1. Estimating reservoir Evaporation: Current practice and state-of-the-art
Justin Huntington,
Associate Research Professor, DRI
Agencies Contributing to This Work
&
Lake Tahoe, NV/CA
Workshop Collaborators: Robert Grossman, Katjia Friedrich, Peter Blanken, Ben Livneh

2. Introduction
Open water evaporation is one of the most difficult surface energy/water fluxes to quantify, and is rarely directly measured in the natural environment
Reservoir operations and the development of new storage and water accounting strategies require estimates of evaporation and net evaporation (E minus PPT)
Changes in open water evaporation under future climate are uncertain, especially if we don’t know current evaporation rates
Boca Reservoir, CA – Truckee River Basin

3. Introduction
Primary factors that govern open water evaporation include
net radiation
heat storage
air temperature
water surface “skin” temperature
humidity
wind speed
stability of the atmosphere
advection of water and heat in and out of the water body
salinity
All of these factors are important to consider when deciding which technique is most appropriate given the application and data requirements
Washoe Lake, NV – Truckee River Basin
Windiest valley in Nevada!!

4. Introduction – Heat Storage
Heat storage can alter both the rate and timing of evaporation depending on the volume, geometry, clarity, and the surrounding environment
For shallow water bodies, the heat storage impact on seasonal evaporation is minor, however for deep water bodies it can be significant
Recent research at Lake Tahoe has found that the peak evaporation is in September-November, rather than summer months
Heat storage causes evaporation to be NEGATIVE in the summer over the Great Lakes!!
Lake Tahoe temperature profile
Figure modified from UC Davis

5. Introduction – Heat Storage
Evaporation is a surface process
Heat storage occurs from penetration of solar radiation beneath the water surface
Energy not readily available at the surface for immediate consumption by evaporation
Heat storage is only available to the surface when transferred by conduction or convection
Large amount of energy is transferred to the surface during fall in winter when Tsurface > Tair
Significant amount of stored energy can be partitioned into heating the air or long-waver emission rather than to evaporation
Important to consider heat storage with any evaporation method..

6. Introduction – Common Indirect Techniques
Common indirect techniques for estimating evaporation include:
Pan evaporation and pan coefficients
Water budget
Energy budget
Bulk mass transfer
Combination of energy and mass transfer techniques
The water budget technique is considered the most accurate indirect approach in arid environments where inflows are minimal and evaporation is a large component of the water budget
Folsom Reservoir, CA - American River
Funded by Reclamation and Cal. DWR

7. Introduction – Common Direct Techniques
The eddy covariance technique is a direct technique, and considered the most accurate if environmental conditions, physical setting, and experimental design is ideal
Eddy Covariance, EC (Enough Corrections!!)
Hard to collect data on shore due to fetch issues
Hard to collect data over water with float/buoy due moving horizontal plane
Subject to energy balance blues…
(Rn – G) = (LE + H)
Energy fluxes = Turbulent fluxes… but never do..
Eddy Covariance Station at American Falls Reservoir, ID - Snake River
Funded by Reclamation

8. Pan Evaporation
Historically, evaporation for operations has been estimated using average pan evaporation data
Pan data are widely known to have significant uncertainty both in magnitude and timing
Freezing conditions limit use of the pans
No heat storage in a pan
Often poorly sited and maintained
Must determine the pan Coef..
Rye Patch Reservoir, Nevada - Humboldt River Basin

9. Typical Operations Table – Truckee River

10. Pan Evaporation Proper pan siting is extremely important
Floating Pans?
Proper pan siting is extremely important
But how do you accurately represent open water and fetch with pans? Floating pans have issues too…
Lake Tahoe Pan and NWS COOP Station

11. Pan Evaporation Example for Tahoe
Old Pan
Example of pan differences due to pan siting
Pier location is double that of the NWS pan station
New Pan.. But discontinued..
Figures from Trask (2007)

12. Energy Balance
Evaporation is estimated as a residual of the water body energy balance
Most widely used in research
Most data intensive and complex approach due the need to consider the entire water body as a control volume rather than just the surface in the case of a land surface energy balance
LE is the latent energy consumed for evaporation
Rn is the net radiation
Qv is the net advected energy to the water body from surface and groundwater inflows and outflows, and direct precipitation (dependent on temperature and amount of flux…linked to water budget!)
Qb is the energy exchange from bottom sediments to the water body,
H is the energy convected and conducted between the air and water as sensible heat
Qx is the energy that is stored in the water body
Qw is the energy advected by evaporating water (dependent on temperature and amount of evaporating water)
Qp is the energy of precipitation
Figure modified from Stannard (2014)

13. Energy Balance
Lake Mead Example
Modified from Moreo and Swancar, 2013

14. Energy Balance Energy Balance - required measurements
Net radiometer to measure net radiation
Air temperature and vapor gradient over the water
Temperature profile to measure heat storage per layer
Bathymetry to estimate volume weighted heat storage per layer
Inflow and outflow volumes and water temperature
Precipitation volume and temperature (usually ignored in arid environments)
Sediment heat storage (usually ignored in deep reservoirs)
Lahontan Reservoir, NV – Carson River Basin
Funded by Reclamation

15. Energy Balance Example - Walker Lake
Walker Lake – USGS energy balance Bowen Ratio station
Walker Lake Bowen Ratio Energy Balance Station, Photo by Allander , 2009 (USGS)
Funded by Reclamation

16. Water Balance Inflows and outflows must all be measured or modeled
Evaporation can be expressed as a residual of the water budget volume or depth per unit time following the continuity equation for a generalized water body as
P is direct precipitation on the water surface
SWin and GWin are surface and groundwater inflows
SWout and GWout are surface and groundwater outflows
B is bank storage
DS is the change in storage
Inflows and outflows must all be measured or modeled
Using modeled inflows, and measured outflows and lake stages can serve as a reality check for other independent methods
Figure modified from Robertson et al. (2003)

17. Aerodynamic / Bulk Mass Transfer
Aerodynamic / Bulk Mass Transfer is probably the most widely and commonly used approach for estimating evaporation
Function of:
surface temperature
humidity
wind speed
atmospheric stability
surface roughness
thermally induced turbulence
barometric pressure
and the density and viscosity of the air
Stampede Reservoir, CA – Truckee River Basin
Funded by Reclamation and DRI

18. Aerodynamic / Bulk Mass Transfer
Dalton’s (1802) general form of the mass transfer equation can be expressed as
E is the evaporation rate,
es is the saturation vapor pressure at the temperature of the water surface
ea is the actual vapor pressure of the air
M is the mass transfer coefficient and is a function of wind speed, atmospheric stability, surface roughness, thermally induced turbulence, barometric pressure, and the density and viscosity of the air

19. Aerodynamic / Bulk Mass Transfer
The direct aerodynamic equation:
The merit of using the aerodynamic method:
Surface temperature can be estimated relatively accurately to estimate qsat
qact and windspeed can be measured and/or estimated fairly easily
Ce can be estimated using Monin-Obukhov similarity theory or computed (calibrated) from evaporation estimates using Energy Balance or Eddy Covariance approaches
Accounts for heat storage by using surface temperature (i.e. seasonal surface temperature variations is the result of heat storage..)
Can be applied at the sub-hourly or daily time steps for near real-time operational monitoring.. Energy balance and water balance can only be applied at longer time steps (weekly, monthly, annual..)
𝐸= 𝜌 𝑎 𝐶 𝑒 𝑢 𝑞 𝑠 − 𝑞 a ,
ra = density of the air
u = windspeed
qs is specific humidity at saturation, estimated from water surface temperature
qa is actual specific humidity, estimated from relative humidity and air temperature measured over water
Ce is the exchange coefficient – a function of atmospheric stability, surface roughness, thermally induced turbulence, barometric pressure, and the density and viscosity of the air

20. Aerodynamic / Bulk Mass Transfer
Pros:
Probably the most cost effective approach for operational monitoring due to low instrument maintenance needs, low instrument cost, and daily time steps
Cons:
Theory is not ideal for reservoirs with limited fetch and located in highly advective environments..
But no other approach works well in these situations either..
Data needs to be collected over water

22. Notable Aerodynamic Study
Harbeck (1962) USGS Study

23. Combination Approaches
Combination energy-aerodynamic mass transfer methods are commonly used for land applications and are typically based on Penman (1948; 1956) and Penman-Monteith formulations
Radiative component:
SW radiation
LW radiation
Ground heat flux
Advective component:
Temperature
Wind Speed
Humidity
λE= Δ Δ+γ 𝑅 𝑛 −𝑄 𝑥 + γ Δ+γ 𝐸 𝐴
Rn = net radiation (shortwave + longwave)
Qx = heat storage of water
λ = latent heat of vaporization
Δ = slope of saturation vapor pressure-temperature curve
γ = psychrometric constant
EA = drying power of air term f (uz)(es – ea))

24. Combination Approaches
Morton modified Penman’s combination approach for practical and operational estimates of evaporation (Morton, 1983)
Morton’s approach is based on Penman’s combination equation with a simple heat storage procedure, and relies on feedbacks between the over passing air and the evaporating surface termed the complementary relationship lake evaporation model (CRLE) (Morton, 1983a,b; Morton et al., 1985; Morton, 1986)
Because CRLE requires limited input data (monthly solar radiation, temperature, humidity) the approach has been fairly popular
Also, the CRLE is not very sensitive to differences in temperature, humidity, and windspeed from land to water, therefore it overcomes shortcomings of the mass transfer method and combination approach, and instead relies on the strength of the Priestly-Taylor available energy approach in which wind speed and vapor pressure are not used

25. CRLE Model

26. CRLE Model

27. CRLE Model

28. Approach Fortran program and well documented
Recently converted to Python by DRI funded by Reclamation

29. Annual Evaluation of CRLE
Evaluated CRLE model for historical periods - compares well to previous estimates of evaporation (water budget, mass transfer, Bowen ratio energy balance)
Average Annual Comparison
- Stannard, 2013
Ratio of Average Annual CRLE Estimated to Research E:
Range =
Average = 1.02
STD = 0.08
WWCRA - Reclamation, 2015

30. Seasonal Evaluation of CRLE
Compared CRLE to USGS bi-weekly energy balance estimates of E at Upper Klamath Lake (Stannard, 2014) funded by Reclamation
WWCRA - Reclamation, 2015

31. Other CRLE Comparison Studies
Most studies found CRLE to be fairly accurate at the annual time step, and less accurate at the seasonal
Very useful for water bodies with limited weather and hydrologic data

32. Limitations of CRLE
Advected heat from inflows and outflows should be considered
Does not account for advection of heat from surrounding water limited environment (negative H)
Does not account for freezing conditions
Heat storage function is based on mean water body depth (area weighted depth) and a simple lag function so seasonal evaporation estimates are more uncertain than annual estimates

33. Seasonal Variation Lake Superior Lake Tahoe
Heat storage is a significant source of uncertainty in most models
Aerodynamic approaches relies on surface temperature… so it must be observed or modeled…
Evaporation can be negative in summer!!
Like condensation on a cold beer glass on a hot humid day
Lake Superior
Lake Tahoe
Trask (2007) – UC Davis Dissertation
Blanken et al. (2015) – Journal of Great Lakes Research

34. Heat storage variations for different lakes in Japan
Lakes in Japan - Kondo (1994)

35. What Model To Use? Depends on the situation and data available…
Don’t pick a highly empirical model that is known to have issues.. be smart about it..
For a model to be good it must have the right ingredients..

36. Reality of Uncertainties – Example for Lake Tahoe
3.6 ft
2.6 ft
Figure from Trask (2007)

37. Uncertainties Uncertainties are important to consider…
Skin temperature and windspeed (aerodynamic)
Heat storage (energy balance)
Inflows outflows (water balance / energy balance)
Solar radiation / Net radiation (combination energy / aerodynamic)
Humidity and temperature (combination / aerodynamic)

38. Future Directions
Combining remote sensing of water surface temperature with gridded weather data and physically based models of evaporation
Model has to have the right type and with the right ingredients to be useful for operations
Model
Samani et al. (2009)

39. Use Satellite Imagery to Track Freezing Conditions
From Huntington and McEvoy (2011)

40. Landsat TIRS Based Open Water Evaporation
Initial tests of Landsat /MODIS TIRS aerodynamic evaporation from Lake Mead compared to USGS eddy flux measurements
Photo by M. Moreo - USGS
Initial results suggest that a TIRS based aerodynamic approach can simulate open water evaporation fairly well, while capturing the lag in evaporation due to the heat storage effect…
Looking into Nov, 2010 and

41. How Good is the Gridded Weather Data for Lakes and Reservoirs??
NLDAS/METDATA (Abatzoglou, 2011) comparison to in-situ eddy station weather data at Lake Mead
Huntington et al. (2015) – in prep.

42. How Good is the Gridded Weather Data for Lakes and Reservoirs??
NLDAS/METDATA (Abatzoglou, 2011) comparison to in-situ buoy station weather data at Lake Tahoe -- Bias Correction / Conditioning??
Huntington et al. (2015) – in prep.

43. Net Evaporation; Net E = E - PPT
Ultimately we need to estimate net evaporation for operations and for predicting lake stage
Precipitation varies significantly across many lakes
We need to measure / model precipitation!!
Quality precipitation measurements or modeling is often forgotten about during our rigorous and expensive flux measurement campaigns..

44. Future Projections of Evaporation
Try to understand the past before we understand the future…
How do we estimate evaporation in the future using a defensible approach (water budget, energy budget, aerodynamic methods)
Water budget requires estimating all future inflows, outflows, and storage changes..
Energy budget requires estimating lots of future variables (net radiation, heat storage, and sensible heat flux… hard ones..), and water inflows and outflows
Aerodynamic requires future surface temperature, windspeed, and humidity

45. Future Projections of Lake Tahoe Evaporation
CRLE forced with 112 climate model projections (many estimated variables)
Lake Tahoe ensemble median and 5th and 95th percentile annual precipitation, temperature, reservoir evaporation, and net evaporation.
WWCRA - Reclamation (2015)

46. Future Projections of Lake Tahoe Evaporation
Lake Tahoe mean monthly ensemble median and 5th and 95th percentile reservoir evaporation and net evaporation.
WWCRA - Reclamation (2015)

47. Need for Representative Data!!
In talking about the lack of data to estimate evaporation Morton (1994) says….
“After all, if physicists can discuss such esoteric subjects as quarks and black holes, why can't the environmental scientists know what is happening right under their noses?”
“The tragedy is that this misapprehension causes the financial resources and scientific skills to be directed toward the development of techniques that can only provide trivial or dubious answers, at the expense of the kind of long-term original research that is needed before worthwhile answers can be expected”

48. Recap Common indirect techniques include:
Pan evaporation and pan coefficients
Proper sitting important
Must select pan coefficients based on lake/reservoir characteristics..
Not really reliable..
Water budget
Inflows outflows and stages must be measured..
Energy budget
Net radiation, heat storage, inflows / ouflows, humidity & temperature gradients must be measured
Aerodynamic
Surface temp, over water wind, and humidity
Combination of energy and mass transfer techniques (all of the above…or use model that estimates.. like CRLE..)
DRI / Reclamation / California Department of Water Resources buoy weather station for energy balance and bulk mass transfer evaporation estimates, Folsom Reservoir, American River, CA.

49. Summary Estimating open water evaporation is not easy!!
True measurements of it are rare..
Important that models be chosen wisely based on environmental conditions and available data
Weather models and gridded weather data have biases and need to be known
Remote sensing of surface temperature can help constrain models or be used directly
Perhaps multiple models should be used… but only those that have “good” ingredients..

50. Thanks – Questions/Suggestions?
Justin Huntington, Desert Research Institute,