Kröcher et al. (2026) Monitoring changes in the landscape water balance: validation of satellite- and model-based evapotranspiration data in Lusatia, Germany

Identification

• Journal: Environmental Monitoring and Assessment
• Year: 2026
• Date: 2026-03-06
• Authors: Jenny Kröcher, Gohar Ghazaryan, Uwe Spank, Beate Zimmermann, F Beyrich, Ottfried Dietrich, Christian Markwitz, Justus van Ramshorst, José Ángel Callejas-Rodelas, Gunnar Lischeid
• DOI: 10.1007/s10661-026-15120-8

Research Groups

• Leibniz Centre for Agricultural Landscape Research, Müncheberg, Germany
• Institute of Environmental Sciences and Geography, University of Potsdam, Potsdam, Germany
• Geography Department, Humboldt-University of Berlin, Berlin, Germany
• Chair of Hydrogeology and Hydrochemistry, Technische Universität Bergakademie Freiberg, Freiberg, Germany
• Institute of Hydrology and Meteorology, Dresden University of Technology, Dresden, Germany
• Research Institute of Post-Mining Landscapes, Finsterwalde, Germany
• German Meteorological Service (Deutscher Wetterdienst, DWD), Meteorological Observatory Lindenberg – Richard-Aßmann-Observatory (MOL-RAO), Lindenberg, Germany
• Bioclimatology, Faculty of Forest Sciences and Forest Ecology, University of Göttingen, Göttingen, Germany
• Quanterra Systems Ltd., Exeter, UK

Short Summary

This study systematically validates three satellite- and model-based evapotranspiration (ET) products (CERv2, MODIS, Landsat) against long-term in situ measurements in Lusatia, Germany. It finds that while all products consistently capture spatio-temporal ET patterns, their accuracy in absolute ET values varies significantly, particularly under water-limited conditions.

Objective

• To systematically validate three publicly available satellite- and model-based evapotranspiration (ET) datasets (MODIS, Landsat, and CERv2) against each other and with long-term in situ measurements in an anthropogenically shaped German lowland region.
• To assess how accurately satellite- and model-based datasets capture ET and its spatio-temporal patterns under different land uses.
• To evaluate the extent to which satellite- and model-based data distinguish between the main effects of spatio-temporal variability of evapotranspiration.

Study Configuration

• Spatial Scale: Lower Lusatia, Germany, covering approximately 10,000 km². Datasets used: CERv2 (2 km resolution), MODIS (500 m resolution), Landsat (30 m resolution). In-situ measurements are point/field scale.
• Temporal Scale:
  • CERv2: 1980–2023 (hourly data, aggregated to weekly for analysis).
  • MODIS: 2001–2024 (8-day composite data, aggregated to weekly for analysis).
  • Landsat: 1982–2024 (16-day revisit cycle, aggregated to weekly for analysis).
  • In-situ measurements: 4 to 16 years of data (5 minutes to hourly resolution, aggregated to weekly for analysis).
  • Principal Component Analysis (PCA) performed for 1993-2022 (CERv2, Landsat) and 2001-2022 (MODIS).

Methodology and Data

• Models used:
  • CERv2: Weather Research and Forecasting Model (WRF), version 4.3.3 (physically based atmospheric model simulating ET indirectly from coupled energy and water balance processes).
  • MODIS (MOD16AGF Version 6.1): Biophysical energy balance model based on the Penman–Monteith framework.
  • Landsat (Collection 2 Provisional Actual Evapotranspiration Science Product): Simplified Energy Balance Operational (SSEBop) model, using land surface temperature as a proxy for ET.
• Data sources:
  • Satellite-based ET products: CERv2, MODIS Terra Net Evapotranspiration (MOD16), Landsat Provisional Actual Evapotranspiration Science Product.
  • Reanalysis data: ERA5 (for CERv2), NASA’s Global Modelling and Assimilation Office (GMAO) meteorological reanalyses (for MODIS).
  • In-situ measurements: Seven sites in Lusatia, Germany, utilizing weighable lysimeters and eddy covariance (EC) stations for direct ET measurements.
  • Ancillary data: CORINE Land Cover/Land use data (100 m spatial resolution), soil moisture time series.

Main Results

• Accuracy of ET Estimates:
  • Lowest relative deviations from measured ET were found at grassland sites (Median relative deviation: 5–54%; Root Mean Square Difference (RMSD): 0.58–1.02 mm d⁻¹).
  • MOD16 consistently showed the lowest deviations across nearly all sites (Median relative deviation: 10–54%; RMSD: 0.58–0.90 mm d⁻¹).
  • All satellite- and model-based datasets tended to overestimate ET during water-limited drought conditions, particularly in pine forests and agricultural sites during summer.
  • At the Spreewald grassland site, characterized by unlimited water supply, all products showed the best agreement with in-situ measurements.
  • High deviations were observed for the Bautzen reservoir site, indicating poor performance of land surface models for complex water body evaporation.
• Spatio-temporal ET Patterns (Principal Component Analysis - PCA):
  • The first three principal components (PCs) explained approximately 98% (CERv2), 83% (MODIS), and 87% (Landsat) of the total variance, indicating consistent capture of dominant ET patterns.
  • PC1 captured the regionally coherent seasonal ET cycle (high in summer, low in winter) across all datasets.
  • PC2 reflected spatial patterns correlated with land use (e.g., forest/water vs. arable land/opencast mines) and the influence of soil water availability on ET dynamics.
  • PC3 represented long-term changes in ET driven by land use changes, such as increasing ET trends in areas of natural succession/afforestation and decreasing trends in active mining areas.
  • CERv2 showed limited ability to capture the spatial patterns of PC3 compared to MODIS and Landsat, likely due to its coarser spatial resolution.
• Scale Effects: Spatial aggregation of ET products introduced minimal systematic bias at the regional scale (mean difference close to zero) but increased the dispersion of ET estimates, particularly in heterogeneous land-use classes (e.g., water bodies, sparsely vegetated areas). Landsat aggregation from 30 m to 2 km resulted in standard deviations up to 47 mm yr⁻¹, while MODIS aggregation from 500 m to 2 km showed lower deviations (6–7 mm yr⁻¹).

Contributions

• Presents a systematic multi-product validation of three publicly available satellite- and model-based ET products (CERv2, MODIS, Landsat), including two recently developed products (CERv2 and Landsat Provisional Actual ET) that had not been comprehensively evaluated against long-term in situ measurements in an anthropogenically shaped German lowland region.
• Evaluates the reliability of ET products in representing spatial and temporal dynamics of ET using Principal Component Analysis (PCA), providing insights into their suitability for regional water management and hydrological applications.
• Offers new insights into the strengths and limitations of widely used ET datasets for monitoring landscape water balance dynamics in heterogeneous and human-impacted regions.
• Provides a robust foundation for the targeted application of these ET datasets in hydrological modeling, water resources management, and land use planning, emphasizing the necessity to consider dataset-specific limitations, particularly the systematic overestimation of ET under water-limited conditions.

Funding

• BMFTR Land-Innovation-Lausitz project ‘Monitoring der Landschaftsraumoberflächen und Biodiversität mittels Drohnen-und Satellitendaten in der Modellregion Lausitz’ (Grant number 03WIR3020B).
• German Science Foundation (Deutsche Forschungsgemeinschaft, DFG) projects:
  • “Greenhouse Gas Emissions from Reservoirs: Mechanisms and Quantification – TregaTa” (project number: 288267759).
  • “Meteorological Drivers of Mass and Energy Exchange between Inland Waters and the Atmosphere — MEDIWA” (project number: 445326344).
• SIGNAL project, funded by the German Ministry of Education and Research (BMBF), through the project BonaRes (Module A, SIGNAL 031A562A, 031B0510A and 031B1063A).

Citation

@article{Kröcher2026Monitoring,
  author = {Kröcher, Jenny and Ghazaryan, Gohar and Spank, Uwe and Zimmermann, Beate and Beyrich, F and Dietrich, Ottfried and Markwitz, Christian and Ramshorst, Justus van and Callejas-Rodelas, José Ángel and Lischeid, Gunnar},
  title = {Monitoring changes in the landscape water balance: validation of satellite- and model-based evapotranspiration data in Lusatia, Germany},
  journal = {Environmental Monitoring and Assessment},
  year = {2026},
  doi = {10.1007/s10661-026-15120-8},
  url = {https://doi.org/10.1007/s10661-026-15120-8}
}