We evaluate a recent region-specific model of export production (Kelly et al., 2018) and some similar fits to in situ data for the California Current Ecosystem using satellite data. The model is a simple linear function of net primary productivity (NPP): Export = 0.08 × NPP + 72 where EF is export flux in mg C m⁻² d⁻¹. We confirm that contrary to several global algorithms export efficiency (e-ratio = export/primary productivity) is negatively correlated with net primary productivity. We find that linear models with a steeper slope of EF relative to NPP produce better estimates of the variability range. Choice of the EF model parameterization can more than double the estimate of temporal variability (standard deviation) in satellite-derived EF time series. The best estimates of EF were obtained when using average NPP during a preceding period of ~7–8 days. This is in contrast with NPP where the best satellite estimates of in situ NPP were obtained using same day satellite data and the coefficient of determination was monotonically decreasing with increasing time lag. We also find that there is substantial unexplained variability in EF that cannot be explained by existing models.

我们评估了最近特定于区域的出口生产模型（Kelly et al。，2018），并且使用卫星数据对加利福尼亚当前生态系统的原位数据进行了一些类似的拟合。该模型是净初级生产力（NPP）的简单线性函数：出口= 0.08×NPP + 72其中EF是出口通量，单位为mg C m ^-2 d ^-1。我们确认，与几种全局算法的出口效率（e-比率=出口/初级生产力）与净初级生产力负相关。我们发现，相对于NPP而言，EF斜率更大的线性模型可以更好地估计变异范围。EF模型参数化的选择可以使卫星衍生的EF时间序列中的时间变异性（标准偏差）的估计值增加一倍以上。当使用前约7-8天的平均NPP时，可以得到对EF的最佳估计。这与NPP相反，在NPP中，使用当日卫星数据获得了最佳的原位NPP卫星估计，并且确定系数随着时间滞后的增加而单调递减。我们还发现，EF中存在大量无法解释的可变性，现有模型无法解释。