Period: 2000 – present

Sensor: MODIS Terra+Aqua

Frequency: Yearly

Resolutions: 500m

Extent: Global

Data can be accessed here.

Algorithm description: MOD17A3H V006 MODIS annual NPP product is derived from 8-Day composite MOD17A2H product, which is 8-Day GPP composite, but it also consists of Net photosynthesis (PSN). NPP is calculated as PSN – Maintenance respiration (MR).

The method is based on ‘radiation conversion theory’ and uses MODIS FPAR product and independently estimated PAR. Conversion of PAR to NPP requires FPAR and radiation conversion efficiency. This parameter is also modified as a function of temperature and water pressure deficit stress factors. The parameter values are biome specific and are derived empirically from Biome-BGC model simulations. The respiration components (maintenance and growth) are calculated by allometric relationships between MODIS LAI and annual growth rates (based on Biome BGC simulations). All the requisite parameters are used from the Biome-BGC model.  The biome characterisation is done using MODIS land-cover map.

Accuracy: Heinsch et al (2006) report that the MODIS annual GPP algorithm had strong correlation (r = 0.859) with tower based annual GPP over a range of eddy covariance sites. However, MOD17 (earlier versions) have been found to be overestimating GPP and NPP for most of the sites, except most productive sites where it is underestimated. Authors also show that use of DAO meteorological data (as in MODIS algorithm) leads to about 28 to 45% error in GPP against the tower based meteorology input simulations. They found that the role of land cover related errors were reasonably small, but other sources of errors such as errors in the MODIS LAI product, FAPAR product and also scaling related issues with Biome-BGC model parameters and allometric equations could have significant effect on the accuracy.

Turner et al (2006) also report no overall bias in the MODIS NPP and GPP but overestimation in low productive sites.

Neumann et al (2015) showed that MODIS NPP compared better with the National Inventory based NPP, when it is estimated with local met-data than global datasets.


  • Validation is not available for recent version of the product.
  • Product is prone to error propagation considering applications of MODIS LAI, FPAR and Landcover map as as well as global PAR data.


Heinsch, F. A. et al. Evaluation of remote sensing based terrestrial productivity from MODIS using regional tower eddy flux network observations. IEEE Trans. Geosci. Remote Sens. 44,1908–1925 (2006).

Turner, D. P. et al. Evaluation of MODIS NPP and GPP products across multiple biomes. Remote Sens. Environ. 102, 282–292 (2006).

Running, S. W., Nemani, R., Glassy, J. M. & Thornton, P. E. MODIS Daily Photosynthesis (PSN) and Annual Net Primamy Production (NPP) Product (MOD17) Algorithm Theoretical Basis Document Version 3.0. 1–59 (1999).

Neumann, M., Zhao, M., Kindermann, G. & Hasenauer, H. Comparing MODIS Net Primary Production Estimates with Terrestrial National Forest Inventory Data in Austria. Remote Sens.7, 3878–3906 (2015).