Professional Documents
Culture Documents
net/publication/334282912
CITATIONS READS
0 256
6 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Birgitta Putzenlechner on 07 July 2019.
(a) (b)
(c)
Figure 5: Daily fAPAR of WSN vs. S2 at (a) Graswang, (b) Peace River and
(c) Santa Rosa. Root Mean Square Error (RMSE), Mean Absolute Error
Figure 1: Locations of the three sites equipped with WSNs: (a) At Graswang (MAE), coefficient of determination (R²), Major Axis Regression (M.A.R.) and
TERENO site, the reference sensor for incoming PAR is located on grassland percent bias (%BIAS) are shown. The continuous black lines correspond to
outside the forest, transmitted PAR was measured inside the forest with 16 slopes and intercepts of the M.A.R.s, the dashed lines display the 95%-
WSN sensors; (b) The Peace River Environmental Super Site was equipped significance level. The red line marks the 1:1 line.
with 22 WSN sensors for transmitted PAR inside the forest, incoming PAR was
measured on top of a flux tower; (c) At Santa Rosa National Park
Environmental Monitoring Super Site, data of 16 sensors for transmitted PAR CONCLUSIONS
were used; incoming PAR was measured on top of a flux tower.
• fAPAR monitoring performed with WSNs can be used to
Figure 4: Daily fAPAR as spatial average and at individual sensor locations retrieve in-situ estimates that are urgently needed to
METHODS measured with WSNs compared to the S2 fAPAR product for the year 2017 at
validate and improve satellite-derived fAPAR products.
(a) Graswang, (b) Peace River and (c) Santa Rosa sites. Panel (d) shows
• Quantum PAR sensors were arranged in Wireless Sensor fAPAR distinguished for sensor locations with a higher percentage of spruces • Uncertainties of ground data (< 5%) could not explain high
Networks (WSNs) to ensure synchronized and multi-sensor and beeches, respectively. discrepancies with S2 fAPAR product.
data acquisition (Fig. 2).
Table 1: Overview of the MEAN and SD of distributions of different temporal • The choice of temporal fAPAR estimate should be well-
• Incoming and transmitted PAR was measured above and fAPAR estimates of spatially aggregated fAPAR at different sites; correlation argued and included in validation protocols.
inside the canopy at 1.3 m every 10 min (Fig. 2). analysis and performance metrics of temporal fAPAR estimates vs. the
fAPAR S2 product are shown. Significant differences (n = 206, KS test: p < • The actual S2 fAPAR product at the forested sites holds
→ Calculation of a two-flux fAPAR estimate 0.05) between value distributions of fAPAR estimates and daily fAPAR are limited opportunities to feed production efficiency models
indicated with *. but showed strong capabilities for monitoring phenological
• Cloud-free S2 images were atmospherically corrected
("Sen2Cor" algorithm); fAPAR retrieval with “Biophysical changes as well as spatial variability attributed to different
Processor” (S2 toolbox in ESA's software SNAP) (Fig. 3). species composition.
• Uncertainty assessment of ground data: investigations on
the bias of fAPAR during high solar zenith angles and REFERENCES
colored autumn leaves (PUTZENLECHNER ET AL. 2019); • D’ODRICO, P., GONSAMO, A., PINTY, B., GOBRON, N., COOPS, N., MENDEZ, E.,
calculation of different temporal aggregation schemes, i.e. SCHAEPMAN, M. E. (2014): Intercomparison of fraction of absorbed
daily integrated fAPAR and fAPAR acquired at solar noon photosynthetically active radiation products derived from satellite data over
Europe, Remote Sensing of Environment, 142, 141-154.
• GLOBAL CLIMATE OBSERVING SYSTEM (GCOS) (2011): Systematic
observation requirements for satellite-based data products for climate.
• PUTZENLECHNER, B., MARZAHN, P., KIESE, R., LUDWIG, R., SANCHEZ-
AZOFEIFA, A. (2019): Assessing the variability and uncertainty of two-flux
FAPAR measurements in a conifer-dominated forest, Agricultural and Forest
Meteorology, 264, 149-163.
(a) • WEISS, M., BARET, F. (2016): S2ToolBox Level 2 products: LAI, FAPAR,
FCOVER.
• WIDLOWSKI, J.-L. (2010): On the bias of instantaneous fAPAR estimates in
open-canopy forests, Agricultural and Forest Meteorology, 150, 1501-1522.
ACKNOWLEDGEMENTS
This work was supported by the Natural Science and Engineering Research
(b) Council of Canada (NSERC) Discovery Grant Program, the Inter-American
Institute for Global Change Research (IAI) Collaborative Research Network
Program (CRN3-023), the Canada Foundation for Innovation (CFI), and a
MICMoR Fellowship to B.P. through KIT/IMK-IFU (Garmisch-Partenkirchen,
Figure 2: Experimental setup of the study sites: WSN node were equipped Germany). The Helmholtz Association and the German Federal Ministry of
with PAR, temperature, relative humidity and soil moisture sensors (left image) Figure 6: Uncertainty assessment at Graswang: (a) spatially aggregated
measured fAPAR (±SD) for different solar zenith angles at the site; (b) Education and Research (BMBF) in the framework of TERENO (Terrestrial
and arranged in hexagonal sampling schemes (here shown for the site Environmental Observatories) (grant no. 01LL0801B) provided support also to
Graswang) (right image). distributions of measured fAPAR under diffuse light conditions for different
periods of leaf colors of beech leaves (PUTZENLECHNER ET AL. 2019). the German site.