAI product from MODIS. However, MODIS product is in 1×1 km spatial resolution. We will have to rescale the LAI product to Landsat cell size based on NDVI and the land-cover map. Essentially, within a 1×1km grid, no LAI will beallocated to water or urban land. LAI from MODIS will then be allocated to Landsat pixels with LAI greater than zero. No LAI will be allocated to pixels with NDVI less thanzero.The above calculations will be carried out on a daily basis. After we have daily GPP fromthe above calculation, we will estimate canopy conductance based on Leuning’s model)()(0DfCAmggsc, (3)where gc is the canopy conductance, g0 is 0.001, and A is GPP, Cs is CO2 concentration at the leaf surface, assuming it is the same as in the air. Γ is CO2 compensation point, which is taken the value of 4.4ppm. f(D)=1/(1+D/D0), where D is vapor pressure deficit in Pa. and D0=1500 pa. A is GPP which is estimated from flux tower measurements at the Duke Forest site. The empirical parameter m will be estimated on a daily basis using Eq (3) andthe measured data from flux tower. The canopy conductance can be solved from: Et=1.56*gc*D, (4)Where Et is measured transpiration, and.Once we calculated gc for every Landsat pixel, we will estimate Et=1.56*gc*D. We will than sum up the Et on each day for the whole year for each pixel, and then sum up the annual total Et for each pixel to get the total Et for the entire watershed. The next BIG question is “How good is the estimated total Et for the watershed?”We will evaluate our estimate of Et from the stream flow record. Assuming that the soil water content is stable on an annual basis, we can estimate ET=P-R, (5)Where ET is the evapotranspiration, including both transpiration (Et) and evaporation from intercepted rain water, impervious surfaces, and water surfaces. P is the total precipitation for the entire watershed. The challenge is to subtract the evaporation from ET. Assuming the following:(1) Water is evaporating at its potential rate as estimated by: Ev=0.033*VPD*u20.76 mm/day Where VPD is in mm mercury, u2 is wind speed at 2 m height in miles/day.(2) Evaporation of Intercepted rainfall is 0.05mm/LAI for any rainfall separated 24 hours apart. Therefore the total intercepted evaporation is 0.05*LAI. If the rainfallis smaller than this. All rainfall is evaporated.(3) Evaporation of intercepted rainfall on the concrete or barren surfaces is 1 mm for any rainfall that is separated 24 hours apart. With the above assumption, we can estimate Et from the stream flow. We will now be able to evaluate Et estimated based on Eq (4).Project schedule1. Download 2001 NLCD from USGS (http://seamless.usgs.gov) and reclassify existing LCLU maps into the following classes: conifers, hardwoods, pasture/ag, barren/urban, water. We will also produce NDVI map from a landsat image collected on May 24, 2002 for the watershed. 2. Download MODIS LAI for the Eno River watershed for the three years, 2001 (Su), 2002 (Ben) and 2003 (Josh). 3. Interpolated MODIS LAI to Landsat spatial resolution through out the year.4. Estimate evaporation for nonvegetated areas.5. Estimate ε each day based on the flux tower measurement.Please refer to Chapter 15 of Campbell and Norman (1998) we used last semester.6. Estimate canopy conductance (gc) and the empirical parameter m in Eq (3) using data from the flux towers.7. Compute A for each Landsat cell using Eq (1)8. Compute gc for each Landsat grid cell using Eq (3).9. Compute Et for each landsat grid cell10. Download stream flow data for Eno River watershed. and calculate ET for the watershed. Subtract evaporation from ET.11. Compare two transpiration estiamtes.12. repeat the same process for another watershed within the Neuse River