Agroforestry meets Remote Sensing


Our study is part of the citizen-science agroforestry project where we try to find out how we could apply the satellite-based Normalized Difference Vegetation Index (NDVI) as a proxy for productivity and thus crop yield of agroforestry systems being able to

  1. investigate the spatial variation across the agroforestry sites and to compare those to the control sites,
  2. make predictions about the effect of additional trees/shrubs on cropland and pastureland, respectively by monitoring the productivity, as well as
  3. give advice for reduction of fertilizer applications by predicting crop yield of agroforestry systems in a long-term and precise way.


In an effort to do that, we depend on the access to high-resolution satellite data, such as WorldView-2 and GeoEye-1 due to the study design and distances between tree rows or our measurement points of field data, respectively (e.g. 2.25 m distances). The NDVI is calculated as NDVI = (NIR-R)/(NIR+R). Regarding to Yang et al. (2013), our study areas and NDVI, respectively will be represented by their colour-infrared image. Second, we will compile classification maps due to thresholds based on the NDVI and field data. Finally, we will be able to predict yield maps by combining NDVI, field data and literature.


We will contribute pointing out the advantages of agroforestry systems with our study by visualizing the results obtained from satellite images in a comprehensible and convincing way:

  • R script for calculating and visualizing the NDVI of a specific cropland or pastureland, respectively of our cooperating farms
  • Table of measurement periods for arable crop-specific NDVI values
  • Recommendations for reduction of fertilizer application due to NDVI thresholds

Our developed method will be embedded in the catalogue of methods.