Primary Production Algorithms Research Articles

A new global oceanic multi-model net primary productivity data product

Abstract. Net primary production of the oceans contributes approximately half of the total global net primary production, and long-term observational records are required to assess any climate-driven changes. The Ocean Colour Climate Change Initiative (OC-CCI) has proven to be robust whilst also being one of the longest records of ocean colour. However, to date, only one primary production algorithm has been applied to this data product, with other algorithms typically applied to single-sensor missions. The data product presented here addresses this issue by applying five algorithms to the OC-CCI data product, which allows the user to interrogate the range of distributions across multiple models and to identify consensus or outliers for their specific region of interest. Outputs are compared to single-sensor data missions, highlighting good overall global agreement, with some small regional discrepancies. Inter-model assessments address the source of these discrepancies, highlighting the choice of the mixed-layer data product as a vital component for accurate primary production estimates. The datasets are published in the Zenodo repository at https://doi.org/10.5281/zenodo.7849935, https://doi.org/10.5281/zenodo.7858590, https://doi.org/10.5281/zenodo.7860491 and https://doi.org/10.5281/zenodo.7861158 (Ryan-Keogh et al., 2023a, b, c, d).

Impact of subsurface chlorophyll maxima on satellite-based Arctic spring primary production estimates

In recent years several regional and pan-Arctic ocean color algorithms, which consider the unique bio-optical properties of the Arctic Ocean, have been developed to accurately extract surface chlorophyll a (Chl a), a key input into net primary production (NPP) algorithms, from spectral remote sensing reflectance (Rrs). However, most satellite derived NPP (NPPSAT) algorithms used in the Arctic do not account for production at deep subsurface chlorophyll maxima (SCMs), a prominent feature in the Arctic Ocean during the summer months, leading to underestimations of NPP during the post-bloom period. In this study, a shallow SCM was observed in early summer (June/July 2016) in Baffin Bay at the time of the sea ice edge bloom, raising the question to what extent this SCM contributes to Rrs and how it is captured in NPPSAT estimates. Radiative transfer simulations based on in situ data of inherent and apparent optical properties and Chl a concentration in the water column were used to examine the effects of heterogeneous vertical Chl a profiles of various strength and depth on spectral reflectance, algorithm-derived surface Chl aSAT and NPPSAT. NPPSAT estimates for varying Chl a profile shapes were then compared to NPPSAT estimates of reference simulations for homogenous Chl a profiles, and to measured NPP calculated using observed Chl a profiles. Results show a significant impact of shallow SCMs on Rrs with the depth of maximum Chl a < 30 m, causing an increase in Chl aSAT by 17 ± 6% relative to a vertically homogenous ocean. Interestingly, SCMs significantly influencing Rrs were found down to the fifth light penetration depth. The increase in Chl aSAT reduced the difference between predicted and measured NPP estimates to −25 ± 12% at stations with a shallow SCM. Furthermore, due to the significant contribution of these shallow SCM stations to regional NPP, regional predicted NPPSAT was only 16% lower than measured NPP. Our results demonstrate that the partial representation of a shallow and very productive SCM in Chl aSAT minimizes errors in satellite-based NPPSAT estimates of the Arctic spring bloom during the sea ice melt period.

Performance of primary production algorithm using absorption coefficient of phytoplankton in the Pacific Arctic

Performance of primary production algorithm using absorption coefficient of phytoplankton in the Pacific Arctic

Estimating estuarine primary production using satellite data and machine learning

• First attempt to use machine learning to estimate estuarine NPP from satellite data. • Machine learning algorithm greatly outperformed VGPM for NPP estimation. • Tampa Bay NPP shows seasonality and an increasing trend from 2003 to 2020. • Seasonality and long-term increasing trend are largely driven by water temperature. • NPP shows a spatial pattern that likely reflects differences in river discharge. Net primary production (NPP) plays an important role in estuarine carbon cycling, which has been increasingly impacted by human activities and global climate change. Spatiotemporal trends of NPP in the open ocean have been well studied using satellite data and standard primary production algorithms such as the Vertically Generalized Production Model (VGPM), but these algorithms are generally not suitable for estuarine and coastal waters. Previous remote sensing studies on estuarine NPP mainly focused on the tuning of the standard VGPM based on extensive local in situ data. Here we make a first attempt to use machine learning algorithms to estimate NPP in an estuarine environment from satellite measurements. Tampa Bay, the largest estuary in Florida (United States), has abundant in situ measurements of NPP, although the spatiotemporal variability of NPP within the bay remains unrevealed. Combining these data with concurrent MODIS/Aqua image data, we developed and evaluated seven machine learning algorithms (support vector regression, random forest, decision tree, bagging, adaptive boosting, gradient tree boosting, and neural network), and applied the one with the least root mean square error (RMSE) and highest correlation coefficient to establish a time-series NPP record for Tampa Bay from 2002 to 2020. Results of the best performing algorithm (Pearson’s r = 0.82, RMSE = 151.0 mgC m −2 d −1 ) showed substantial improvement over the standard or tuned VGPM (Pearson’s r = 0.51, RMSE = 364.9 mgC m −2 d −1 ), using the same independent variables for NPP ranging from 157.5 to 1368.8 mgC m −2 d −1 . MODIS NPP shows temporal variations that are largely driven by temperature: lowest values in winter, highest values in summer, and an increasing trend from 2003 to 2020, highlighting the impact of global warming on estuarine NPP. The spatial distribution of MODIS NPP shows higher values in Hillsborough Bay, Middle Tampa Bay and Lower Tampa Bay, and relatively lower values in Old Tampa Bay, a pattern that likely reflects differences in river discharge. The long-term NPP product derived from machine learning algorithms and satellite data can complement existing field-based monitoring programs and help to understand estuarine responses to climate changes and human impacts, and design relevant mitigation strategies.

Seasonal Variability and Estimation of Annual East Siberian Sea Phytoplankton Primary Production and Comparison with the Other Siberian Seas

Studies of seasonal variability of the East Siberian Sea (ESS) water column primary production (IPP) and evaluation of its total annual value (PPtot) were performed using MODIS-Aqua data (2002–2018). Region-specific primary production and chlorophyll algorithms were used for the first time to achieve that result. Northeastern and Southwestern regions were distinguished in ESS based on multiyear average daily primary production. Seasonal variations in IPP in the Northeastern region were characterized by the maximum in June (273 mgC m–2 d–1). The maximum of water column primary production value in the Southwestern region was recorded in May (311 mgC m–2 d–1). The maximum of IPP for the entire ESS was registered in June (273 mgC m–2 d–1). Values of daily primary production and PPtot in the Southwestern region were, respectively, 1.8 and 2 fold higher than in the Northeastern region. The multiyear averaged value of ESS IPP was equal to 91 mgC m–2 d–1 and PPtot value was equal to 9 TgC. Productivity of the Siberian seas decreased eastward.

Validation of a Primary Production Algorithm of Vertically Generalized Production Model Derived from Multi-Satellite Data around the Waters of Taiwan

Basin-scale sampling for high frequency oceanic primary production (PP) is available from satellites and must achieve a strong match-up with in situ observations. This study evaluated a regionally high-resolution satellite-derived PP using a vertically generalized production model (VGPM) with in situ PP. The aim was to compare the root mean square difference (RMSD) and relative percent bias (Bias) in different water masses around Taiwan. Determined using light–dark bottle methods, the spatial distribution of VGPM derived from different Chl-a data of MODIS Aqua (PPA), MODIS Terra (PPT), and averaged MODIS Aqua and Terra (PPA&amp;T) exhibited similar seasonal patterns with in situ PP. The three types of satellite-derived PPs were linearly correlated with in situ PPs, the coefficients of which were higher throughout the year in PPA&amp;T (r2 = 0.61) than in PPA (r2 = 0.42) and PPT (r2 = 0.38), respectively. The seasonal RMSR and bias for the satellite-derived PPs were in the range of 0.03 to 0.09 and −0.14 to −0.39, respectively, which suggests the PPA&amp;T produces slightly more accurate PP measurements than PPA and PPT. On the basis of environmental conditions, the subareas were further divided into China Coast water, Taiwan Strait water, Northeastern upwelling water, and Kuroshio water. The VPGM PP in the four subareas displayed similar features to Chl-a variations, with the highest PP in the China Coast water and lowest PP in the Kuroshio water. The RMSD was higher in the Kuroshio water with an almost negative bias. The PPA exhibited significant correlations with in situ PP in the subareas; however, the sampling locations were insufficient to yield significant results in the China Coast water.

Climate change scenarios and projected impacts for forest productivity in Guanacaste Province (Costa Rica): lessons for tropical forest regions

The Guanacaste Province of Costa Rica is home to highly diverse forests which are under threat of degradation due to ongoing climatic changes. There is concern that increasing temperatures and changes in precipitation will force these forests outside of their optimal growth ranges leading to widespread degradation of forest productivity. The objectives of this study are, therefore, to project and assess climatic changes in Guanacaste and to build a relationship between these climatic changes and forest productivity with the goal of projecting productivity trends into the future. The ClimateSA model was used to project RCP4.5 and 8.5 scenarios from 2018 to 2080 and then assess these projections for the mean and extreme future conditions. Furthermore, the MODIS gross primary productivity (GPP) algorithm was used to build a relationship between forest productivity and the vapour pressure deficit scalar (VPD scalar) and project GPP alteration under future climatic scenarios both seasonally and annually. Results indicate that Guanacaste’s mean annual precipitation will stay within the historic levels for both the RCP4.5 and 8.5 scenarios. The monthly and annual temperatures increase in all projections. Results also indicate that the productivity-climate relationship follows a quadratic relationship between GPP and the VPD scalar. This quadratic relationship leads to areas with higher precipitation (high VPD scalar) experiencing an increase in GPP as they dry in the future. In drier areas (low VPD scalar), reduced precipitation will stabilize or decrease GPP.

Seasonal variability and annual Laptev sea phytoplankton primary production using MODIS-aqua data

Studies of seasonal variability of Laptev Sea water column primary production and evaluation of its annual values were performed using MODIS-Aqua data (20022018). To reach that result regional-specific primary production and chlorophyll algorithms were used for the first time. Based on multiyear averaged daily primary production Northwestern and Southeastern regions were distinguished in the Laptev Sea. Seasonal variations in water column primary production in the Northwestern region were characterized by the maximum in June (245 mgC m‑2 d‑1). In the Southeastern region and for all Laptev Sea the maximum water column primary production values were denoted from May to July, 273282 mgC m‑2 d‑1 and 256281 mgC m‑2 d‑1, respectively. Daily primary production and annual values of total primary production in the Southeastern region were, respectively, 1.9 and 3 fold higher than in the Northwestern region. Multiyear averaged value of Laptev Sea water column primary production was equal to 125 mgC m‑2 d‑1 and total annual primary production was equal to 8 1012 gC.

Spatial distribution of common Minke whale (Balaenoptera acutorostrata) as an indication of a biological hotspot in the East Sea

The minke whale (Balaenoptera acutorostrata) is the most common baleen whale among several marine mammal species observed in Korea. Since a high concentrated condition of prey to whales can be obtained by physical structures, the foraging whale distribution can be an indicator of biological hotspot. Our main objective is verifying the coastal upwelling-southwestern East Sea as a productive biological hotspot based on the geographical distribution of minke whales. Among the cetacean research surveys of the National Institute of Fisheries Science since 1999, 9 years data for the minke whales available in the East Sea were used for this study. The regional primary productivity derived from Moderate-Resolution Imaging Spectroradiometer (MODIS) was used for a proxy of biological productivity. Minke whales observed during the sighting surveys were mostly concentrated in May and found mostly (approximately 70%) in the southwestern coastal areas (< 300m) where high chlorophyll concentrations and primary productivity were generally detected. Based on MODIS-derived primary productivity algorithm, the annual primary production (320gCm−2y−1) estimated in the southwestern coastal region of the East Sea belongs to the highly productive coastal upwelling regions in the world. A change in the main spatial distribution of minke whales was found in recent years, which indicate that the major habitats of mink whales have been shifted into the north of the common coastal upwelling regions. This is consistent with the recently reported unprecedented coastal upwelling in the mid-eastern coast of Korea. Based on high phytoplankton productivity and high distribution of minke whales, the southwestern coastal regions can be considered as one of biological hotspots in the East Sea. These regions are important for ecosystem dynamics and the population biology of top marine predators, especially migratory whales and needed to be carefully managed from a resource management perspective.

Evaluating Parameter Adjustment in the MODIS Gross Primary Production Algorithm Based on Eddy Covariance Tower Measurements

How well parameterization will improve gross primary production (GPP) estimation using the MODerate-resolution Imaging Spectroradiometer (MODIS) algorithm has been rarely investigated. We adjusted the parameters in the algorithm for 21 selected eddy-covariance flux towers which represented nine typical plant functional types (PFTs). We then compared these estimates of the MOD17A2 product, by the MODIS algorithm with default parameters in the Biome Property Look-Up Table, and by a two-leaf Farquhar model. The results indicate that optimizing the maximum light use efficiency (εmax) in the algorithm would improve GPP estimation, especially for deciduous vegetation, though it could not compensate the underestimation during summer caused by the one-leaf upscaling strategy. Adding the soil water factor to the algorithm would not significantly affect performance, but it could make the adjusted εmax more robust for sites with the same PFT and among different PFTs. Even with adjusted parameters, both one-leaf and two-leaf models would not capture seasonally photosynthetic dynamics, thereby we suggest that further improvement in GPP estimaiton is required by taking into consideration seasonal variations of the key parameters and variables.

Comparison of primary productivity algorithms for Korean waters

This study compares five primary productivity algorithms for Korean waters. Five algorithms are in the form of vertical generalized production models: One algorithm is for gross primary production and four are for net primary production. The five algorithms were evaluated using 117 in situ primary production datasets observed by 20 cruises from 1994 to 2011 in Korean waters (East Sea, Yellow Sea, East China Sea, and Yeosu Bay). The results show that the regionally-tuned variants give better results than the original formulation. We recommend, among the tested algorithms, YSVGPM (Yellow Sea Vertically Generalized Productivity Model) for gross primary productivity algorithm and Kameda-Ishizaka algorithm for net primary productivity algorithm for estimating primary production in Korean waters.

A reassessment of primary production and environmental change in the Bering Sea

[1] Regarded as one of the world's most productive marine environments, the Bering Sea is widely thought to be rapidly warming and losing sea ice. Such changes would be expected to have dramatic impacts on primary producers, with cascading effects on upper trophic levels, including this region's vast commercial fisheries resources. Here, we use satellite-derived sea ice concentration, sea surface temperature, and ocean color data as input to a primary productivity algorithm to take stock of environmental change and primary production in the Bering Sea. Results show that, rather than declining, mean annual sea ice extent in the Bering Sea has exhibited no significant change over the satellite sea ice record (1979–2009). Furthermore, significant warming during the satellite sea surface temperature record (1982–2009) is mainly limited to the summer months, when all regions of the Bering Sea warmed. This warming suggests increasing stratification during the phytoplankton growth season. Despite certain hot spots of primary production and a strong pulse in the spring, the rate of annual area-normalized primary production in the Bering Sea (124 g C m−2 yr−1) is below the global mean (140 g C m−2 yr−1). Between 1998 and 2007, basin-wide annual primary production ranged from 233 to 331 Tg C yr−1 under the influence of highly variable sea ice and temperature conditions. By comparing warm, low-ice years (2001–2005) with cold, high-ice years (1998–2000 and 2006–2007), we speculate that Bering Sea primary productivity is likely to rise under conditions of future warming and sea ice loss.

Two-component modeling of the optical properties of a diatom bloom in the Southern Ocean

Diatom cells have distinctive optical characteristics, originating from their relatively large cell size, fucoxanthin content and silica cell wall. It has been proposed that diatom-dominated phytoplankton blooms can be identified by optical remote sensing and that specifically tuned chlorophyll and primary production algorithms should be applied in regions where these blooms are present. However there have been few studies on how the optical properties of diatom blooms change as they progress from active growth to senescence, and it is unlikely that measurements on laboratory cultures encompass the full range of physiological states found in natural waters. We have therefore examined the inherent optical properties (IOPs) of the waters around the island of South Georgia at the end of the spring diatom bloom. Considerable variability was found in the relationships between the inherent optical properties and analytically determined chlorophyll a concentrations even in the surface layer, which meant that the usual bio-optical assumptions for Case 1 waters did not apply. To account for this variability, phytoplankton absorption and scattering were modeled as a two-component mixture, with the components representing actively growing and senescent material. The specific inherent optical properties of the two components were derived by linear regression of total IOPs against chlorophyll concentration and a fraction of the suspended mineral concentration. These specific IOPs were used to develop radiative transfer models of diatom blooms in varying stages of growth and senescence. Remote sensing reflectances calculated using this technique confirmed the tendency of the standard algorithms employed in SeaWiFS, MODIS and MERIS data processing to under-estimate near-surface chlorophyll concentrations in diatom blooms. However the inclusion of increasing proportions of senescent material had a significant effect on algorithm performance only at chlorophyll concentrations below 10 mg m − 3 . Optical depths predicted by the model around South Georgia were 9 +/− 2 m at 512 nm, indicating that a large fraction of the phytoplankton biomass was located below the depth from which the remote sensing signals originated.

Site-level evaluation of MODIS-based primary production in an old-growth forest in Northeast China

To improve the accuracy of the moderate resolution imaging spectroradiometer (MODIS) gross primary production (GPP) algorithm, it is critical to evaluate MODIS GPP production for different land cover types using ground-based measurements. In this paper the MODIS primary production products (MOD17) is evaluated by using site-specific input parameters to the algorithm and compared to the results of eddy covariance measurements over an old-growth forest. Direct comparisons suggest that 8-day GPP predicted by the standard MODIS algorithm was highly correlated with flux tower based GPP (r2 = 0.77, P<0.001), while with an average underestimation of 39%. The difference is substantial in magnitude mainly because the inputs of underestimated MODIS biome-specific parameters, maximum light use efficiency emax and MODIS derived fraction of photosynthetically active radiation. The modified MODIS algorithm GPP calculated with site-specific input parameters compares favorably with ground flux tower observations (r2 = 0.92, relative error = 7%). These results suggest that the MODIS GPP production is most likely underpredicted in forest sites with high primary production, and site-specific input parameters could help to improve the accuracy of MODIS GPP algorithm.

Evaluation of Collections 4 and 5 of the MODIS Gross Primary Productivity product and algorithm improvement at a tropical savanna site in northern Australia

In this study, we assessed the accuracy of the MODIS (Moderate Resolution Imaging Spectroradiometer) GPP (gross primary productivity) Collections 4.5, 4.8 and 5 along with Leaf Area Index (LAI), fraction of absorbed Photosynthetically Active Radiation (fPAR), light use efficiency (LUE) and meteorological variables that are used to estimate GPP for a northern Australian savanna site. Results of this study indicated that the MODIS products captured the seasonal variation in GPP, LAI and fPAR well. Using the index of agreement (IOA), it was found that Collections 4.5 and 4.8 (IOA 0.89 respectively) agreed reasonably well with flux tower measurements between 2001 and 2006. It was also found that MODIS Collection 4.5 predicted the dry season GPP well (Relative Predictive Error (RPE) 4.17%, IOA 0.72 and Root Mean Square Error (RMSE) of 1.05 g C m − 2 day − 1 ), whilst Collection 4.8 performed better in capturing wet season dynamics (RPE 1.11%, IOA 0.80 and RMSE of 0.91 g C m − 2 day − 1 ). Although the wet season magnitude of GPP was predicted well by Collection 4.8, an examination of the inputs to the GPP algorithm revealed that MODIS fPAR was too high, but this was compensated by PAR and LUE that was too low. Although LAI and fPAR estimated by Collection 5 were more accurate, GPP for this Collection resulted in a much lower value (RPE 25%) due to errors in other factors. Recalculation of MODIS GPP using site specific input parameters indicated that MODIS fPAR was the main reason for the differences between MODIS and tower derived GPP followed by LUE and meteorological inputs. GPP calculated using all site specific values agreed very well with tower data on an annual basis (IOA 0.94, RPE 6.06% and RMSE 0.83 g C m − 2 day − 1 ) but the early initiation of the growing season calculated by the MODIS algorithm was improved when the vapor pressure deficit (VPD) function was replaced with a soil water deficit function. The results of this study however, reinforce previous findings in water limited regions, like Australia, and incorporation of soil moisture in a LUE model is needed to accurately estimate the productivity.

Trends in primary production in the California Current detected with satellite data

Several ocean primary production algorithms using satellite data were evaluated on a large archive of net primary production (NPP) and chlorophyll‐a (Chl‐a) measurements collected by the California Cooperative Fisheries Investigations program in the California Current. The best algorithm matching in situ data was found by empirically adjusting the Behrenfeld‐Falkowski Vertically Generalized Production Model. Satellite‐derived time series of NPP were calculated for the California Current area. Significant increase in NPP and Chl‐a annual peak levels, i.e., the “bloom magnitude,” were found along the coasts of the California Current as well as other major eastern boundary currents for the period of modern ocean color data (1997–2007). The reasons for this increase are not clear but are associated with various environmental conditions.

Determination of ocean primary productivity using support vector machines

A major task of ocean colour observations is to determine the distribution of phytoplankton primary production. At present, the global coverage of the sea surface chlorophyll concentration, sea surface temperature, photosynthetically available radiation (PAR) can nominally be achieved every 1 to 2 days with standard algorithms from satellite data. From these standard products, a variety of bio‐optical algorithms has been developed to estimate ocean primary productivity. In this communication, we have investigated the possibility of using a novel universal approximator‐support vector machine (SVM) as the nonlinear transfer function between ocean primary productivity and the information that can be retrieved from satellite data, including chlorophyll concentration, PAR, maximum carbon fixation rate and day length, which is the same as the vertically generalized production model (VPGM). The VGPM dataset was used to evaluate the proposed approach. The primary production algorithm round robin 2 (PPARR2) dataset was used to further compare the precision between the VGPM and the SVM model. The results suggest that the SVM model is more accurate than the VGPM. Using the SVM model to calculate the global ocean primary productivity, the result is 45.5 Pg C yr−1, which is a little higher than the VGPM result.

Impact of a shrinking Arctic ice cover on marine primary production

Loss of Arctic sea ice has accelerated recently, culminating in a 2007 summer minimum ice extent that was 23% below the previous low. To quantify the impact of this unprecedented loss of ice on marine primary production, we have coupled satellite‐derived sea ice, SST, and chlorophyll to a primary production algorithm parameterized for Arctic waters. Annual primary production in the Arctic has increased yearly by an average of 27.5 Tg C yr−1 since 2003 and by 35 Tg C yr−1 between 2006 and 2007. 30% of this increase is attributable to decreased minimum summer ice extent and 70% to a longer phytoplankton growing season. Should these trends continue, additional loss of ice during Arctic spring could boost productivity &gt;3‐fold above 1998–2002 levels, potentially altering marine ecosystem structure and the degree of pelagic‐benthic coupling. Changes in carbon export could in turn modify benthic denitrification on the vast continental shelves.

Primary production in the Arctic Ocean, 1998–2006

Sea ice in the Arctic Ocean has undergone an unprecedented reduction in area and thickness in the last decade, exposing an ever increasing fraction of the sea surface to solar radiation and increasing the habitat suitable for phytoplankton growth. Here we use a primary production algorithm that utilizes remotely sensed chlorophyll a, sea surface temperature, and sea ice extent data to quantify interannual changes in phytoplankton production in the Arctic Ocean between 1998 and 2006. Our results show that since 1998, open water area in the Arctic has increased at the rate of 0.07 × 106 km2 a−1 (where a is years), with the greatest increases in the Barents, Kara, and Siberian sectors, particularly over the continental shelf. Although pan‐Arctic primary production averaged 419 ± 33 Tg C a−1 during 1998–2006, recent increases in open water area have lead to higher rates of annual production, which reached a 9‐year peak in 2006. Annual production was roughly equally distributed between pelagic waters (less productive but greater area) and waters located over the continental shelf (more productive but smaller area). Interannual differences are most tightly linked to changes in sea ice extent, with changes in sea surface temperature (related to the Arctic Oscillation) and incident irradiance playing minor roles. Estimation of primary production in the Arctic will aid the assessment of air‐sea CO2 fluxes and improve our understanding of the ecological and biogeochemical changes that could take place if ice cover continues to decrease.

Primary production in the Southern Ocean, 1997–2006

Estimates of primary production in the Southern Ocean are difficult to obtain but are essential if we are to understand its role in the global carbon cycle. Here we present a 9‐year time series of daily primary production calculated from remotely sensed ocean color, sea surface temperature, and sea ice concentration using a primary production algorithm parameterized specifically for use in Southern Ocean waters. Results suggest that total annual production in waters south of 50°S averaged 1949 ± 70.1 Tg C a−1 (where a is years) between 1998 and 2006, approximately half that of previous estimates. The large but relatively unproductive pelagic province accounted for ∼90% of Southern Ocean production, while area normalized rates of production were greatest on the much smaller continental shelf (109 g C m−2 a−1). Surprisingly, production in the marginal ice zone was only slightly higher than in the pelagic province. The Ross Sea was the most productive sector of the Southern Ocean (mean = 503 Tg C a−1), followed closely by the Weddell Sea (mean = 477 Tg C a−1). Unlike the Arctic Ocean, there was no secular trend in either sea ice cover or annual primary production in the Southern Ocean during our 9‐year study. Interannual variability in annual production was most closely tied to changes in sea ice cover, although changes in sea surface temperature also played a role. Only 31% of the variation in annual production was explained by the Southern Annular Mode. Annual primary production could increase in the future as stronger winds increase nutrient upwelling.