Comparison of two Statistical Methods for the Derivation of the Fraction of Absorbed Photosynthetic Active Radiation for Cotton

Abstract

The fraction of absorbed photosynthetic active radiation (FAPAR) is an important input for modelling biomass increase and agricultural yield and can be calculated based on optical remote sensing data. In this study two remote sensing based approaches to derive the FAPAR for irrigated cotton in Fergana valley, Uzbekistan, are tested and compared: (i) FAPAR rescale from the normalized difference vegetation index (NDVI) (“percentile approach”), and (ii) an empirical regression approach based on NDVI. In the rescaling approach FAPAR was derived by relating upper and lower percentiles derived from the NDVI distribution of cotton fields from the entire study area to fixed FAPAR minima (bare soil) and maxima. NDVI was derived from multi-temporal 6.5 m RapidEye data acquired throughout 2011. For the regression approach FAPAR data was collected in situ from cotton fields during the vegetation season. The percentile approach delivered an RMSE of 0.10 whilst regression was only slightly better with an RMSE of 0.07. Hence, the percentile approach could be concluded as being a fast and easy alternative to field data demanding empirical regressions for the derivation of FAPAR on cotton fields.