The current version of GLO-PEM (GLObal Production Efficiency Model) was designed on the basis of the pro- duction efficiency concept (Prince, 1991; Prince, Justice & Moore, 1994) in which gross primary production (Pg photosynthesis-photorespiration) is calculated from the products of the photosynthetically active radiation (PAR) absorbed by the vegetation (APAR) and a potential conver- sion 'efficiency' or carbon yield of absorbed energy in terms of P_g, (Kumar & Monteith, 1982; Landsberg et al., 1995). The model is summarized in equation 1.

whereis the reductioncaused by environmental factors in time interval t (a proportion); is the potentialin terms of gross production (g C MJ ^-1);N_t is the proportion of incident PAR (S_t) absorbed by the vegetation (sometimes calledf_PAR). YgYm are measures of respiration.

     The APAR in time interval t is obtained from incident PAR (S_t) multiplied by N_t, the proportion of the incident PAR absorbed by the vegetation, estimated from the nor- malized difference vegetation index (NDVI = IR - R/ IR + R, where IR is near infrared and R red reflectance).

    When first introduced, s was regarded as an empirical constant. Here we propose to identify it with the quantum yield of a leaf, a well-known variable with defined values and dependancy on the particular bio- chemical carbon fixation pathway and temperature.  for C3 and C4 photosynthesis was calculated separately and then the value of  for a specific pixel was obtained from the estimated proportion of each carbon fixation pathway present in each 10-day period.

    The utilization of APAR in assimilation of carbon diox- ide in P_g is strongly affected by environmental variables such as air temperature, soil moisture and vapour pressure deficits (Runyon et al., 1993). The combined effects of environmental factors in reducing the efficiency from its potential value  to its actual value  were specified by the three components of , by analogy with stomatal models:

The first component()accounts for the reduction in assimilation as a result of low air temperature(T_a)since photosynthesis is known to be sensitive to low tempera- tures, and stomata may remain closed for several days following freezing temperatures;the second()accounts for the reduced stomatal conductance caused by high atmospheric water vapour pressure deficits; and the thirdaccounts for the effect of soil moisture on assimilation. Each component can vary between time intervals (t). Although originally developed for a leaf scale, this model was shown to account for stand scale behaviour in a range of sites in OTTER .

    Seasonal P_g was obtained from the time sum of the products of N_t,S_t, and .eq. 1). Seasonal P_n was calculated by multiplication of P_g by two factors, Y_g and Y_m, representing the proportion of P_g remaining after growth and maintenance respiration, respectively. Y. was modelled from the above-ground biomass of the vegetation using an algorithm by Hunt (1994). Above-ground biomass was obtained from a new algorithm based on minimized reflectance in the visible channel of the AVHRR.

    Unlike apparently similar models (Potter et al., 1993; Running & Hunt, 1993; Ruimy et al., 1994), at no point in GLO-PEM were field observations of primary production or used. These are reserved for validation.