I was wondering what kind of models people use to estimate potential ET for forests? Any reliable models out there? I understand the FAO reference model or Hamon's methods give lower values than actual ET for forests in hot and wet regions.
Thank you for your comments. I used PnET - it is for actual ET and productivity modeling. My question was if any PET model has been used for estimating forest actual ET. I am guessing some sort of calibration to FAO reference ET is needed for peak growing season when ET is highest. Unfortunately, forests are so dynamic. Age, species, LAI, responses to water availability all make the calibration difficult.
There are models such as Shuttleworth PM model, however given the huge variability in water uptake by species, age, phenology, soil moisture conditions etc, a model would have to be calibrated, and there lies the problem, how to get an accurate estimate for tree water uptake. Sapflux is one way, but that has its own set of errors (like assuming uniform crosssectional xylem flux when the measurement is in one side of the trunk, typically). Then there are scaling up issues... logistically one can measure sapflux in upto 16 trees at atime in an area (multiplexer limitation), and then one scales up, unless one has access to a huge amount of equipment. if one has hourly water table data, one can use white's method, but that again has assumptions about lateral flow. Eddy flux is again footprint - dependant, and goes haywire in a tropical canopy. Remote sensing can detect changes in leaf water content (via reflectances), but these again have to be calibrated. One could grow a plant in a pot, upto maybe several meters high, and measure change in waeigh, like a giant lysimeter, but then again scaling up leads to errors...interesting topic this is, especially as ET is such a huge part of the hydrol cycle.
estimating ET for woody vegetation is like the story of seven blind men describing an elephant ! I suppose the choice of model and approacj also depends upon the scale of the basin and the question asked. For instance, if one is looking at a plant water uptake ecophysiology question, greater acuracy is required than if one is computing a water balance of a large catchment.
Appreciate your comments on the difficulty to estimate ecosystem scale forest ET! Forests are not ag crops that are even age and uniform in structure. So, it makes me wondering how wrong it could be if some one uses general PET-based ET models in hydrological modeling (any of the watershed scale models SWAT, MIKESHE, MODFLOW) if one gets the ET wrong. Models models do not calibrate with ET and only streamflow at a watershed outlet is used for calibration/validation.
From a plant ecophysiologist's perspective, hydrological models' handling of ET is potentially laden with high error ! However, at scales larger than a stand, especially when scaling up to a large catchment or river basin, i guess that the errors in all the water balance components also increase, thereby decreasing the relative error of ET with respect to other water inputs and outputs. The error in models' ET estimates culd also depend on the nature of the catchment. For instance, closed catchments like Hubbard Brook watershed have a clear partitioning of precip into runoff and ET, as the underlying bedrock is impermeable to water and the vadose zone is shallow, hence change in soil moisture storage is of a much smaller magnitude than rain, runoff and hence rain-runoff = ET. In this case, at least on a monthly or annual scale, ET estimates would be quite accurate. However in subtropical or tropical catchments with low topo relief, all bets are off, and hydro software could have huge errors in ET estimates. Even though they may have LAI and veg ht, still there are many other plant factors that lead to interspecific variation in transpiration. For instance, increasing salinity in estuaries can shut down transpiration in freshwater lens dependant vegetation for weeks. These trees still have green leaves (higher LAI than drought-stressed leaf drop), hence models would estimate higher ET than there actually is. or there can be species like eucalyptus that transpire much more than other species, while having not that big diff in LAI. In the end, its a matter of scale, but even a 20% error in ET at a basin level can have huge effects on estimation of other components like groundwater recharge, esp in tropical climates. I suppose more work is required to parametrize these models, and calibrate them, thats the hard part.