Abstract

Land and ocean are often treated separately in modeling studies despite their close links through the carbon, water, and energy cycles. However, biospheric models, particularly when used in conjunction with recent satellite datasets, provide a new, fully coupled, global perspective. The current investigation uses a new version of the Grid Enabled Integrated Earth system (GENIE-SF) to compare both the magnitude and the seasonal and zonal variation in water flux [evaporation E and precipitation (PPT)] and carbon flux [net primary productivity (NPP)] above land and ocean. GENIE-SF contains state-of-the-art representations of photosynthesis and is driven by the phenological cycles of leaf area index (LAI) and marine chlorophyll concentration, both recorded with the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite sensors. The current study reveals the striking uniformity of the ocean–atmosphere carbon and water flux exchange, both temporally and spatially, compared to the corresponding land–atmosphere exchange. Although biospheric annual NPP (108 ± 27 GtC yr−1) is split almost equally between land (52% ± 9%) and ocean (48% ± 9%), the oceanic contribution to biospheric annual E exceeds that of the land by a factor of 6.7 ± 1.7. Simulations conducted over a 50-yr period (1951–2000) suggest that a 16% increase in land NPP, owing mainly to CO2 fertilization, may be partially offset by a decline in marine productivity.

1. Introduction

Ocean and land are often modeled independently despite their close links through the carbon, water, and energy cycles. It is becoming increasingly important, however, to gain a biospheric (land + ocean) perspective of the globe to understand how the different components interact, potentially reinforcing or offsetting each other under climate change (e.g., the relative importance of land and ocean as sinks of anthropogenically released carbon; Houghton 2007).

The relative importance of land and ocean within the global carbon cycle has been examined in a small number of studies. For example, Field et al. (1998) estimated the partitioning of biospheric net primary productivity (NPP) using Advanced Very High Resolution Radiometer (AVHRR) satellite data for land plants in combination with the Coastal Zone Color Scanner (CZCS) for ocean chlorophyll concentration.1 Behrenfeld et al. (2001) conducted a similar study using Sea-Viewing Wide Field-of-View Sensor (SeaWiFS) data for both land and ocean. Both of these pioneering studies were highly insightful. However, they adopted a relatively straightforward algorithm for land photosynthesis based on constant light-use efficiency (LUE) that, though useful on annual time scales, has difficulties reproducing observed carbon fluxes on shorter time scales (Medlyn et al. 2003; Alton and Bodin 2010). Land and ocean are also sometimes considered together when examining anomalies in rising CO2 atmospheric concentration and the variability of global carbon sinks (Peylin et al. 2005; Houghton 2007). However, studies of the seasonal and zonal distribution of NPP, comparing land and ocean, are fairly scarce.

Table 1.

An alphabetical list of acronyms and abbreviations used frequently in the current study.

An alphabetical list of acronyms and abbreviations used frequently in the current study.
An alphabetical list of acronyms and abbreviations used frequently in the current study.

In a similar vein, attempts to quantify the role of land and ocean in the global water cycle, by examining the distribution of biospheric evaporation E, for example, are surprisingly rare, although such fluxes are implicit in global simulations of present and future climate. Indeed, as stated recently by Trenberth et al. (2007), few studies attempt to provide a quantitative synthesis of the global water cycle. Conventional, textbook estimates of water fluxes (e.g., Strahler and Strahler 1992; Schlesinger 1997) rely heavily on sparse and nonuniform in situ measurements rather than the global perspective afforded by recent satellite data (Schlosser and Houser 2007) or a biospheric model. Recently, Trenberth et al. (2007) focused on the main global reservoirs of water, and the annual fluxes between them, combining observational and reanalysis datasets with a forced (i.e., noncoupled) land surface model (LSM). Using similar sources, Mehta et al. (2005) quantified the seasonal cross-equatorial transfer of atmospheric moisture. However, explicit estimates of seasonal and zonal water fluxes, comparing land and ocean in a fully coupled biospheric model, are uncommon.

The recent development of integrated earth-modeling systems permits a biospheric view of carbon and water fluxes (Petoukhov et al. 2000; Claussen et al. 2002; Brovkin et al. 2003). The current investigation adopts one such modeling system, the Grid Enabled Integrated Earth system (GENIE), which combines a general circulation model (GCM) of intermediate complexity with an ocean model and a LSM (Lenton et al. 2007). The latest satellite measurements of LAI and marine chlorophyll concentration, both recorded with the same satellite sensor [the Moderate Resolution Imaging Spectroradiometer (MODIS)], are used to drive GENIE. An investigation into the global cycles of carbon, water, and energy is too ambitious in scope, particularly if we wish to compare land and ocean. Therefore, we restrict our focus to biospheric carbon and water fluxes, contrasting both the magnitudes and the spatial and temporal variation in NPP and E between land and ocean. We refer to Trenberth et al. (2009) for a recent and thorough review of biospheric energy exchange and the global radiation budget.

Specific objectives of the current investigation are as follows:

  1. to estimate the division of annual NPP between land and ocean by combining MODIS satellite data with a biospheric model (GENIE) containing state-of-the-art representations of photosynthesis;

  2. to estimate the relative contributions of land and ocean to biospheric E and to compare with the corresponding partitioning of biospheric NPP;

  3. to quantify and compare the spatial and temporal variation in NPP and E between land and ocean; and

  4. to provide an estimate of the earth’s fundamental water cycle using a biospheric model, in particular the transfer of atmospheric moisture from ocean to land and the return of precipitation to the atmosphere (precipitation recycling) over the global landmass.

2. Methods

In the following three sections we discuss the modeling framework, the datasets required to run and validate the model, and the experiment conducted to examine biospheric carbon and water balance.

a. Modeling framework: GENIE

As indicated in the introduction, our objectives are scientific (quantifying carbon and water fluxes) rather than technical (model improvement or intermodel comparison). Therefore, only a qualitative overview of the model components is given here with detailed equations contained in the appendix or supplied via reference. The modeling framework is provided by GENIE-SF, which is a modified version of GENIE containing state-of-the-art representations of biophysical processes, in particular, photosynthesis on land and in the ocean. Realistic processes are important as predicted carbon fluxes differ by at least 20%–30% according to the complexity with which, for example, canopy light interception and photosynthesis are formulated (Cramer et al. 1999; Knorr and Heimann 2001; Medlyn et al. 2003). The three main components of GENIE-SF are ocean, atmosphere, and land, which are all run at a spatial resolution of 5.6°. This coarse spatial resolution provides adequate precision in predicted land fluxes (≤5%; Müller and Lucht 2007) while facilitating the calculation of carbon, water, and energy exchange at regional and global scales. While accurate process representation appears to carry more weight than high spatial resolution within the land component, less consensus exists on the relative importance of these two elements (processes and spatial resolution) within the ocean and atmosphere components. Nevertheless, the value of systems of intermediate complexity, such as GENIE-SF, in providing reliable flux estimates at larger (regional/continental) scales is recognized by the Intergovernmental Panel on Climate Change (IPCC) (Randall et al. 2007). The three components of GENIE-SF are summarized below.

1) Ocean

The ocean component uses prescribed, monthly sea surface temperature (SST) obtained from National Centers for Environmental Prediction (Smith et al. 2008) averaged to a 5.6° resolution. Marine primary productivity is calculated using the vertically generalized production model (VGPM; Behrenfeld and Falkowski 1997), which has been previously tested and calibrated against ocean plankton sampling measurements. VGPM estimates phytoplanktonic primary production according to satellite-measured surface chlorophyll concentration and SST. The main equation for the VGPM is given in the appendix. We do not simulate the full carbon cycle within the ocean (dissolution of CO2, burial of carbonate sediments, etc.) owing to the relatively short time scale of the current simulation and the selection of a simple, noncirculatory ocean component based on monthly SST. Note that an explicit treatment of planktonic primary productivity has only been introduced relatively recently into a few atmosphere–ocean GCMs (e.g., Palmer and Totterdell 2001; Dutkiewicz et al. 2005). As demonstrated in the results, this more process-based approach (as opposed to using prescribed ocean carbon fluxes) allows temporal changes in ocean carbon balance to be monitored using satellite data.

2) Atmosphere

The atmospheric component is the Reading 3D Intermediate General Circulation Model (IGCM3.1), which has, at its adiabatic core, the 3D spectral primitive equation model of Hoskins and Simmons (1975). To expedite the atmospheric calculations, the number of vertical resolution elements is reduced from a standard number of 22 to just 7. The radiation scheme, based on daily averages, uses a lookup table of longwave (LW) transmittance and the two-band scheme (visible and near-infrared radiation) of Morcrette (1990) in the shortwave. The IGCM is run with a Tiedtke convection scheme (Molteni 2003), which performs reasonably well in tuning exercises conducted with GENIE (Annan et al. 2005).

3) Land

The land surface component is based on the Met Office Surface Exchange Scheme (MOSES), adopting the Penman–Monteith approach (Monteith 1965) to ensure energy balance within each of 10 tiles (4 nonvegetation tiles and 6 plant functional types) constituting any given land point (see appendix for detailed equations). Cover type for each land point is taken from the International Geosphere-Biosphere Project classification (IGBP 1992; Hansen and Reed 2000). Vegetation is represented as a multilayer canopy where light interception, foliar nitrogen, leaf photosynthesis, and leaf stomatal conductance are calculated explicitly within each of five leaf layers before being summed to produce canopy totals. Leaf photosynthesis is derived using a colimitation biochemical model (Collatz et al. 1991) and stomatal conductance is linked to photosynthesis through a modified Leuning relation (Cox et al. 1998). Direct sunlight is calculated as an exponential probabilistic function (Norman 1992; Alton and Bodin 2010) while the distribution of diffuse sunlight follows from the two-stream radiative calculation of Sellers et al. (1996). The fraction of diffuse solar irradiance, incident at the top of the canopy, is fixed on a daily basis by comparing the daily average surface shortwave (SW) irradiance within the IGCM with the top-of-atmosphere daily average solar radiation (Roderick et al. 2001). For simplicity, a single 2-m soil layer is adopted, which is represented by a leaky bucket hydrology (Huang et al. 1996; Gedney et al. 2000).

The meteorology required to force the land surface component is supplied by the IGCM on a 24-h time step. From this, hourly SW irradiance is calculated according to time, day, and latitude, assuming clear-sky conditions (Campbell and Norman 1998). These hourly values of SW irradiance are then normalized, so that their average over the 24-h cycle matches the daily SW irradiance calculated by the IGCM. Canopy photosynthesis, respiration, and evaporation are calculated on an hourly basis and their daily averages returned to the IGCM. The importance of modeling SW irradiance across the diurnal cycle has been demonstrated at eddy covariance flux towers (Hollinger et al. 1994). However, the diurnal cycle is present in only 3 out of 12 large-scale LSMs (Medlyn et al. 2003), many of which are involved in GCM climate simulations. Furthermore, observed carbon and water fluxes at eddy covariance sites underline the importance of simulating light interception at multiple heights within the canopy (Alton et al. 2005; Mercado et al. 2009) as is the case for GENIE-SF. Currently, 9 of 12 of radiative-transfer schemes adopted in large-scale LSMs implement either the Big Leaf (Schulze et al. 1994) or the constant LUE approach (Yuan et al. 2007) when scaling leaf photosynthesis to canopy level (Medlyn et al. 2003). Field campaigns reveal, however, incomplete light acclimation of leaf nitrogen in tree canopies in violation of the key assumption of the Big Leaf paradigm (Carswell et al. 2000; Lewis et al. 2000; Meir et al. 2002). Constant LUE, while widely adopted with satellite data, may be sufficient for annual totals of NPP but has difficulties reproducing observations on shorter time scales (Medlyn et al. 2003; Alton and Bodin 2010), particularly under varying conditions of direct and diffuse solar radiation (Alton et al. 2007; Knohl and Baldocchi 2008; McCallum et al. 2009).

4) Model inputs

To summarize the model inputs, meteorological forcing for land and ocean is provided by the IGCM. Phenology on land and in the ocean is provided by satellite-derived annual cycles of, respectively, LAI and chlorophyll concentration, both of which are discussed below (section 2b). Parameters for the ocean component (e.g., optimal rate of carbon fixation and the euphotic zone depth) are fixed by the original model authors (see appendix) using measured productivity rates present in ocean samples. For parameters in the land component we follow Friend et al. (1997) and Zaehle et al. (2005), by assigning average field measurements reported in the literature (Table 2). Our primary reason for eschewing parameter “tuning” is that we prefer to use the few observational datasets available for regional NPP and E for validation purposes (see below) rather than for calibration of the model. Furthermore, covariance amongst key parameters (equifinality) makes it difficult to assign a unique parameter configuration in LSMs simply by tuning against observed fluxes (Medlyn et al. 2005).

Table 2.

Key parameter values for the land component of GENIE-SF, based on average field measurements reported in the literature (Alton and Bodin 2010). PFTs are tropical broadleaf forest (TBL), extratropical broadleaf forest (BL), needleleaf forest (NL), C3 grasses (C3), C4 grasses (C4), and shrubs (SH). Where single parameter values are given, the value applies to all PFTs. PAR and NIR are, respectively, photosynthetically active radiation and near-infrared radiation. Both Vcrit and Vwilt are set to quite low values to ensure the onset of plant water stress at fairly low soil moisture content, consistent with field measurements (Newman 1969).

Key parameter values for the land component of GENIE-SF, based on average field measurements reported in the literature (Alton and Bodin 2010). PFTs are tropical broadleaf forest (TBL), extratropical broadleaf forest (BL), needleleaf forest (NL), C3 grasses (C3), C4 grasses (C4), and shrubs (SH). Where single parameter values are given, the value applies to all PFTs. PAR and NIR are, respectively, photosynthetically active radiation and near-infrared radiation. Both Vcrit and Vwilt are set to quite low values to ensure the onset of plant water stress at fairly low soil moisture content, consistent with field measurements (Newman 1969).
Key parameter values for the land component of GENIE-SF, based on average field measurements reported in the literature (Alton and Bodin 2010). PFTs are tropical broadleaf forest (TBL), extratropical broadleaf forest (BL), needleleaf forest (NL), C3 grasses (C3), C4 grasses (C4), and shrubs (SH). Where single parameter values are given, the value applies to all PFTs. PAR and NIR are, respectively, photosynthetically active radiation and near-infrared radiation. Both Vcrit and Vwilt are set to quite low values to ensure the onset of plant water stress at fairly low soil moisture content, consistent with field measurements (Newman 1969).

b. Datasets

We use measurements from MODIS to determine the phenological cycles of LAI (m2 m−2) on land (Terra satellite) and chlorophyll-a concentration (mg Chl m−3) within the ocean (Aqua satellite). The MOD15 LAI product for the year 2002 (Morisette et al. 2006) is averaged to produce a phenology of 5.6° spatial resolution and 16-day temporal resolution. The MODIS/Aqua Level 3 standard mapped image data for ocean chlorophyll concentration, covering the period 2003–08 (Esias et al. 1998; Franz et al. 2005), are averaged in a similar manner. However, GENIE only calculates ocean productivity on a monthly time step owing to the fixed monthly SST used in the ocean component. Although leaf phenology is only derived from a single year of satellite measurements, simulations based on decadal satellite observations of LAI (Alton et al. 2009) exhibit an interannual variability in LAI of 5% (5°)−1 latitudinal zone. This variation is small compared to the uncertainties we estimate below (section 3a) for predicted annual zonal fluxes (20%–25%).

There is a dearth of observational datasets available for the validation of carbon and water fluxes at regional and global levels. Furthermore, the uncertainties associated with such datasets are quite large (15%–30%; Coe 2000; Adler et al. 2003; Fisher 2007). To validate the land–atmosphere fluxes, we compare simulated NPP against the Ecosystem Model-Data Intercomparison (EMDI) archive (Scurlock and Olson 2002; Olson et al. 2008), which contains site measurements of annual NPP recorded mostly between 1960 and 1990. For each 5.6° latitudinal band, the EMDI values are grouped and averaged according to PFT and an “observed” value assigned according to the fractions composing the latitude band in question. This procedure is carried out separately for both the 632 class B sites and the 81 higher quality class A sites to gauge the uncertainty in the observed profile. Note this comparison makes use of the new release of the EMDI dataset (end 2008), which corrects a large error in tropical NPP present in previous versions. Predicted continental E is compared against the measurement-based estimates of Baumgartner and Reichel (1975) and Henning (1989). Runoff is compared to measured river discharge (Labat et al. 2004; Trenberth et al. 2007; Dai et al. 2009). Ocean productivity is checked against previous estimates based on both satellite data (Field et al. 1998; Behrenfeld et al. 2001) and ocean phytoplanktonic sampling (Lieth 1975). Ocean E and global seasonal PPT are compared against satellite-derived products (Schlosser and Houser 2007 and references therein). For consistency we refer to evaporation when discussing both land and ocean. However, it should be noted that land E includes water vapor released through plant stomata (transpiration).

c. Experiments

The main GENIE-SF simulation is conducted over a 13-yr period from 1990–2002. The selected time period necessarily represents a compromise between the satellite observing periods for LAI and marine chlorophyll and ground-based measurements, such as NPP and runoff, used to validate the model output. In section 3c, we examine the sensitivity of our results to this temporal offset. The first 3 years of the main simulation act as spinup for the land component. This spinup applies only to soil moisture since carbon stocks within the model are prescribed at currently observed levels. Over the remaining 10 years, the annual fluxes for 5.6° latitudinal bands are convergent within 3%. Therefore, monthly and annual fluxes are averaged over this 10-yr period to compare the temporal and zonal distribution of NPP and E for land and ocean.

As indicated in the introduction, the current study focuses on the contemporary global cycles of carbon and water, which are still quite poorly quantified (Trenberth et al. 2007). However, we make a tentative estimate of how the partitioning of biospheric fluxes between the ocean and continents may be changing. The crude spatial resolution of the biospheric model makes such simulations computationally feasible. Thus, we run GENIE-SF over a 50-yr period (1951–2000) for which LAI has been reconstructed using an algorithm based on annual precipitation (PPT) and near-surface air temperature (Los et al. 2006). This LAI time series has been calibrated against satellite-derived LAI for the period 1982–2000. For the same period we adopt atmospheric CO2 concentration from the Law Dome ice core (Etheridge et al. 1996) for 1951–58 and from Keeling et al. (2009) for 1959–2000. For SST, we adopt the extended reconstructed dataset of Smith et al. (2008) based on a reanalysis of ship and buoy measurements.

3. Results and discussion

a. Model validation

GENIE-SF reproduces latitudinal profiles of E and land NPP with a fair amount of success when comparing against observation-based estimates [r2 = 0.64–0.85 (p < 0.001); Fig. 1]. However, there is a tendency to underestimate land E (by 22% on average), the shortfall most notable in the tropics. Furthermore, observation-based land NPP within the tropics shows a smoother spatial distribution compared to the model. The uneven model NPP may relate to soil moisture stress, suggesting that GENIE-SF, like many other fully coupled models (Dai 2006), may have difficulty reproducing tropical rainfall patterns [see comparison against the Global Precipitation Climatology Project (GPCP) below]. While the error bars in Fig. 1 take some account of observational uncertainties, considerable bias may still be present in the measurements. For example, Alton et al. (2009) found that a high spatial resolution (1°) LSM underestimated both land E and continental discharge over the zonal range 35°–70° when compared against separate observational datasets for E and runoff (by 14% and 27%, respectively), suggesting substantial inconsistencies between regional, observational datasets. On average, ocean E is underestimated by 15% compared to satellite-based estimates (Yu and Weller 2007), although E over the tropical ocean does not appear to be underestimated (Fig. 1).

Fig. 1.

(a) Annual NPP for land, (b) annual E for land, and (c) annual E for ocean, all predicted with the GENIE-SF biospheric model and averaged over 5.6° latitudinal bands. Observation-based estimates (Obs) for NPP are derived from EMDI measurements using class A and class B sites to define the uncertainty range. Observation-based estimates of land E are defined according to the range given by Baumgartner and Reichel (1975) and Henning (1989). Observation-based estimates of ocean E derive from satellite data (Yu and Weller 2007) using the dispersion of zonal estimates in Chou et al. (2003) to quantify the measurement uncertainty. For all coefficients of determination r2, p < 0.001 and n = 24.

Fig. 1.

(a) Annual NPP for land, (b) annual E for land, and (c) annual E for ocean, all predicted with the GENIE-SF biospheric model and averaged over 5.6° latitudinal bands. Observation-based estimates (Obs) for NPP are derived from EMDI measurements using class A and class B sites to define the uncertainty range. Observation-based estimates of land E are defined according to the range given by Baumgartner and Reichel (1975) and Henning (1989). Observation-based estimates of ocean E derive from satellite data (Yu and Weller 2007) using the dispersion of zonal estimates in Chou et al. (2003) to quantify the measurement uncertainty. For all coefficients of determination r2, p < 0.001 and n = 24.

Integrated land NPP (56 Gt yr−1; Table 3) lies right in the middle of the range predicted by 16 other global LSMs (55 ± 11 Gt yr−1; Cramer et al. 1999).2 Since most of these LSMs are run at a higher spatial resolution (~1°) compared to GENIE-SF, this result confirms that the coarse resolution used by the biospheric model is sufficient for estimating global and regional fluxes. Indeed, controlled experiments in spatial degradation indicate a ≤5% loss in the accuracy of predicted zonal land fluxes if a spatial resolution of 5°–6°, rather than 1°, is adopted (Müller and Lucht 2007). This precision is tolerable within the errors indicated below by the model validation (≤25%). Integrated ocean NPP (52 Gt yr−1) lies midway between the values of 48.5 and 56.5 Gt yr−1 estimated from CZCS (Field et al. 1998) and SeaWiFS (Behrenfeld et al. 2001) satellite measurements.

Table 3.

Annual totals of E, PPT, and NPP, divided into land and ocean components, estimated with GENIE-SF for the biosphere. Note that 1 Gt = 1012 kg.

Annual totals of E, PPT, and NPP, divided into land and ocean components, estimated with GENIE-SF for the biosphere. Note that 1 Gt = 1012 kg.
Annual totals of E, PPT, and NPP, divided into land and ocean components, estimated with GENIE-SF for the biosphere. Note that 1 Gt = 1012 kg.

Global runoff is estimated as 5.6 × 104 Gt yr−1, which is significantly larger than recent estimates based on observed river discharge (3.7–4.2 × 104 Gt yr−1; Labat et al. 2004; Trenberth et al. 2007; Schlosser and Houser 2007; Dai et al. 2009). A direct zonal comparison between predicted and observed river discharge is confounded by the coarse land surface pixel within the simulation (5.6°), which may straddle more than one watershed. However, a tendency to overestimate runoff is apparent in Fig. 2 with, for example, total continental discharge underestimated by 25% between −15° and +15°. Over the same latitude band, E is underestimated by 26% (Fig. 1), confirming that too little moisture evaporates from the land surface within the model.

Fig. 2.

Simulated (GENIE) and observed (Dai et al. 2009) annual river discharge (km3 yr−1) averaged over 5.6° latitudinal bands. Model runoff has been linked to continental discharge via the global river routing map of Oki and Sud (1998). The observational record is averaged over a longer time span (1948–2004) compared to the simulations (1990–2002), but observed annual discharge has changed by only 1%–2% over the latter half of the twentieth century (Dai et al. 2009). The observed peak at latitude −3°, owing to the Amazon river, is reproduced poorly by the model owing to the coarse pixel size (5.6°), which may straddle more than one watershed.

Fig. 2.

Simulated (GENIE) and observed (Dai et al. 2009) annual river discharge (km3 yr−1) averaged over 5.6° latitudinal bands. Model runoff has been linked to continental discharge via the global river routing map of Oki and Sud (1998). The observational record is averaged over a longer time span (1948–2004) compared to the simulations (1990–2002), but observed annual discharge has changed by only 1%–2% over the latter half of the twentieth century (Dai et al. 2009). The observed peak at latitude −3°, owing to the Amazon river, is reproduced poorly by the model owing to the coarse pixel size (5.6°), which may straddle more than one watershed.

Total land PPT and total ocean PPT lie within 10% of the corresponding values that we derive, respectively, from the reanalysis climatologies of the Global Soil Wetness Project (1.0 × 105 Gt yr−1; Dirmeyer et al. 1999) and the Princeton dataset (3.0 × 105 Gt yr−1; Sheffield et al. 2006). Our prediction of annually averaged latent heat release from the tropical ocean (discussed below) is within 10% of the value predicted by Betts et al. (1986) using a thermodynamic model. However, annual global PPT and E over ocean appear to be systematically underestimated, by 20%–25% and 15%, respectively, when compared against recent satellite-derived estimates (Fig. 1c). We note, however, that significant bias still exists for measurements of PPT over both land and ocean (~10% and ~20%, respectively; Schlosser and Houser 2007). The IGCM appears to reproduce fairly well the biospheric distribution of annual PPT when compared against the GPCP (Adler et al. 2003) product, the latter combining microwave and infrared satellite measurements with surface gauge observations. This includes the intense band of precipitation associated with the intertropical convergence zone (ITCZ; latitude −30° to +30°; Fig. 3). However, there is a tendency to underestimate global precipitation toward the poles especially during the austral summer (Fig. 4).

Fig. 3.

Global rates of (top) precipitation and (bottom) evaporation estimated with GENIE-SF (mm yr−1), corresponding to (left) the austral summer [December–February DJF)] and (right) the boreal summer [June–August (JJA)].

Fig. 3.

Global rates of (top) precipitation and (bottom) evaporation estimated with GENIE-SF (mm yr−1), corresponding to (left) the austral summer [December–February DJF)] and (right) the boreal summer [June–August (JJA)].

Fig. 4.

Latitudinal profiles of seasonal biospheric (land + ocean) precipitation rate (mm yr−1). Predictions with GENIE are compared to observation-based estimates (Obs) from GPCP (Adler et al. 2003). (a) Austral summer (DJF) and (b) boreal summer (JJA). For all r2, p < 0.001 and n = 32.

Fig. 4.

Latitudinal profiles of seasonal biospheric (land + ocean) precipitation rate (mm yr−1). Predictions with GENIE are compared to observation-based estimates (Obs) from GPCP (Adler et al. 2003). (a) Austral summer (DJF) and (b) boreal summer (JJA). For all r2, p < 0.001 and n = 32.

Our validation focuses on observations over quite broad latitude zones since we wish to use model output over large spatial scales (e.g., for the tropics or globally). At smaller scales we must be much more cautious about drawing conclusions, especially for the hydrological components. A pixel-by-pixel comparison of GPCP and GENIE (Fig. 5) reveals a fair reproduction of many of the global PPT patterns but, for example, significant underestimation of annual PPT in the western Amazon. In many cases the GPCP-GENIE discrepancies exceed the uncertainties associated with the observation-based dataset. For example, we find an average difference of no more than 5% between the Global Soil Wetness Project and GPCP for annual PPT except over the zone 40°–70° where the difference is 20%. Sharma et al. (2007) recognize the problems of predicting accurate PPT in individual GCM grid cells. In their study of the Ping River Basin in northern Thailand, their ECHAM model underestimates gridcell PPT by 1000 mm yr−1.

Fig. 5.

Annual PPT (mm yr−1) (top) predicted by GENIE and (bottom) estimated by the GPCP product. In both cases the adopted pixel size is 5.6°.

Fig. 5.

Annual PPT (mm yr−1) (top) predicted by GENIE and (bottom) estimated by the GPCP product. In both cases the adopted pixel size is 5.6°.

Uncertainties in predicted annual flux exchange over land and ocean are difficult to quantify since the “truth” often consists of sparse, local measurements, which are then extrapolated globally. When we scale the EMDI profile in Fig. 1, by accounting for the landmass at each latitude, we obtain an observation-based estimate for global land NPP of 58 ± 9 Gt yr−1. This deviates by no more than 20% from the simulation (56 Gt yr−1). Global land NPP simulated with 16 different LSMs exhibits a standard deviation of 20% from the mean (Cramer et al. 1999). The ocean model adopted in the present investigation reproduces primary productivity measured within ocean samples to within 30% (Behrenfeld et al. 2006). As demonstrated above, GENIE-SF has a 15%–22% bias against observation-based estimates of latitudinal E and a comparable bias against continental discharge. Conservatively, therefore, we assume a 25% error in our estimated total annual fluxes for both water and carbon. We emphasize, however, that many of our results are based on the relative distribution of fluxes, which are much less sensitive to the modeling errors. Across the last 10 years of the 13-yr GENIE-SF simulation, predicted values of total land NPP and total land E exhibit an interannual variability (IAV) of only about 1%. The IAV for the separate north, south, and tropical regions discussed below is somewhat higher (2%–3%), but the GENIE-SF simulation is likely to underestimate the true IAV owing to the use of average values for monthly SST and LAI. Observational data across the El Niño–Southern Oscillation suggest an IAV of no more than 4% for both total land NPP (Peylin et al. 2005) and total ocean NPP (Behrenfeld et al. 2001; Behrenfeld et al. 2006).

b. Global analysis

1) Water exchange

Our simulation shows the striking difference in the seasonal release of water vapor to the atmosphere from northern (latitude > +23.5°) land points compared to ocean at the same latitude (Fig. 3). In the boreal summer, increased leaf area and high surface temperatures permit higher E rates over the land compared to the ocean despite the high surface conductance associated with the latter. In the boreal winter, the E rate drops almost to zero for northern land points but, over the northern ocean, it increases by a factor of 2.4 compared to the boreal summer. Indeed, 21% of total ocean E during the boreal winter originates from the northern ocean (Fig. 6a). Global ocean E, which is dominated in all seasons by the tropical ocean (Fig. 6), is higher than global land E by a factor of 6.7 ± 1.7 despite the large evaporation rates of the northern continents during the boreal summer (Table 3). This annual ocean–land E ratio is somewhat higher than, but nevertheless overlaps with, the corresponding estimate of Mehta et al. (2005) who combined observational and reanalysis datasets with results from a forced (i.e., noncoupled) LSM (3.0–5.7).

Fig. 6.

Evaporation ΣE and net primary productivity ΣNPP integrated over land and ocean, as predicted by GENIE-SF (Gt). Totals are shown separately for DJF and JJA and for southern (S; latitude < −23.5°), tropical (T; −23.5 ≤ latitude ≤ +23.5°), and northern (N; latitude > +23.5°) zones.

Fig. 6.

Evaporation ΣE and net primary productivity ΣNPP integrated over land and ocean, as predicted by GENIE-SF (Gt). Totals are shown separately for DJF and JJA and for southern (S; latitude < −23.5°), tropical (T; −23.5 ≤ latitude ≤ +23.5°), and northern (N; latitude > +23.5°) zones.

In Fig. 7, the seasonal zonal peak of ocean E is offset from the warmest tropical waters associated with maximum PPT and the ITCZ. This latitudinal inversion is fairly well established (e.g., Yu 2007) and has been attributed to reduced SW irradiance under deep convective cloud and lower surface wind speed over the warmer water (Seager et al. 2003). Our simulation provides some insights into the sources of land PPT. In Fig. 7, the distributions of water vapor column and E rate are much more closely matched over land compared to ocean (r2 = 0.72 and r2 = 0.46, respectively, with p < 0.001). This suggests a strong local source for air moisture over the continents. Over ocean, the difference in total annual E and PPT (i.e., E − PPT) is +5.6 (±1.4) × 104 Gt yr−1 (Table 3), and this excess supplies the advection of moisture from the ocean over the land (e.g., Chahine 1992). However, this advection is only half of the annual PPT over land. Thus, global land PPT consists of 50% ± 13% of moisture that has previously evaporated from the land surface and has been “recycled.”

Fig. 7.

Latitudinal profiles of PPT, E, and atmospheric water vapor column Q predicted using GENIE-SF. Graphs are shown separately for (top) land, (middle) ocean, and (bottom) all biospheric (land + ocean) points. Profiles are given for (left) the austral summer (DJF) and (right) the boreal summer (JJA). The key for all profiles is given in the bottom-left panel.

Fig. 7.

Latitudinal profiles of PPT, E, and atmospheric water vapor column Q predicted using GENIE-SF. Graphs are shown separately for (top) land, (middle) ocean, and (bottom) all biospheric (land + ocean) points. Profiles are given for (left) the austral summer (DJF) and (right) the boreal summer (JJA). The key for all profiles is given in the bottom-left panel.

Previous estimates of moisture recycling over land are lower than the current study (10%–35%; Eltahir and Bras 1996; Trenberth 1999), but this is expected given that they apply to smaller spatial scales (1000–2500 km). For the northern landmasses, the present investigation indicates that moisture recycling is high during the boreal summer (77% ± 18%) and low (5% ± 2%) during the boreal winter. A similar conclusion is drawn by Trenberth et al. (2007) who force an LSM with the GPCP precipitation dataset. The same pattern in seasonal recycling is also noted by Dirmeyer et al. (2009) although their recycling ratio, defined over 1° (≃100 km) cells, is much smaller (≤0.3). These authors define the recycling ratio as the fraction of PPT that originates from moisture evaporated over the same area. Under this definition, our regional recycling values are upper limits since the back trajectory of moisture advection, as carried out by Dirmeyer et al., is beyond the limits of the current study. The recycling, as we have defined it (E/PPT), simply reveals whether PPT is sufficiently large to account for E. For the northern landmasses some of summer PPT is almost certainly lost as surface runoff. Therefore, to account for the relatively high summer recycling, a significant fraction of winter moisture must be stored in the model soil layer for releases as E during the growing season. A few site and regional studies confirm that the land acts as a reservoir in this way, storing winter and spring PPT, which is returned to the atmosphere as E during the growing season (Williams et al. 2004; Yu et al. 2004).

Dividing the air moisture content by the E rate (Fig. 7) yields a replenishment time scale (Trenberth 1999) that is significantly longer over the landmasses (24 ± 6 days) compared to over the oceans (11 ± 3 days). Therefore, E rates, PPT rates, and recycling rates through the atmosphere are all relatively high over the ocean.

2) Net productivity

In a manner analogous to E, the amplitude of seasonal net productivity rate is larger over land compared to over ocean (Fig. 8). This is especially true of the northern latitudes, where the ratio of (boreal) winter-to-summer productivity rates is 0.01 for land but 0.42 for ocean. The large amplitude for land is such that 26% of annually integrated land NPP originates from photosynthesis within the northern continents during the 3-month period June–August (JJA) (Fig. 6). The southern (latitude < −23.5°) continents exhibit a somewhat weaker seasonal change compared to the northern landmasses (Fig. 8).

Fig. 8.

Seasonal NPP simulated with GENIE-SF for (top) land and (bottom) ocean during DJF and JJA.

Fig. 8.

Seasonal NPP simulated with GENIE-SF for (top) land and (bottom) ocean during DJF and JJA.

Spatially, productivity rates are also more uniform within the ocean. Thus, over the latitude range −50° to +75°, the seasonal productivity rate possesses a standard deviation of 97% from the mean for land but only 55% from the mean for ocean (Fig. 8). Averaging the absolute gradient |d(NPP)/d(Lat)| over the same latitude range, yields 0.19 and 0.034 kg m−2 yr−1 (5.6°)−1 for land and ocean, respectively. Dividing this absolute gradient by the corresponding mean yields 0.43 and 0.20 kg m−2 yr−1 (5.6°)−1, respectively. Therefore, in any given period (boreal summer or winter), the latitudinal gradient relative to the mean is twice as large for land compared to ocean. The relative spatial uniformity of ocean productivity is evident from in situ sampling of phytoplankton (Lieth 1975). Given that ocean productivity models, such as the VGPM within GENIE-SF, are constructed on the basis of such in situ measurements it is not too surprising that our results indicate a relative uniformity to ocean productivity over large spatial scales. S. Running and W. Esaias (2008, personal communication) also note the uniformity of ocean productivity compared to the landmasses on the basis of MODIS satellite images.

The tropical landmasses possess higher productivity rates compared to the tropical ocean, although the tropical ocean is so large that NPP integrated over the tropics is divided fairly equally between land and ocean (Fig. 6). Low nutrient status is often invoked to explain the low productivity rates in tropical waters compared to the midlatitudes (Behrenfeld et al. 2006). NPP integrated over the southern biosphere (latitude < −23.5°) is dominated throughout the seasons by the ocean, owing to the relatively small area of vegetated land within this zone.

Land and ocean contribute, respectively, 52% ± 9% and 48% ± 9% to an estimated annual biospheric productivity of 108 ± 27 Gt yr−1 (Table 3). The same proportions, within 2%–3%, are found by Behrenfeld et al. (2001) who used SeaWiFS satellite measurements for both ocean chlorophyll and the normalized difference vegetation index (NDVI) on land. These authors adopt the same ocean productivity model as the current study. However, they employ an ecological model for land NPP based on monthly solar irradiance and constant LUE. Previous studies demonstrate that, after suitable calibration, this constant-LUE algorithm may be sufficient for annual totals of NPP but that it may be less suitable on shorter (e.g., seasonal) time scales (Medlyn et al. 2003; McCallum et al. 2009; Alton and Bodin 2010). In the current study, this algorithm is replaced by a state-of-the-art multilayer light interception model, which calculates canopy photosynthesis under direct and diffuse sunlight at hourly time steps.

c. Temporal change (50-yr simulation)

1) Water exchange

Our modeled global water balance does not change significantly over the 50-yr simulation period (1951–2000). Simulated global runoff is consistent with the observed change in continental discharge over the latter half of the twentieth century, which is less than 1% (Dai et al. 2009). Simulations by some previous authors predict significant increases in global runoff (6%–8%) over the course of the entire twentieth century (Betts et al. 2007; Gerten et al. 2008). However, these trends are both model and forcing dependent. Furthermore, these previous simulations attempted to reproduce the 3% increase in continental discharge estimated by Labat et al. (2004). This earlier estimate is based on no more than 50%, rather than 73% (Dai et al. 2009), coverage of global runoff.

2) Net productivity

Over the 50-yr simulation period we predict a 16% increase in land NPP (Fig. 9). We attribute this trend to the “fertilization” of gross primary productivity (GPP) owing to increased atmospheric CO2 concentration (Fig. 9). Averaging the seasonal LAI over the period 1970–79, and applying this average phenology to the whole 50-yr simulation, reduces the NPP trend by only 1%. Thus, the increase in land productivity is relatively insensitive to any trend inherent in the reconstructed LAI. The decadal trend simulated for land NPP is also highly significant compared to the observed IAV across the ENSO cycle (≤4%; Behrenfeld et al. 2001; Peylin et al. 2005). Temporal change in land NPP has been examined by several previous authors. Cao et al. (2004) estimate a 10% increase in global land NPP over the period 1981–2000 by combining satellite (AVHRR) reflectance data with a LUE model. The increase recorded in that study correlates strongly with atmospheric CO2 concentration, implying that land NPP would rise by 20% over our 50-yr simulation period. On the same basis, simulations with a LUE approach (Nemani et al. 2003) and a more complex, process-based model (Piao et al. 2009) are both consistent with a 14%–16% increase over the 50-yr period. Both are very close to our estimate. Validating the model response to increased atmospheric CO2 concentration is difficult because free-air CO2 enrichment (FACE) field experiments do not provide an estimate of NPP change. However, the average FACE diurnal photosynthetic carbon assimilation increases by 28% for a 125-ppm increase in atmospheric CO2 concentration (Ainsworth and Long 2004). If plant respiration increases pro rata, then our 50-yr simulation (16% NPP increase for a 60-ppm increase) is approximately consistent with FACE. In principle, increased NPP may lead to an increase in both the amount and the cover fractions of above-ground biomass causing a feedback on NPP through a change in either primary productivity or autotrophic respiration. However, the LSM within GENIE is static with respect to carbon stocks. Although dynamic vegetation models are useful for decadal simulations of this type, they contain a large number of uncertainties concerning carbon allocation and vegetation cover change (Moorcroft 2006; Purves and Pacala 2008), which lie beyond the scope of the current study.

Fig. 9.

ΣNPP for land and ocean simulated with GENIE-SF for 1951–2000. Annual values have been averaged over a 10-yr period.

Fig. 9.

ΣNPP for land and ocean simulated with GENIE-SF for 1951–2000. Annual values have been averaged over a 10-yr period.

In contrast to land productivity, ocean NPP is stable (within 1%) despite an increase in average global SST. The temperature dependency within the VGPM reaches a peak at approximately 20°C (Behrenfeld and Falkowski 1997). Thus, increased productivity over the simulation period in the mid- and northern latitudes is offset by a simultaneous reduction in warmer (tropical) water productivity. Our estimate for the ocean neglects any change in chlorophyll concentration. Gregg and Conkright (2002) record a 6% decrease in marine chlorophyll concentration between 1979 and 2000 by comparing CZCS and SeaWiFS data, while Behrenfeld et al. (2006) record a decrease in ocean NPP of approximately 1% decade−1. Thus, the negative feedback to increased atmospheric CO2 concentration, provided by increased NPP on land, may be at least partially offset by reduced marine productivity once the changes in ocean chlorophyll are accounted for.

Our main simulation period used for comparing zonal and seasonal fluxes (1990–2002) necessarily represents a compromise since the datasets used to force the model (LAI and ocean chlorophyll) and the datasets used for validation (EMDI, GPCP, etc.) span different periods. Nevertheless, our results appear relatively insensitive to this temporal offset. For example, EMDI measurements used to validate model NPP stem principally from 1960–90. Our 50-yr simulation suggests a 6% change between 1975 (average EMDI year) and 1996 (average year in the main simulation), which is modest compared to the 25% error assumed in the model output. Similarly, the change in ocean productivity between 1996 and 2005 (average chlorophyll year from Aqua) is unlikely to exceed a few percent on the basis of both our 50-yr simulation (stability within 1%) and the observed decrease in marine chlorophyll (≤3% decade−1; Gregg and Conkright 2002; Behrenfeld et al. 2006). As discussed above, the changes in historical global river discharge are small (<1% decade−1) compared to the assumed model error (25%). Furthermore, zonal PPT has generally changed by no more than 1% decade−1 over the course of the twentieth century (Huntington 2006).

4. Summary and conclusions

We have used a new version of the GENIE biospheric model (GENIE-SF) to compare carbon and water fluxes over land and ocean. The ocean and land components are driven by measurements of LAI and marine chlorophyll concentration acquired with MODIS sensors aboard the Terra and Aqua satellites. The land component contains state-of-the-art representations of multilayer light interception and canopy photosynthesis. We order our conclusions according to the scientific objectives enumerated in the introduction:

  1. Division of biospheric NPP: annual biospheric NPP is estimated at 108 ± 27 Gt yr−1 and is split almost equally between land (56 Gt yr−1) and ocean (52 Gt yr−1). For the latter half of the twentieth century, a small increase in the land contribution to biospheric NPP is predicted (from 48% to 52%) mainly owing to increased atmospheric CO2 concentration.

  2. Division of biospheric E: the oceanic contribution to biospheric E is 6.7 ± 1.7 times greater than that of the land surface.

  3. Spatial and temporal variation in NPP and E: for both NPP and E, the land is subject to a more pronounced seasonal cycle compared to the ocean. This is especially true for the northern latitudes (>+23.5°) where the ratio of (boreal) winter-to-summer productivity rates is 0.01 for land but 0.42 for ocean. In the boreal winter, the E rate drops almost to zero for northern land points but increases by a factor of 2.4 over the northern ocean. The northern continents contribute just over a quarter of annually integrated land NPP during the 3-month period June–August. Spatially, primary productivity rates are also more uniform within the ocean compared to on land. Indeed, in any given period (boreal summer or winter), the latitudinal gradient in seasonal NPP is twice as large for land compared to ocean.

  4. Estimates of the earth’s fundamental water cycle using a biospheric model: the advection of moisture from the ocean over the land is estimated at 5.6 (±1.4) × 104 Gt yr−1, a value which suggests that global land PPT consists 50% ± 13% of moisture that has previously evaporated from the land surface and been “recycled.” During the summer, the ratio E/PPT is no more than 77% over the northern landmasses, which suggests that seasonal recycling may be important in those regions.

Acknowledgments

Gethin Williams at Bristol University is thanked for his assistance in installing GENIE. We thank Les Hook and Dr. R. Olson for the revised set of NPP values from the EMDI archive. We are grateful to Sietse Los for supplying his reconstructed LAI time series and useful comments on the manuscript.

APPENDIX

Modeling Technical Description

a. Land component

The LSM is based on MOSES, for which detailed equations are already given by Cox et al. (1999). MOSES is forced by the following meteorological variables: downwelling shortwave radiation, downwelling longwave radiation, precipitation, air temperature, wind speed, air humidity, and pressure. The energy calculation central to the model is based on a Penman–Monteith approach (Monteith 1965), which ensures that the downwelling fluxes of shortwave (SW) and longwave (LW) radiation are balanced by the outgoing fluxes of reflected shortwave radiation, sensible heat H, latent heat (LE), radiant thermal energy, and conduction into the ground (all in W m−2). Thus

 
formula

where σ is the Stefan–Boltzmann constant, Ts is the surface temperature, and α is the surface albedo. Energy balance is conducted separately for each of 10 categories of surface cover (tiles) constituting each land point. The 10 tiles consist of 6 PFTs (tropical broadleaf forest, extratropical broadleaf forest, needleleaf forest, C3 grass, C4 grass, and shrub) and 4 nonvegetation tiles (urban, water, ice, and bare ground). Cover type for each land point is taken from the International Geosphere-Biosphere Project classification (IGBP 1992; Hansen and Reed 2000) simplified for use with MOSES (Cox et al. 1999; Lawrence and Slingo 2004). Fluxes from each tile are combined to yield a gridbox mean value for each land point. Land surface albedo is derived using the two-stream radiative calculation of Sellers et al. (1996).

Leaf photosynthesis is estimated using a colimitation biochemical model (Collatz et al. 1991). Light interception within the canopy is calculated at multiple heights taking account of sunfleck penetration and diffuse sunlight. A sunfleck reaches a given canopy layer if a randomly generated number, between 0 and 1, is less than or equal to P, where

 
formula

and LAIc is the cumulative LAI lying above the leaf layer, θs is the solar zenith angle, and kext is the light extinction coefficient (Norman 1992; Jones 1992). Assuming a spherical leaf angle distribution (Campbell and Norman 1998), kext = 0.5. Diffuse light in the canopy follows from the two-stream formulation, the equations for which are given by Sellers et al. (1996). At any given time step, the fraction of diffuse sunlight incident at the top of the canopy is derived using the ratio of observed surface irradiance and top-of-atmosphere irradiance, as given by Roderick et al. (2001).

Leaf photosynthesis for the C3 and C4 pathways are derived using the colimitation model of Collatz et al. (1991), which is conceptually similar to the biochemical model of Farquhar et al. (1980). The leaf photosynthetic rate Al (μmol m−2 s−1) is the smoothed minimum of three limits: the photosynthetic rate due to incident light JPAR; photosynthetic capacity due to the concentration and chemical activity of Ribulose-1, 5-bisphosphate carboxylase/oxygenase (i.e., Rubisco) Jr; and the photosynthetic rate based on the ability of the leaf to export the products of photosynthesis Je. Thus,

 
formula

where QE is the quantum efficiency and IL (μmol m−2 s−1) is the leaf photosynthetically active radiation (PAR) irradiance. Also, ci (mol mol−1) and c0 (mol mol−1) are, respectively, the CO2 concentration internal to the leaf and the photorespiratory compensatory point.

The Rubisco-limited rate of leaf photosynthesis is given by

 
formula

where Kc (mol mol−1) and Ko (mol mol−1) are the Michaelis constants determining the competing rates of carboxylation and oxygenation, and Oa is the oxygen concentration (mol mol−1).

The parameter Vm (μmol m−2 s−1) describes the chemical activity of Rubisco at leaf temperature TL (°C):

 
formula

where Q10 is a dimensionless coefficient for leaf respiration, Tl and Th are the inhibition temperatures (°C) for photosynthesis, and Vcmax (μmol m−2 s−1) is the leaf photosynthetic capacity at the top of the canopy. Both Vcmax and QE decline exponentially according to the nitrogen allocation coefficient krub and cumulative LAI (Hirose and Werger 1987).

Finally, the export-limited rate of leaf photosynthesis is determined by

 
formula

Leaf respiration RL is set to Fd × Vm (Collatz et al. 1991). Total plant respiration RP consists of respiration for maintenance RPM and for growth RPG. The former includes root respiration based on a Q10 relationship with soil temperature (Law et al. 1999). Growth respiration is a prescribed fraction Rgrow of the difference between gross primary productivity (GPP) and maintenance respiration (Ryan 1991). Thus

 
formula

Leaf stomatal conductance gl (mol m−2 s−1) depends on humidity deficit at the leaf surface according to a modified Leuning relation (Cox et al. 1998). Leaf photosynthesis is dependent on the soil moisture content within a 2-m soil layer according to a linear ramp function (FSMC). Thus

 
formula

where V is the fractional soil moisture content with respect to saturation and Vcrit and Vwilt correspond, respectively, to the fractional soil moisture content at the critical and wilting points for leaf photosynthesis. The key parameters for the land component, including those for soil representation, are given in Table 2.

b. Ocean component

The ocean model, which has a monthly time step, depends on SST and surface chlorophyll concentration. Productivity is calculated according to the Vertically Generalized Production Model (VGPM; Behrenfeld and Falkowski 1997) as follows:

 
formula

where PPeu is the daily carbon fixation within the euphotic water column (mg C m−2), is the optimal rate of carbon fixation within the water column (mg C mg−1 Chl h−1), E0 is the sea surface daily PAR (mol quanta m−2 day−1), Csat is the monthly satellite-measured surface chlorophyll concentration (mg Chl m−3), Zeu is the physical depth of the euphotic zone (m), and Dirr is the daily photoperiod (h). Note that is a polynomial function of SST, which is described and tested in detail by Behrenfeld and Falkowski (1997). Also, E0 and Dirr are supplied to the ocean component from the atmospheric module within GENIE-SF, and Zeu is calculated from Csat using relations developed and tested by Morel and Berthon (1989).

REFERENCES

REFERENCES
Adler
,
R.
, and
Coauthors
,
2003
:
The version 2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present)
.
J. Hydrometeor.
,
4
,
1147
1167
.
Ainsworth
,
E.
, and
S.
Long
,
2004
:
What have we learned from 15 years of free-air CO2 enrichment (FACE)? A meta-analytic review of the responses of photosynthesis, canopy properties and plant production to rising CO2
.
New Phytol.
,
165
,
351
372
.
Alton
,
P.
, and
P.
Bodin
,
2010
:
A comparative study of a multilayer and a productivity (light-use) efficiency land-surface model over different temporal scales
.
Agric. For. Meteor.
,
150
,
182
195
.
Alton
,
P.
,
P.
North
,
J.
Kaduk
, and
S.
Los
,
2005
:
Radiative transfer modeling of direct and diffuse sunlight in a Siberian pine forest
.
J. Geophys. Res.
,
110
,
D23209
,
doi:10.1029/2005JD006060
.
Alton
,
P.
,
P.
North
, and
S.
Los
,
2007
:
The impact of diffuse sunlight on canopy light-use efficiency, gross photosynthetic product and net ecosystem exchange in three forest biomes
.
Global Change Biol.
,
13
,
776
787
.
Alton
,
P.
,
R.
Fisher
,
S.
Los
, and
M.
Williams
,
2009
:
Simulations of global evapotranspiration using semiempirical and mechanistic schemes of plant hydrology
.
Global Biogeochem. Cycles
,
23
,
GB4023
,
doi:10.1029/2009GB003540
.
Annan
,
J.
,
D.
Lunt
,
J.
Hargreaves
, and
P.
Valdes
,
2005
:
Parameter estimation in an atmospheric GCM using the Ensemble Kalman Filter
.
Nonlinear Processes Geophys.
,
12
,
363
371
.
Baumgartner
,
A.
, and
E.
Reichel
,
1975
:
Die Weltwasserbilanz Oldenburg
.
Verlag, 179 pp
.
Behrenfeld
,
M.
, and
P.
Falkowski
,
1997
:
Photosynthetic rates derived from satellite-based chlorophyll concentration
.
Limnol. Oceanogr.
,
42
,
1
20
.
Behrenfeld
,
M.
,
J.
Randerson
, and
C.
McClain
,
2001
:
Biospheric primary production during and ENSO transition
.
Science
,
291
,
2594
2597
.
Behrenfeld
,
M.
,
K.
Worthington
,
R.
Sherrell
,
F.
Chavez
,
P.
Strutton
,
M.
McPhaden
, and
D.
Shea
,
2006
:
Controls on tropical Pacific Ocean productivity revealed through nutrient stress diagnostics
.
Nature
,
442
,
1025
1028
.
Betts
,
A.
,
J.
Ball
,
A.
Beljaars
,
M.
Miller
, and
P.
Veterbo
,
1986
:
The land surface–atmosphere interaction: A review based on observational and global modelling perspectives
.
J. Geophys. Res.
,
101
,
7209
7225
.
Betts
,
R.
, and
Coauthors
,
2007
:
Projected increase in continental runoff due to plant responses to increasing carbon dioxide
.
Nature
,
448
,
1037
1041
.
Brovkin
,
V.
,
S.
Levis
,
M.
Loutre
,
M.
Crucifix
,
M.
Claussen
,
A.
Ganopolski
,
C.
Kubatzki
, and
V.
Petoukhov
,
2003
:
Stability analysis of the climate-vegetation system in the northern high latitudes
.
Climate Change
,
57
,
119
138
.
Campbell
,
B.
, and
J.
Norman
,
1998
:
Environmental Biophysics
.
2nd ed. Springer-Verlag, 286 pp
.
Cao
,
M.
,
S.
Prince
,
J.
Small
, and
S.
Goetz
,
2004
:
Remotely sensed interannual variations and trends in terrestrial net primary productivity 1981–2000
.
Ecosystems
,
7
,
233
242
.
Carswell
,
F.
,
P.
Meir
, and
E.
Wandelli
,
2000
:
Photosynthetic capacity in a central Amazonian rain forest
.
Tree Physiol.
,
20
,
179
186
.
Chahine
,
M.
,
1992
:
The hydrological cycle and its influence on climate
.
Nature
,
359
,
373
380
.
Chou
,
S.
,
E.
Nelkin
,
J.
Ardizzone
,
R.
Atlas
, and
C.
Shie
,
2003
:
Surface turbulent heat and momentum fluxes over global oceans based on the Goddard satellite retrievals, version 2 (GSSTF2)
.
J. Climate
,
16
,
3256
3273
.
Claussen
,
M.
,
L.
Mysak
, and
A.
Weaver
,
2002
:
Earth system models of intermediate complexity: Closing the gap in the spectrum of climate system models
.
Climate Dyn.
,
18
,
579
586
.
Coe
,
M.
,
2000
:
Modeling terrestrial hydrological systems at the continental scale: Testing the accuracy of an atmospheric GCM
.
J. Climate
,
13
,
686
704
.
Collatz
,
C.
,
J.
Ball
,
C.
Grivet
, and
J.
Berry
,
1991
:
Physiological and environmental regulation of stomatal conductance, photosynthesis and transpiration: A model that includes laminar boundary layer
.
Agric. For. Meteor.
,
54
,
107
136
.
Cox
,
P.
,
C.
Huntingford
, and
R.
Harding
,
1998
:
A canopy conductance and photosynthesis model for use in a GCM land surface scheme
.
J. Hydrol.
,
212–213
,
79
94
.
Cox
,
P.
,
R.
Betts
,
C.
Bunton
,
R.
Essery
,
P.
Rowntree
, and
J.
Smith
,
1999
:
The impact of new land surface physics on the GCM simulation of climate and climate sensitivity
.
Climate Dyn.
,
15
,
183
203
.
Cramer
,
W.
,
D.
Kicklighter
, and
A.
Bondeau
,
1999
:
Comparing global models of terrestrial net primary productivity (NPP): overview and key results
.
Global Change Biol.
,
5
(
Suppl. 1
),
1
15
.
Dai
,
A.
,
2006
:
Precipitation characteristics in eighteen coupled climate models
.
J. Climate
,
19
,
4605
4630
.
Dai
,
A.
,
T.
Qian
, and
K.
Trenberth
,
2009
:
Changes in continental freshwater discharge from 1948 to 2004
.
J. Climate
,
22
,
2773
2792
.
Dirmeyer
,
P.
,
A.
Dolman
, and
N.
Sato
,
1999
:
The global soil wetness project: A pilot project for global land surface modeling and validation
.
Bull. Amer. Meteor. Soc.
,
80
,
851
878
.
Dirmeyer
,
P.
,
A.
Schlosser
, and
K.
Brubaker
,
2009
:
Precipitation, recycling, and land memory: An integrated analysis
.
J. Hydrometeor.
,
10
,
278
288
.
Dutkiewicz
,
S.
,
A.
Sokolov
,
J.
Scott
, and
P.
Stone
,
2005
:
A three-dimensional ocean–sea ice–carbon cycle model and its coupling to a two-dimensional atmospheric model: Uses in climate change studies
.
MIT Joint Program on Science and Policy of Global Change Rep. 122, 1–51
.
Eltahir
,
E.
, and
R.
Bras
,
1996
:
Precipitation recycling
.
Rev. Geophys.
,
34
,
367
378
.
Esias
,
W.
,
M.
Abbott
, and
I.
Barton
,
1998
:
An overview of MODIS capabilities for ocean science observations
.
IEEE Trans. Geosci. Remote Sens.
,
36
,
1250
1265
.
Etheridge
,
D.
,
L.
Steele
,
R.
Langenfelds
,
R.
Francey
,
J.
Barnola
, and
V.
Morgan
,
1996
:
Natural and anthropogenic changes in atmospheric CO2 over the last 1000 years from air in Antarctic ice and firn
.
J. Geophys. Res.
,
101
,
4115
4128
.
Farquhar
,
G.
,
S.
von Caemmerer
, and
J.
Berry
,
1980
:
A biochemical model of photosynthetic CO2 assimilation in leaves of C3 species
.
Planta
,
149
,
78
90
.
Field
,
C.
,
J.
Behrenfeld
,
J.
Randerson
, and
P.
Falkowski
,
1998
:
Primary production of the biosphere: Integrating terrestrial and oceanic components
.
Science
,
281
,
237
240
.
Fisher
,
B.
,
2007
:
Statistical error decomposition of regional-scale climatological precipitation estimates from the Tropical Rainfall Measuring Mission (TRMM)
.
J. Appl. Meteor. Climatol.
,
46
,
791
813
.
Franz
,
B.
,
S.
Bailey
, and
R.
Eplee
,
2005
:
The continuity of ocean color measurements from SeaWiFS to MODIS
.
Earth Observing Systems X, J. J. Butler, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 5882), 304–316
.
Friend
,
A.
,
A.
Stevens
,
R.
Knox
, and
M.
Cannell
,
1997
:
A process-based, terrestrial biosphere model of ecosystem dynamics (Hybrid v3.0)
.
Ecol. Modell.
,
95
,
249
287
.
Gedney
,
N.
,
P.
Cox
,
H.
Douville
,
J.
Polcher
, and
P.
Valdes
,
2000
:
Characterizing GCM land surface schemes to understand their responses to climate change
.
J. Climate
,
13
,
3066
3079
.
Gerten
,
D.
,
S.
Rost
,
W.
von Bloh
, and
W.
Lucht
,
2008
:
Causes of change in 20th century global river discharge
.
Geophys. Res. Lett.
,
35
,
L20405
,
doi:10.1029/2008GL035258
.
Gregg
,
W.
, and
M.
Conkright
,
2002
:
Decadal changes in global ocean chlorophyll
.
Geophys. Res. Lett.
,
29
,
L014689
,
doi:10.1029/2002GL014689
.
Hansen
,
M.
, and
B.
Reed
,
2000
:
A comparison of the IGBP DISCover and University of Maryland 1 km global land cover products
.
Int. J. Remote Sens.
,
21
,
1365
1373
.
Henning
,
D.
,
1989
:
Atlas of the Surface Heat Balance of the Continents
.
Gebrueder Borntraeger, 402 pp
.
Hirose
,
T.
, and
M.
Werger
,
1987
:
Maximizing daily canopy photosynthesis with respect to leaf N allocation pattern in the canopy
.
Oecologia
,
72
,
520
526
.
Hollinger
,
D.
,
F.
Kelliher
,
J.
Byers
,
J.
Hunt
,
T.
McSeveny
, and
P.
Wier
,
1994
:
Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere
.
Ecology
,
75
,
134
150
.
Hoskins
,
B.
, and
A.
Simmons
,
1975
:
A multi-layer spectral model and the semi-implicit method
.
Quart. J. Roy. Meteor. Soc.
,
101
,
637
655
.
Houghton
,
R.
,
2007
:
Balancing the global carbon budget
.
Annu. Rev. Earth Planet. Sci.
,
35
,
313
347
.
Huang
,
J.
,
H.
van den Dool
, and
K.
Georgakakos
,
1996
:
Analysis of model calculated soil moisture over the United States (1931–1993) and applications to long-range temperature forecasts
.
J. Climate
,
9
,
1350
1362
.
Huntington
,
T.
,
2006
:
Evidence for intensification of the global water cycle: Review and synthesis
.
J. Hydrol.
,
319
,
83
95
.
IGBP
,
1992
:
Improved global data for land applications
.
IGBP Global Change Rep. 20, J. R. G. Townshend, Ed., International Geosphere Biosphere Programme, 75 pp
.
Jones
,
H.
,
1992
:
Plant and Microclimate
.
Cambridge University Press, 428 pp
.
Keeling
,
R.
,
S.
Piper
,
A.
Bollenbacher
, and
J.
Walker
,
2009
:
Atmospheric CO2 records from sites in the SIO air sampling network
.
Trends: A Compendium of Data on Global Change, Carbon Dioxide Information Analysis Center, 1–3
.
Knohl
,
A.
, and
D.
Baldocchi
,
2008
:
Effects of diffuse radiation on canopy gas exchange processes in a forest ecosystem
.
J. Geophys. Res.
,
113
,
G02023
,
doi:10.1029/2007JG000663
.
Knorr
,
W.
, and
M.
Heimann
,
2001
:
Uncertainties in global terrestrial biosphere modeling, Part I: A comprehensive sensitivity analysis with a new photosynthesis and energy balance scheme
.
Global Biogeochem. Cycles
,
15
,
207
225
.
Labat
,
D.
,
Y.
Godderis
,
J.-L.
Probst
, and
J.-L.
Guyot
,
2004
:
Evidence for global runoff increase relating to climate warming
.
Adv. Water Resour.
,
27
,
631
642
.
Law
,
B.
,
M.
Ryan
, and
P.
Anthoni
,
1999
:
Seasonal and annual respiration of a ponderosa pine ecosystem
.
Global Change Biol.
,
5
,
169
182
.
Lawrence
,
D.
, and
J.
Slingo
,
2004
:
An annual cycle of vegetation in a GCM. Part 1: Implementation and impact on evaporation
.
Climate Dyn.
,
22
,
87
105
.
Lenton
,
T.
,
R.
Marsh
, and
A.
Price
,
2007
:
Effects of atmospheric dynamics and ocean resolution on bi-stability of the thermohaline circulation examined using the Grid ENabled Integrated Earth system modelling (GENIE) framework
.
Climate Dyn.
,
29
,
591
613
.
Lewis
,
J.
,
R.
McKane
,
D.
Tingey
, and
P.
Beedlow
,
2000
:
Vertical gradients in photosynthetic light response within an old-growth douglas-fir and western hemlock canopy
.
Tree Physiol.
,
20
,
447
.
Lieth
,
H. F. H.
,
1975
:
Primary production of the major vegetation units of the world
.
Primary Productivity of the Biosphere, H. Lieth and R. H. Whittaker, Eds., Ecological Studies and Synthesis, Vol. 14, Springer-Verlag, 203–215
.
Los
,
S.
,
G.
Weedon
,
P.
North
,
J.
Kaduk
,
C.
Taylor
, and
P.
Cox
,
2006
:
An observation-based estimate of the strength of rainfall–vegetation interactions in the Sahel
.
Geophys. Res. Lett.
,
33
,
L16402
,
doi:10.1029/2006GL027065
.
McCallum
,
I.
,
W.
Wagner
,
C.
Schmullius
,
A.
Shvidenko
,
M.
Obersteiner
,
S.
Fritz
, and
S.
Nilsson
,
2009
:
Satellite-based terrestrial production efficiency modeling
.
Carbon Balance Manag.
,
4
,
8
.
Medlyn
,
B.
,
D.
Barrett
,
J.
Landsberg
,
P.
Sands
, and
R.
Clement
,
2003
:
Conversion of canopy-intercepted radiation to photosynthate: Review of modeling approaches for regional scales
.
Funct. Plant Biol.
,
30
,
153
169
.
Medlyn
,
B.
,
A.
Robinson
,
R.
Clement
, and
E.
McMurtie
,
2005
:
On the validation of models of forest CO2 exchange using eddy covariance data: Some perils and pitfalls
.
Tree Physiol.
,
25
,
839
857
.
Mehta
,
V.
,
A.
DeCandis
, and
A.
Mehta
,
2005
:
Remote-sensing-based estimates of the fundamental global water cycle: Annual cycle
.
J. Geophys. Res.
,
110
,
D22103
,
doi:10.1029/2004JD005672
.
Meir
,
P.
,
B.
Kruijt
,
M.
Broadmeadow
,
E.
Barbosa
,
O.
Kull
,
F.
Carswell
,
A.
Nobre
, and
P.
Jarvis
,
2002
:
Acclimation of photosynthetic capacity to irradiance in tree canopies in relation to leaf nitrogen concentration and leaf mass per unit area
.
Plant, Cell Environ.
,
25
,
343
357
.
Mercado
,
L.
,
N.
Bellouin
,
S.
Sitch
,
O.
Boucher
,
C.
Huntingford
,
M.
Wild
, and
P.
Cox
,
2009
:
Impact of changes in diffuse radiation of the global land carbon sink
.
Nature
,
458
,
1014
1018
.
Molteni
,
F.
,
2003
:
Atmospheric simulations using a GCM with simplified physical parameterisations: I. Model climatology and variability in multi-decadal experiments
.
Climate Dyn.
,
20
,
175
191
.
Monteith
,
J. L.
,
1965
:
Evaporation and environment
.
Symp. Soc. Exp. Biol.
,
19
,
205
234
.
Moorcroft
,
P.
,
2006
:
How close are we to a predictive science of the biosphere?
Trends Ecol. Evol.
,
21
,
400
407
.
Morcrette
,
J.
,
1990
:
Impact of changes to the radiative transfer parameterizations plus cloud optical properties in the ECMWF model
.
Mon. Wea. Rev.
,
118
,
847
873
.
Morel
,
A.
, and
J.
Berthon
,
1989
:
Surface pigments, algal biomass profiles and potential production of the euphotic layers: Relationships re-investigated in view of remote-sensing applications
.
Limnol. Oceangor.
,
34
,
1545
1562
.
Morisette
,
J.
,
F.
Baret
, and
J.
Privette
,
2006
:
Validation of global moderate-resolution LAI products: A framework proposed within the CEOS land product validation subgroup
.
IEEE Trans. Geosci. Remote Sens.
,
44
,
1804
1817
.
Müller
,
C.
, and
W.
Lucht
,
2007
:
Robustness of terrestrial carbon and water cycle simulations against variations in spatial resolutions
.
J. Geophys. Res.
,
112
,
D06105
,
doi:10.1029/2006JD007875
.
Nemani
,
R.
,
C.
Keeling
,
H.
Hashimoto
,
W.
Jolly
,
S.
Piper
,
C.
Tucker
,
R.
Myneni
, and
S.
Running
,
2003
:
Climate-driven increases in global terrestrial net primary production from 1982 to 1999
.
Science
,
300
,
1560
.
Newman
,
E.
,
1969
:
Resistance to water flow in soil and plant. II. A review of experimental evidence on the Rhizosphere resistance
.
J. Appl. Ecol.
,
6
,
261
272
.
Norman
,
J.
,
1992
:
Scaling processes between leaf and canopy levels
.
Scaling Physiological Processes: Leaf to Globe, J. Ehleringer and C. Field, Eds., Academic Press, 41–76
.
Oki
,
T.
, and
Y.
Sud
,
1998
:
Design of Total Runoff Integrating Pathways (TRIP)—A global river channel network
.
Earth Interactions
,
2
.
[Available online at http://EarthInteractions.org.]
Olson
,
R.
,
J.
Scurlock
,
S.
Prince
,
D.
Zheng
, and
K.
Johnson
, Eds., cited
2008
:
NPP multi-biome: NPP and driver data for ecosystem model-data intercomparison, Revision 1
.
Dataset. Oak Ridge National Laboratory, Distributed Active Archive Center. [Available online at http://www.daac.ornl.gov.]
Palmer
,
J.
, and
J.
Totterdell
,
2001
:
Production and export in a global ocean ecosystem model
.
Deep-Sea Res.
,
48
,
1169
1198
.
Petoukhov
,
V.
,
A.
Ganopolski
,
V.
Brovkin
,
M.
Claussen
,
A.
Eliseev
,
C.
Kubatzki
, and
S.
Rahmstorf
,
2000
:
CLIMBER-2: A climate system model of intermediate complexity. Part I: Model description and performance for present climate
.
Climate Dyn.
,
16
,
1
17
.
Peylin
,
P.
,
P.
Bousquet
, and
C.
Le Quere
,
2005
:
Multiple constraints on regional CO2 flux variations over land and oceans
.
Global Biogeochem. Cycles
,
19
,
GB1011
,
doi:10.1029/2003GB002214
.
Piao
,
S.
,
P.
Ciais
,
P.
Friedlingstein
,
N.
de Noblet-Ducoudre
,
P.
Cadule
,
N.
Viovy
, and
T.
Wang
,
2009
:
Spatiotemporal patterns of terrestrial carbon cycle during the 20th century
.
Global Biogeochem. Cycles
,
23
,
GB4026
doi:10.1029/2008GB0033339
.
Purves
,
D.
, and
S.
Pacala
,
2008
:
Predictive models of forest dynamics
.
Science
,
320
,
1452
1453
.
Randall
,
D. A.
, and
Coauthors
,
2007
:
Climate models and their evaluation
.
Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 589–662
.
Roderick
,
M.
,
G.
Farquhar
,
S.
Berry
, and
I.
Noble
,
2001
:
On the direct effect of clouds and atmospheric particles on the productivity and structure of vegetation
.
Oecologia
,
129
,
21
30
.
Ryan
,
M.
,
1991
:
The effects of climate change on plant respiration
.
Ecol. Appl.
,
1
,
157
167
.
Schlesinger
,
W.
,
1997
:
Biogeochemistry: An Analysis of Global Change
.
Academic Press, 432 pp
.
Schlosser
,
C.
, and
P.
Houser
,
2007
:
Assessing a satellite-era perspective of the global water cycle
.
J. Climate
,
20
,
1316
1338
.
Schulze
,
E. D.
,
F.
Kelliher
,
C.
Korner
,
J.
Lloyd
, and
R.
Leuning
,
1994
:
Relationships among maximum stomatal conductance, ecosystem surface conductance, carbon assimilation rate and plant nitrogen nutrition: A global ecology scaling exercise
.
Annu. Rev. Ecol. Syst.
,
25
,
629
660
.
Scurlock
,
J.
, and
R.
Olson
,
2002
:
Terrestrial net primary productivity—A brief history and a new worldwide database
.
Environ. Rev.
,
10
,
91
109
.
Seager
,
R.
,
R.
Murtugudde
,
A.
Clement
, and
C.
Herweijer
,
2003
:
Why is there an evaporation minimum at the equator
.
J. Climate
,
16
,
3793
3802
.
Sellers
,
P.
,
D.
Randall
, and
G.
Collatz
,
1996
:
A revised land surface parameterization (SiB2) for atmospheric GCMs. Part I: Model formulation
.
J. Climate
,
9
,
676
705
.
Sharma
,
D.
,
A.
Das Gupta
, and
M.
Babel
,
2007
:
Spatial disaggregation of bias-corrected GCM precipitation for improved hydrologic simulation: Ping River Basin, Thailand
.
Hydrol. Earth Syst. Sci. Discuss.
,
4
,
35
74
.
Sheffield
,
J.
,
G.
Goteti
, and
E.
Wood
,
2006
:
Development of a 50 year high-resolution global dataset of meteorological forcings for land surface modelling
.
J. Climate
,
19
,
3088
3111
.
Smith
,
T.
,
R.
Reynolds
,
T.
Peterson
, and
J.
Lawrimore
,
2008
:
Improvements to NOAA’s historical merged land–ocean surface temperature analysis (1880–2006)
.
J. Climate
,
21
,
2283
2296
.
Strahler
,
A.
, and
A.
Strahler
,
1992
:
Modern Physical Geography
.
John Wiley & Sons, 656 pp
.
Trenberth
,
K.
,
1999
:
Atmospheric moisture recycling: Role of advection and local evaporation
.
J. Climate
,
12
,
1368
1381
.
Trenberth
,
K.
,
L.
Smith
,
T.
Qian
,
A.
Dai
, and
J.
Fasullo
,
2007
:
Estimates of the global water budget and its annual cycle using observational and model data
.
J. Hydrometeor.
,
8
,
758
769
.
Trenberth
,
K.
,
J.
Fasullo
, and
J.
Kiehl
,
2009
:
Earth’s global energy budget
.
Bull. Amer. Meteor. Soc.
,
90
,
311
323
.
Williams
,
D.
, and
Coauthors
,
2004
:
Evapotranspiration components determined by stable isotope, sap flow and eddy covariance techniques
.
Agric. For. Meteor.
,
125
,
241
258
.
Yu
,
F.
,
K.
Price
,
J.
Ellis
,
J.
Feddema
, and
P.
Shi
,
2004
:
Interannual variations of the grassland boundaries bordering the eastern edges of the Gobi Desert in central Asia
.
Int. J. Remote Sens.
,
25
,
327
346
.
Yu
,
L.
,
2007
:
Global variations in oceanic evaporation (1958–2005): the role of the changing wind speed
.
J. Climate
,
20
,
5376
5390
.
Yu
,
L.
, and
R.
Weller
,
2007
:
Objectively analyzed air-sea heat fluxes (OAFlux) for the global oceans
.
Bull. Amer. Meteor. Soc.
,
88
,
527
539
.
Yuan
,
W.
,
S.
Liu
, and
G.
Zhou
,
2007
:
Deriving a light-use efficiency model from eddy covariance covariance flux data for predicting daily gross primary production across biomes
.
Agric. For. Meteor.
,
143
,
189
207
.
Zaehle
,
S.
,
S.
Sitch
,
B.
Smith
, and
F.
Hatterman
,
2005
:
Effects of parameter uncertainties on the modeling of terrestrial biosphere dynamics
.
Global Biogeochem. Cycles
,
19
,
GB3020
,
doi:10.1029/2004GB002395
.

Footnotes

1

Acronyms used frequently in the text are listed in Table 1.

2

We show productivity rates (kg m−2 yr−1) and E rates (mm yr−1 or mm month−1) when plotting zonal fluxes but, when integrating globally or regionally, we refer to mass-integrated (total) productivity or E, both measured in gigatons (Gt).