Did you know that over 60% continental precipitation is lost through evapotranspiration (ET) globally?  This is why modelling ET has great significance. But, doing it is difficult, particularly considering the large number of models available. This article illustrates how estimating ET is made easy with this R package. The R package enables the use of 17 well-known ET models in a consistent manner.

Firstly, why do we worry about using multiple ET models?

If we have a perfect model that can capture all the physical processes that are relevant to ET, and always produces accurate estimates, then we should never worry about any other models. But as we all noticed (sadly), models are only conceptualisations of the real world. Simplifications of the actual physical processes and assumptions are unavoidable. Furthermore, there are still physical processes that we do not know, or fully understand, which cannot be included in our models. This also applies to the ET models, as no model consists a perfect representation of all the ET processes.

Therefore, we need to use multiple ET models to understand their impact on ET estimation, and to eventually inform our model selection.

If the models are compared with historical climate and ET data, we can then see which model closely resembles the historical relationships between the climate variables and ET. But this is not possible for projecting the future levels of ET, due to the lack of data. In this sense, using multiple ET models is a good way to inform the range of ET projections we can get when using different models.

Why are multiple ET models difficult to implement?

The ET models currently available consist of substantial diversity in aspects including:

  • process representations and assumptions
  • terminology
  • units and data requirements.

Consequently, it is particularly difficult to implement alternative ET models in a consistent way.

As an example, three commonly used ET models are summarised in the figure below, which illustrates varying requirements for input data, as well as different ET quantities produced as output.

Summary of three commonly used ET models, with varying requirements for input data and different ET quantities as output.

How can package Evapotranspiration help?

The open-source R package Evapotranspiration enables ET consistently to be estimated from 17 well-known models.

Data input is flexible, and customised data checking and pre-processing methods are provided. Results are presented as summary text and plots. Comparison of multiple sets of ET estimates can be easily visualized, which can include estimates from alternative models as well as different input data sets. The ET estimates also can be exported for further analysis, and used as input to rainfall-runoff models.

Since it was released in January 2014, the package has been installed over 5,000 times. There has been much positive feedback and many requests via e-mail, which relate to different areas of application. Some of these include hydrology, meteorology and agricultural studies.

Package structure

The schematic figure below summarizes the functions, data inputs and outputs, and graphical features of the package.

The data pre-processing function ReadInputs() is used for loading and processing sub-daily and daily raw climate data.

The processed data are then ready to feed into the generic function ET.… (), where each of the 17 different methods can be called by substituting the ‘…’ by the function name (e.g. ‘ET.Penman ()’ to call the Penman model). The function performs calculations for the relevant ET model,  generates a calculation summary and also saves the full calculation results into a *.csv file.

Having calculated the ET quantity, the function ETPlot() can then be called to plot the original estimates, as well as aggregations and averages at different time scales. Function ETComparison() facilitates comparison of results and visualisation of uncertainties from using different models and/or different input data.

Finally, ETForcing() enables the association between estimated ET and different climate variables to be plotted.

Schematic of package 'Evapotranspiration'

ET models included

17 alternative ET models are included in package Evapotranspiration, which covers PET – potential ET (PET), reference crop ET (ET0) and actual ET (AET).

These include eight PET models, six ET0 models and six AET models (within which three are also capable of estimating PET).

The PET and ET0 models consider different sets of relevant sub-processes, including:

  • incoming radiation
  • vapour gradient
  • heat exchanges with the atmosphere and the ground
  • advection processes
  • aerodynamic resistance
  • surface resistance for vegetated surface.

The six AET models are all based on an observed complementary relationship between PET and AET. To estimate ET, these models use different subsets of a number of climate variables as input, including temperature, solar radiation, relative humidity and wind speed.

Now, let’s now use the package to compare some ET models…

Here is a comparison of the distribution of monthly ET estimates at Adelaide and Alice Springs between 1995-2000, from both the Penman-Monteith model and the Priestley-Taylor model.

For both sites the ET estimates from the Priestley-Taylor model are consistently lower compared with those from the Penman-Monteith model. We can see that the greatest differences between estimates from different models at one site occur at the peak estimates, which is over 100 mm/month at Adelaide. That isapproximately twice the amount of its average rainfall!

Comparison of ET estimates at Adelaide and Alice Springs between 1995-2000, from both the Penman-Monteith and the Priestley-Taylor models.

Further Applications

The output from this package is formatted as time-series-like dataset in R, as well as saved to a *.csv file. Therefore, the output can either be extracted and used as an input to other R packages, or imported to external software for a range of further analyses.

Examples include:

  • Hydrologic modelling (for example with R package hydromad);
  • Investigating the sensitivity of ET estimates to changes in the input climate data (for example with R package sensitivity), which can be used to assess a range of impacts including climate change and data errors.

Related Publications

Package Evapotranspiration is freely available online at CRAN.

For more detailed descriptions and demonstrations of the package please see my recent publication on this package.

We would love to see how this package can help you and having your feedback on it!

 

Recommended Reading

Package Evapotranspiration at CRAN:

https://cran.r-project.org/web/packages/Evapotranspiration/index.html

Guo, D., Westra, S., Maier, H.R., 2016. An R package for modelling actual, potential and reference evapotranspiration. Environmental Modelling & Software, 78: 216-224. DOI:http://dx.doi.org/10.1016/j.envsoft.2015.12.019

 

Dingman, S.L., 2015. Physical Hydrology: Third Edition. Waveland Press.

McMahon, T.A., Peel, M.C., Lowe, L., Srikanthan, R., McVicar, T.R., 2013. Estimating actual, potential, reference crop and pan evaporation using standard meteorological data: a pragmatic synthesis. Hydrology and Earth System Sciences, 17(4): 1331-1363. DOI:10.5194/hess-17-1331-2013

Stocker, T.F. et al., 2013. Climate change 2013: The physical science basis. Intergovernmental Panel on Climate Change, Working Group I Contribution to the IPCC Fifth Assessment Report (AR5),Cambridge University Press, New York.