direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

Modelling of evapotranspiration of urban vegetation using thermal and optical remote sensing - a physically-based approach using SCOPE model

Postdoc: Dr. Alby Duarte Rocha

Supervisor: Prof. Dr. Birgit Kleinschmit,


Evapotranspiration (ET) is a measurement of mass or energy exchanged between surfaces and atmosphere, which combines evaporation and plant transpiration (Nouri et al., 2012). ET has a vital role in the water cycle and energy fluxes, regulating precipitation, temperature and vegetation productivity. ET is affected by sunlight intensity (i.e. radiation), air temperature, CO2 concentration (for photosynthesis) and water availability (soil and air).

Fields such as meteorology, hydrology and agriculture have developed different instruments and methods to estimate ET at a local scale (Nouri et al., 2012). However, ET estimations are necessary at a regional and global scale, which often rely on remote sensing data for upscaling (Zhang et al., 2016). ET is challenging to model at broad resolution due to complex ecological processes in different scales. For instance, transpiration occurs at a cellular level and is directly measured at a leaf or plant level, but mainly modelled at the canopy or plot level.

Evapotranspiration does not directly change the spectra (reflectance). Therefore, estimation of ET using exclusively remote sensing data is still not possible, although many input variables for modelling it are currently derived from optical and thermal sensors (Timmermans et al., 2013). Empirical models using spectral indices may be overly simplistic to capture soil-vegetation-atmosphere interactions to predict ET fully accurate (Zhang et al., 2016). Physical and semi-empirical models based on energy balance equations supported by remote sensing data are widely used, but they focus mostly on the atmospheric interface and meteorological data.

The Soil-Canopy-Observation of Photosynthesis and Energy fluxes (SCOPE) model accounts for a wide range of surface-atmosphere interactions as it integrates radiative transfer and energy balance models (Tol et al., 2009). SCOPE has been successfully applied to predict ET (Timmermans et al., 2013, Bayat et al., 2019). However, it assumes a landscape of homogeneous canopies without anthropogenic elements, which is rare in urban areas (Tol et al., 2009). Urban environments are fragmented and present heterogeneous surfaces (vertically and horizontally). The effect of surface heterogeneity in the horizontal direction may not be addressed well with the (1D) model SCOPE (Tol et al., 2009). For mitigating this limitation, the models will be run in fine spatial and temporal resolutions and then aggregated to characterise a typical urban vegetation landscape composed by different species of grasses, shrubs and trees.


This study aims to assess the possibility of a SCOPE model to predict ET accurately in an urban environment. It will also evaluate the contributions of different remote sensing data, such as optical and thermal images from UAV and satellite platform or hyperspectral measurements of leaves and canopies from field spectrometers to predict ET with SCOPE models. This project will explore which of the four ET limiting factors (light, temperature, CO2 and water) is more critical for estimation purely based on remote sensing data.


The study area is a research garden from the Technische Universität Berlin (TUB) located at Steglitz in Berlin. This area represents typical urban backyard vegetation composed by grassland, shrubs and trees, but surrounded by buildings and sealed surfaces (Figure 1). An eddy flux tower is installed, and many data collection initiatives with a comprehensive set of instruments are hosted in the area.

Figure 1: UAV-based multispectral image (left) and a thermal image (right) captured by a field campaign of 2019

The data cover the period from April to September of the years 2019 and 2020. For this period, biochemical and biophysical plant trait measurements were collected for different vegetation types. Thermal and optical remote sensing data captured from different platforms (UAV, field spectrometer and satellites) provide spectra information for canopies and leaves. These data, combined with meteorological, radiative and turbulence fluxes from an eddy flux tower, will be modelled to predict evapotranspiration using SCOPE.

SCOPE is a physical model which integrates radiative transfer models (RTM) of soil, leaf and canopy with energy balance models (Tol et al., 2009). The model simulates soil, leaf and canopy reflectance in optical domain (0.4–2.4 m), and canopy and soil emitted radiance in the thermal domain (2.5–50 m) as a function of plant traits and heat fluxes parameters. Spectra cannot detect variations of latent heat flux directly, but some of the SCOPE parameters used for estimating ET (forward model) can be determined by inverse modelling or vegetation indices using remote sensing data. Many SCOPE parameters such as chlorophyll, carotene and anthocyanins content are strongly correlated. Therefore, the retrieval of biochemical and biophysical parameters using the inverse modelling will be done hierarchically to create a more realistic dependency among the parameters (Figure 2).

Figure 2: Inversion modelling scheme using SCOPE (figure based on ARTMO guidelines)

This study will be constrained by measurements at daytime from spring to autumn and limited for periods without precipitation, as under wet canopy, at night or on winter (temperate climate), ET approach zero (Figure 3 - left). ET varies significantly during the day and along the seasons according to climatological variables (Figure 3 - right). The temporal variability of ET makes hard the alignment of all the essential SCOPE parameters in time and space with remote sensing data (Pacheco-Labrador et al., 2020).

For instance, the time lag of the spectral data and ET (response variable) may vary differently across the year driven by sun zenith angle and the fraction of sunlit and shaded leaves. Water deficient in the high summer season may also affect the ET relationship with the SCOPE parameters in different vegetation types across the city according to the uses of irrigation systems (Nouri et al., 2012). Therefore, soil moisture measurements and atmospheric vapour pressure will be used as model parameters to minimise the drought effect.

Figure 3: Evapotranspiration (mm h-1) variation per hour within a day (left) and cross the days in a season (right) between April and September of 2019

Preliminary results

The global sensitivity analysis (GSA) for SCOPE shows that the most influential parameters to latent heat flux do not significantly affect the spectral signal and vice-versa (figure 4 and 5). The only exception is LAI and solar zenith angle that can affect both ET and spectra. Meteorological parameters are the most important for ET estimation according to the GSA, especially incoming shortwave radiation (Rin), air temperature (Ta) and atmospheric vapour pressure (ea). The range of meteorological parameters used in the GSA was based on eddy tower measurements for the mentioned period.

Ball-Berry stomatal conductance parameter (m), the maximum rate of carboxylation (Vcmax) and the respiration parameter (Rdparam) are related to photosynthetic processes and relatively important for ET, while parameters related to canopy structure (LAI and LIDFa) or pigments and leaf structure (Cab, Cw, Cdm) are important for different spectral regions over the hyperspectral wavelengths or the six bands of the MCA sensor used in the UAV campaign (Figure 4).

Figure 4: Global sensitivity analysis (i.e. normalised total SI) for the most critical SCOPE parameters for a hyperspectral sensor

ET measurements often have many more points over time than space. In SCOPE, values of a time series and pixel from an image are both treated independently, temporally and spatially. However, based on the results for this area, the model will be upscaled for the vegetation surfaces in the entire city using landcover maps for different vegetation classes and climatological data.

Figure 5: Global sensitivity analysis (total SI and first-order SI) for the most critical SCOPE parameters related to latent heat flux



Fieldwork in Steglitz (W3)

Data collaboration and publications based in Steglitz (W1, W3, Postdoc Dr. Basem Aljoumani, MOSAIC)

Eddy heat flux from Climatology department, TU Berlin


Bayat, B., van der Tol, C., Yang, P., & Verhoef, W. (2019). Extending the SCOPE model to combine optical reflectance and soil moisture observations for remote sensing of ecosystem functioning under water stress conditions. Remote sensing of environment, 221, 286-301.

Pacheco-Labrador, J., El-Madany, T. S., Martin, M. P., Gonzalez-Cascon, R., Carrara, A., Moreno, G., ... & Kolle, O. (2020). Combining hyperspectral remote sensing and eddy covariance data streams for estimation of vegetation functional traits. Biogeosciences Discussions.

Nouri, H., Beecham, S., Kazemi, F., & Hassanli, A. M. (2013). A review of ET measurement techniques for estimating the water requirements of urban landscape vegetation. Urban Water Journal, 10(4), 247-259.

Timmermans, J., Su, Z., Van der Tol, C., Verhoef, A., & Verhoef, W. (2013). Quantifying the uncertainty in estimates of surface-atmosphere fluxes through joint evaluation of the SEBS and SCOPE models. Hydrology & Earth System Sciences, 17(4).

Van der Tol, C., Verhoef, W., Timmermans, J., Verhoef, A., & Su, Z. (2009). An integrated model of soil-canopy spectral radiances, photosynthesis, fluorescence, temperature and energy balance. Biogeosciences, 6(12).

Zhang, K., Kimball, J. S., & Running, S. W. (2016). A review of remote sensing based actual evapotranspiration estimation. Wiley Interdisciplinary Reviews: Water, 3(6), 834-853.

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions