BE-HyperSpecPrediction

Prediction of biodiversity variables using hyperspectral observations.

Workflow for the prediction of plant traits and biodiversity on grasslands in the biodiversity exploratories.

Prepare hyperspectral data

Scripts:

1. set_environment
2. get_hyperspectral_data
   • locate 10m centers in GIS, avoid roads, climate stations etc.
3. crop_hyperspectral_data

Indices and plot based summary

Scripts:

1. vegetation_indices
2. narrow_range_indices
3. single_bands

Output format of the data frames:

EPID	WL / Index	mean	sd

Script: plot_based_summary
Converted to a data frame with one row per Plot. Because of previous model namings, the column names need to be:

• Single bands: mean400.4
• Vegetation index: Carter_mean, Carter_sd
• NRI: mean_wl_400_404 (Wavelength rounded)

Initial modelling with AEG

the 26 plots of the Alb Exploratory are used in the Scripts:

1. 07_merge_predictors
2. 08_model_preparation
3. 09a_independent_fold_training
4. 09b_independent_fold_control

Model validation

the other two exploratories are used afterwards for model validation on independent regions Scripts:

1. field_data_summary
2. extract_aeg_model_variables
3. predict_aeg_hai_sch