Period: 1981 – present

Sensor: AVHRR

Frequency: 1 day

Resolutions: 5km

Extent: Global

Data can be accessed here.

Algorithm description: The product has been developed to ensure continuity of AVHRR based LAI/FAPAR products. The algorithm relies on Artificial Neural Networks (ANN) relating LAI to BRDF corrected AVHRR surface reflectance products for five biomes. MODIS LAI/FAPAR data was used as training dataset. It consists several steps: (1) input data normalisation using VJB model (2) ANN execution per class (3) output normalisation (4) class fusion according to IGBP land cover

Accuracy:  Accuracy assessment was done for BELMANIP and DIRECT global sites. Reproducibility of the algorithm was demonstrated to achieve overall uncertainty performance of 0.54. But per-biome scores were contrasting. Best performances are computed for croplands, grassland and non-vegetated surfaces. Overall uncertainty of 1.03 was found for LAI on average (R2 ~ 0.7).

These data represent global longest daily time series derived from a single sensor type.

Limitations:

  • Incapacity of the algorithm to reproduce variability in dense vegetated covers
  • Significant saturation of the algorithm for high LAI (>4.5) values.
  • Very coarse spatial resolution.

Reference:

Claverie, M., Matthews, J., Vermote, E. & Justice, C. A 30+ Year AVHRR LAI and FAPAR Climate Data Record: Algorithm Description and Validation. Remote Sens. 8, 263 (2016).