EGU2020-20725
https://doi.org/10.5194/egusphere-egu2020-20725
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Evapotranspiration assessments from drone-based thermography - a method comparison in an oil palm plantation and a look ahead

Alexander Röll1, Florian Ellsäßer1, Christian Stiegler2, Tania June3, Hendrayanto Hendrayanto4, Alexander Knohl2, and Dirk Hölscher1
Alexander Röll et al.
  • 1University of Goettingen, Tropical Silviculture and Forest Ecology, Germany
  • 2University of Goettingen, Bioclimatology, Germany
  • 3Bogor Agricultural University, Geophysics and Meteorology, Indonesia
  • 4Bogor Agricultural University, Forest Management, Indonesia

Evapotranspiration (ET) is a key flux in hydrological cycles; it is affected by both climate and land-use change. A recent study across 42 study sites in four land-use types in lowland Sumatra (Indonesia) reported that local and regional transpiration are on the rebound due to the high water use and continuing expansion of oil palm plantations. Conventional ET assessment methods such as satellite-based thermography or the eddy covariance (EC) technique lack the high spatial resolution and spatial replicability, respectively, that are required for ET assessments in dynamic and heterogeneous, mosaic-like landscapes. For such assessments of ET, near-surface airborne thermography offers new opportunities for studies with high numbers of spatial replicates and in a fine spatial resolution. In our study, we tested drone-based thermography and the subsequent application of three energy balance models (DATTUTDUT, TSEB-PT, DTD) using the widely accepted EC technique as a reference method. The study site was a mature oil palm plantation in lowland Sumatra. For 61 flight missions, latent heat flux estimates of the DATTUTDUT model with measured net radiation agreed well with eddy covariance measurements (r²=0.85; MAE=47; RMSE=60) across variable weather conditions and daytimes. Confidence intervals for slope and intercept of a Deming regression suggest no difference between drone-based and eddy covariance method, thus indicating interchangeability. TSEB-PT and DTD yielded agreeable results, but all three models are highly sensitive to the configuration in which net radiation is assessed. Overall, we conclude that drone-based thermography with energy-balance modeling is a reliable method complementing available methods for ET studies. It offers promising, additional opportunities for fine grain and spatially explicit studies. Further steps in the near future will include the testing and if necessary calibrating of the method across different biomes as well as ecological applications.

How to cite: Röll, A., Ellsäßer, F., Stiegler, C., June, T., Hendrayanto, H., Knohl, A., and Hölscher, D.: Evapotranspiration assessments from drone-based thermography - a method comparison in an oil palm plantation and a look ahead, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20725, https://doi.org/10.5194/egusphere-egu2020-20725, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 30 Apr 2020
  • CC1: Comment on EGU2020-20725, Stenka Vulova, 07 May 2020

    Very interesting work! I am also working with UAV data (thermal and multispectral) to characterize ET, but for urban vegetation. How did you implement each algorithm (which coding language?). Do all 3 algorithms require only thermal data as input (e.g. DATTUTDUT, TSEB-PT, DTD)?

    • AC1: Reply to CC1, Florian Ellsäßer, 07 May 2020

      Thanks :) For TSEB-PT and DTD we basically used the pyTSEB implementation of Hector Nieto https://github.com/hectornieto/pyTSEB , for the DATTUTDUT model we wrote or own software that you can just download in QGIS3 as a Plugin called QWaterModel and that comes with an easy to use graphical interface (no coding required!): https://github.com/FloEll/QWaterModel/blob/master/README.md and here the link to the repository: https://plugins.qgis.org/plugins/qwatermodel/

      All of this is implemented in Python 3.

      No, you'll definitely need more input for the TSEB-PT and the DTD model (e.g. meteorological measurements, LAI).  You can technically run DATTUTDUT only from a thermal image, but make sure it is a sunny day without clouds. The DATTUTDUT model computes better results if you have some measured data such as short-wave solar irradiance or net radiation.

      • CC3: Reply to AC1, Stenka Vulova, 07 May 2020

        Thanks a lot; that's really helpful! Thankfully I also have many other measurements in my field site (LAI, meteorological, soil moisture, etc.).

  • CC2: Comment on EGU2020-20725, Stenka Vulova, 07 May 2020

    Also, do you think these methods would also be applicable for urban areas/urban vegetation (which is more heterogeneous)? And what were some challenges of your study?

    • AC2: Reply to CC2, Florian Ellsäßer, 07 May 2020

      I'm not experienced in urban evapotranspiration modelling, but thermal maps of urban environments might contain lots of thermal outliers (hot asphalt, hot car roofs, shiny surfaces that reflect long wave radiation) which will definitly affect the prediction outcome of e.g. the DATTUTDUT model.

      I would still give it a try and maybe mask or remove different areas from the thermal images.

      Good luck, this is going to be a very interesting study!

      • CC4: Reply to AC2, Stenka Vulova, 07 May 2020

        Thank you! :-)