![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Difficulties of biomass estimation over natural grassland |
VerfasserIn |
Péter Kertész, Bernadett Gecse, Krisztina Pintér, Szilvia Fóti, Zoltán Nagy |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250153941
|
Publikation (Nr.) |
EGU/EGU2017-18980.pdf |
|
|
|
Zusammenfassung |
Estimation of biomass amount in grasslands using remote sensing is a challenge due to the
high diversity and different phenologies of the constituting plant species. The aim of this
study was to estimate the biomass amount (dry weight per area) during the vegetation period
of a diverse semi-natural grassland with remote sensing. A multispectral camera (Tetracam
Mini-MCA 6) was used with 3 cm ground resolution. The pre-processing method includes
noise reduction, the correction for the vignetting effect and the calculation of the
reflectance using an Incident Light Sensor (ILS). Calibration was made with ASD
spectrophotometer as reference. To estimate biomass Partial Least Squares Regression
(PLSR) statistical method was used with 5 bands and NDVI as input variables.
Above ground biomass was cut in 15 quadrats (50×50 cm) as reference. The best
prediction was attained in spring (r2=0.94, RMSE: 26.37 g m−2). The average biomass
amount was 167 g m−2. The variability of the biomass is mainly determined by the
relief, which causes the high and low biomass patches to be stable. The reliability of
biomass estimation was negatively affected by the appearance of flowers and by the
senescent plant parts during the summer. To determine the effects of flower’s presence
on the biomass estimation, 20 dominant species with visually dominant flowers
in the area were selected and cover of flowers (%) were estimated in permanent
plots during measurement campaigns. If the cover of flowers was low (<25%),
the biomass amount estimation was successful (r2 >0,9), while at higher cover of
flowers (>30%), the estimation failed (r2 <0,2). This effect restricts the usage of the
remote sensing method to the spring – early summer period in diverse grasslands. |
|
|
|
|
|