These models enable the user to estimate forage quantity (green biomass, gBM) and forage quality (metabolisable energy, ME) from hyperspectral reflectance data of vegetation canopies measured using an ASD FieldSpec 3 (or higher) Hi-Res Portable Spectroradiometer. The models were calibrated in R statistical software and were provided as RData-Files. The models were calibrated using spectral reflectance as well as forage provision data from several sites within the study area, comprising a steep south-north gradient of climatic aridity reaching from northern Ghana to central Burkina Faso between 9.0°N and 13.5°N latitude and 0.1°W and 2.0°W longitude. It is located in the southern and northern Sudanian zone of West Africa’s savanna belt, capturing a precipitation range of 600 mm to 1200 mm corresponding to UNEP aridity indices of 0.31 (semi-arid) to 0.69 (humid). Data collection took place at 21 sites during the rainy season 2012 (June-September). Our sampling design intended to cover diverse vegetation types and a wide range of land-use intensities (including protected and degraded areas). We stratified sampling at sites by topographic position (upslope, footslope and lowland). Vegetation samples were oven-dried (60°C, > 48 h) to obtain gBM. Dried samples were ground for analysis of in vitro gas production (GP) using the Hohenheim gas test (HGT). Crude protein (CP) content was determined by LUFA NRW using Kjeldahl´s method (method 4.1.1). The ME was calculated using the following equation: ME (MJ kg-1 dry matter, DM) = 2.20 + 0.1357 GP + 0.0057 CP + 0.0002859 CP², where GP is expressed as ml 200 mg-1 DM and CP is expressed as g kg-1 DM. A partial least-squares regression was used to model the relations between spectral data and target variables (ME and gBM). We used the PLSR implementation in the R package 'autopls'. During PLSR, we also performed multiplicative scatter correction and brightness normalization of reflectance spectra. Before the models can be applied, spectra have to be smoothed using Savitzky-Golay smoothing filter and noisy regions (bands 940:1170, 1350:1700, 2001:5000) have to be excluded from model application. Further information: Ferner, J., Linstädter, A., Südekum, K. H., & Schmidtlein, S. (2015). Spectral indicators of forage quality in West Africa’s tropical savannas. International Journal of Applied Earth Observation and Geoinformation, 41, 99-106.

The database contains point data from several sites within the study area, comprising a steep south-north gradient of climatic aridity reaching from northern Ghana to central Burkina Faso between 9.0°N and 13.5°N latitude and 0.1°W and 2.0°W longitude. It is located in the southern and northern Sudanian zone of West Africa’s savanna belt, capturing a precipitation range of 600 mm to 1200 mm corresponding to UNEP aridity indices of 0.31 (semi-arid) to 0.69 (humid). Data collection took place at 44 sites during the rainy seasons in 2012 (June-September) and 2013 (July-October) and the end of the rainy season 2014 (October). Our sampling design intended to cover diverse vegetation types and a wide range of land-use intensities (including protected and degraded areas). We stratified sampling at sites by topographic position (upslope, footslope and lowland). At each slope position per site, 3-5 plots (10x10 m) were placed, containing three random circular subplots. The point data collected at each (sub)plot contains modelled data on forage quality (metabolisable energy) and quantity (green & total biomass) from spectrometric measurements, vegetation relevées (i.e. species data, vegetation clusters, phenology, photosynthetic pathway, height, cover, life span...), interview data on livestock keeping, grazing pressure and soil attributes (N, C, P, soil depth, soil structure...). Based on climate data, aridity indices were calculated. Based on fire occurrences detected by satellite images, fire frequency around sites was calculated. Based on soil moisture values detected by (radar) satellite images, soil moisture was extracted and modelled for the time of data collection. Further information: Ferner, J., Linstädter, A., Südekum, K. H., & Schmidtlein, S. (2015). Spectral indicators of forage quality in West Africa’s tropical savannas. International Journal of Applied Earth Observation and Geoinformation, 41, 99-106.

The time series of maps is showing three differnt forage parameters (green biomass, metabolisable energy, metabolisable energy yield) from 04.12.2015 to 29.09.2016. The covered area matches the Vea catchment at the border region of Ghana and Burkina Faso. For model calibration, data was collected during the rainy season 2012 at 21 sites spread along a north-south climate gradient reaching from northern Ghana to central Burkina Faso. Spectral reflectance measurements of vegetation plots were performed using a FieldSpec 3 Hi-Res Portable Spectroradiometer (ASD Inc., Boulder, CO, USA). Partial least squares regression was used to calibrate spectral models between reflectance data and forage characteristics (green biomass & metabolisable energy). Metabolisable enrgy yield was calculated based on the other two forage parameters. For more details regarding sampling design and model calibration, see Ferner, Linstädter et al. (2015). Sentinel-2 images were acquired for twelve dates. Imagery was atmospherically corrected using the plugin ‘sen2cor’ within the SNAP toolbox, provided by the European Space Agency (ESA). To assure that forage models were only applied on vegetated areas, a vegetation mask was build based on MESMA (multiple endmember spectral mixture analysis) and only pixels with a fractional cover of green vegetation of more than 30% were used. Finally, forage provision models resampled to Sentinel-2 spectral resolution were applied to obtain maps of estimated forage provision. Further information: Ferner, J., Linstädter, A., Südekum, K. H., & Schmidtlein, S. (2015). Spectral indicators of forage quality in West Africa’s tropical savannas. International Journal of Applied Earth Observation and Geoinformation, 41, 99-106.

The time series of maps is showing three differnt forage parameters (green biomass, metabolisable energy, metabolisable energy yield) from 04.12.2015 to 29.09.2016. The covered area matches the Vea catchment at the border region of Ghana and Burkina Faso. For model calibration, data was collected during the rainy season 2012 at 21 sites spread along a north-south climate gradient reaching from northern Ghana to central Burkina Faso. Spectral reflectance measurements of vegetation plots were performed using a FieldSpec 3 Hi-Res Portable Spectroradiometer (ASD Inc., Boulder, CO, USA). Partial least squares regression was used to calibrate spectral models between reflectance data and forage characteristics (green biomass & metabolisable energy). Metabolisable enrgy yield was calculated based on the other two forage parameters. For more details regarding sampling design and model calibration, see Ferner, Linstädter et al. (2015). Sentinel-2 images were acquired for twelve dates. Imagery was atmospherically corrected using the plugin ‘sen2cor’ within the SNAP toolbox, provided by the European Space Agency (ESA). To assure that forage models were only applied on vegetated areas, a vegetation mask was build based on MESMA (multiple endmember spectral mixture analysis) and only pixels with a fractional cover of green vegetation of more than 30% were used. Finally, forage provision models resampled to Sentinel-2 spectral resolution were applied to obtain maps of estimated forage provision. Further information: Ferner, J., Linstädter, A., Südekum, K. H., & Schmidtlein, S. (2015). Spectral indicators of forage quality in West Africa’s tropical savannas. International Journal of Applied Earth Observation and Geoinformation, 41, 99-106.

The time series of maps is showing three differnt forage parameters (green biomass, metabolisable energy, metabolisable energy yield) from 04.12.2015 to 29.09.2016. The covered area matches the Vea catchment at the border region of Ghana and Burkina Faso. For model calibration, data was collected during the rainy season 2012 at 21 sites spread along a north-south climate gradient reaching from northern Ghana to central Burkina Faso. Spectral reflectance measurements of vegetation plots were performed using a FieldSpec 3 Hi-Res Portable Spectroradiometer (ASD Inc., Boulder, CO, USA). Partial least squares regression was used to calibrate spectral models between reflectance data and forage characteristics (green biomass & metabolisable energy). Metabolisable enrgy yield was calculated based on the other two forage parameters. For more details regarding sampling design and model calibration, see Ferner, Linstädter et al. (2015). Sentinel-2 images were acquired for twelve dates. Imagery was atmospherically corrected using the plugin ‘sen2cor’ within the SNAP toolbox, provided by the European Space Agency (ESA). To assure that forage models were only applied on vegetated areas, a vegetation mask was build based on MESMA (multiple endmember spectral mixture analysis) and only pixels with a fractional cover of green vegetation of more than 30% were used. Finally, forage provision models resampled to Sentinel-2 spectral resolution were applied to obtain maps of estimated forage provision. Further information: Ferner, J., Linstädter, A., Südekum, K. H., & Schmidtlein, S. (2015). Spectral indicators of forage quality in West Africa’s tropical savannas. International Journal of Applied Earth Observation and Geoinformation, 41, 99-106.