Interannual variability of global terrestrial primary production: Results of a model driven with satellite observations

Interannual variation in terrestrial net primary production (NPP) was modeled using the global production efficiency model (GLO‐PEM), a semimechanistic plant photosynthesis and respiration model driven entirely with satellite advanced very high resolution radiometer (AVHRR) observations. The model also estimated a wide range of biophysical variables at 10‐day intervals for the period 1982–1989, including air temperature, vapor pressure deficit, soil moisture, biomass, autotrophic respiration, canopy‐absorbed photosynthetically active radiation, gross primary production, and light use efficiency. The accuracy of the simulated variables has previously been shown to be within 10–30% of field measurements, depending on the specific variable. We analyze here interannual changes in NPP, which showed large spatial variability (0–1500 gC m−2 yr−1) and trends that differed regionally over the 8‐year period. Annually integrated global NPP was found to vary as much as 12% between years and was very sensitive to air temperature. The coefficient of variation in NPP of sparsely vegetated areas (mostly semiarid) on an interannual basis was as much as 80%, whereas densely vegetated areas (broadleaf evergreen and seasonally deciduous forests) varied comparatively little (0–10%). Mean annual NPP of the latter decreased 36 gC m−2 yr−1 over the time series examined. There was extreme seasonal and moderate interannual variation (10–60%) in NPP of middle‐ to high‐latitude regions (temperate and boreal forests) with evidence for a slight trend toward increased values through time (+3 to 12 gC m−2 yr−1). The results indicate significant interannual and regional differences in responses to climate variability, with boreal regions increasing 39 gC m−2 yr−1 compared to a decrease of 116 gC m−2 yr−1 in tropical regions for each 1°C rise in air temperature. We explore a few of the possible reasons for these observations and discuss some of the issues and limitations to the use of the current global AVHRR observational record.