Approximately half of global primary production occurs in the ocean. While the large‐scale variability in net primary production (NPP) has been extensively studied, ocean gross primary production (GPP) has thus far received less attention. In this study, we derived two satellite‐based GPP models by training machine learning algorithms (Random Forest) with light‐dark bottle incubations (GPP_LD) and the triple isotopes of dissolved oxygen (GPP_17Δ). The two algorithms predict global GPPs of 9.2 ± 1.3 × 10¹⁵ and 15.1 ± 1.05 × 10¹⁵ mol O₂ yr⁻¹ for GPP_LD and GPP_17Δ, respectively. The projected GPP distributions agree with our understanding of the mechanisms regulating primary production. Global GPP_17Δ was higher than GPP_LD by an average factor of 1.6 which varied meridionally. The discrepancy between GPP_17Δ and GPP_LD simulations can be partly explained by the known biases of each methodology. After accounting for some of these biases, the GPP_17Δ and GPP_LD converge to 9.5 ∼ 12.6 × 10¹⁵ mol O₂ yr⁻¹, equivalent to 103 ∼ 150 Pg C yr⁻¹. Our results suggest that global oceanic GPP is 1.5–2.2 fold larger than oceanic NPP and comparable to GPP on land.

中文翻译：

---

基于实地观测的机器学习扩展的海洋初级总产值的全球估计

全球初级生产的大约一半发生在海洋中。尽管已经对净初级生产力（NPP）的大规模变化进行了广泛研究，但到目前为止海洋初级生产力（GPP）却很少受到关注。在这项研究中，我们通过训练机器学习算法（Random Forest），浅色暗瓶孵化法（GPP _LD）和溶解氧的三重同位素（_GPP17Δ）得出了两个基于卫星的GPP模型。这两种算法预测GPP _LD和_GPP17Δ的全局GPP为9.2±1.3×10 ¹⁵和15.1±1.05×10 ¹⁵  mol O ₂  yr ^-1， 分别。预计的GPP分布与我们对调节初级生产的机制的理解相符。全球_GPP17Δ比GPP _LD高出平均因数1.6，而平均因数在子午线上有所变化。可以通过每种方法的已知偏差来部分解释_GPP17Δ和GPP _LD仿真之间的差异。在考虑了这些偏见后，_GPP17Δ和GPP _LD收敛至_9.5〜12.6 ×10 ¹⁵  mol O ₂  yr ^-1，相当于_103〜150 Pg C yr ^-1。我们的结果表明，全球海洋GPP比海洋NPP大1.5–2.2倍，可与陆上GPP相媲美。