Processing and Data for "Estimating ocean net primary productivity from daily cycles of carbon biomass measured by profiling floats"

Description: These files contain processed BGC-Argo float data, figure data, the radiocarbon productivity subset, bootstrapping results, and the associated Python/Matlab code to calculate net primary productivity from daily cycles of optical backscatter and dissolved oxygen. The raw float data used in this study are available from the Argo Global Data Assembly Centers in Brest, France (ftp://ftp.ifremer.fr/ifremer/argo/dac/coriolis) and Monterey, California (ftp://usgodae.org/pub/outgoing/argo/dac/coriolis). The raw MODIS satellite-based productivity data is available from the Oregon State University Ocean Productivity site (http://orca.science.oregonstate.edu/npp_products.php). The raw MODIS satellite-based euphotic depth estimates are available from the NASA L3 browser (https://oceancolor.gsfc.nasa.gov/l3/). The original ship-based estimates of net primary productivity are available from the Pangaea (https://doi.pangaea.de/10.1594/PANGAEA.932417) and the British Oceanography Data Centre (https://www.bco-dmo.org/dataset/814803). Please cite as: Stoer, A., and Fennel, K. 2022. Processing and Data for Estimating ocean net primary productivity from daily cycles of carbon biomass measured by profiling floats. Zenodo. doi: 10.5281/zenodo.6977161. Python/MATLAB Software Description: dielFit_GOPeqCR.m: This code is from Johnson and Bif (2021). We have added outputs for standard errors for linear and PvE models and sunrise/sunset times. To run this code with the associated Python software a MATLAB engine needs to be installed. Please see: https://www.mathworks.com/help/matlab/matlab-engine-for-python.html argo_so_processing_20220815.py: This code is the first of two pieces of software for estimating net primary productivity from floats in the Southern Ocean. The program below obtains the data from the BGC Argo database (Argo, 2021) and processes it. Simple data quality control, interpolation, biogeochemical calculations, and data binning occur. The processed float data is located in the folder 'Processed Argo Transects'. argo_daily_npp_20220815.py: This code using processed Argo float data that contains oxygen and particle backscatter measurements to infer net primary production. The code combines the float that meet the criteria of sampling at all local hours of the day throughout its lifetime. Then, it constructs diel cycles from this data by finding the median value of each hour and uses the code from Johnson and Bif (2021), which is a modified version from Barone et al. (2019). The algorithm used to convert particle backscatter to particulate organic carbon is from Graff et al. (2015). We assume that dissolved primary productivity accounts for 30% of total primary productivity (Moran et al., 2022). argo_daily_npp_bootstrap_20220815.py: This code using processed Argo float data that contains co-located oxygen and particle backscatter measurements to infer net primary production. This code is very similar to argo_daily_npp_20220815.py but randomly samples a subset of the co-located profiles at different sample sizes before calculating net primary productivity. Productivity is calculated at each sample size 1000 times. The results of this analysis is located in the folder 'Bootstrapped Results'. More details can be found in the code itself. Data Descriptions: Data from 'Processed Argo Transects' Folder | Description for each variable Variable Description Units depth Average depth of depth bin m mid_depth Center of depth bin m pressure Average pressure in depth bin dbar profile_index Profile number or index profile_longitude Average longitude of profile degE profile_latitude Average latitude of profile degN profile_time Average UTC time of profile yyyy-mm-dd hh:mm:ss profile_local_time Average local time of profile yyyy-mm-dd hh:mm:ss profile_local_hour The hour of the local timestamp salinity Seawater salinity PSU temperature Seawater temperature degC oxygen Dissolved oxygen concentration umol kg-1 oxygen_saturation Saturated dissolved oxygen concentration calculated from the Garcia and Gordon (1992) equation. umol kg-1 oxygen_anom The difference between observed dissolved oxygen concentration and saturated oxygen umol kg-1 bbp470 Optical backscatter coefficient at 470 nm. Particulate organic carbon is calculated in argo_daily_npp_20220815.py m-1 Data for Fig. 1 | Description for bbp_profile_locations.csv and oxy_profile_locations.csv Variable Description Units wmo WMO number of float profile_index Profile index or profile number taken by float profile_latitude Average profile latitude degN profile_longitude Average profile longitude degE Data for Fig. 2 Description for curve_fits.csv and hourly_means.csv Variable Description Units fod Fraction of day oxy Sinusoidal curve fit to oxygen mol m-3 poc Sinusoidal curve fit to particulate organic carbon mol m-3 oxy_med Hourly median oxygen mol m-3 oxy_sem Hourly standard error of oxygen mol m-3 poc_med Hourly median particulate organic carbon mol m-3 poc_sem Hourly standard error of particulate organic carbon mol m-3 region Name of data subset (e.g., 30-40 deg N, co-located) Data for Fig. 3-4 and S2-S4 | Description for lat_npp_profiles.csv, seas_npp_profiles.csv, npp_profiles_14c_compare, lat_npp_profiles_coloc.csv, npp_profile_coloc.csv, npp_profile_coloc_30_70.csv, and all files in the folder 'Bootstrapped Results'. Variable Description Units region Name of data subset (e.g., 30-40 deg N) depth Depth of profile m zeu 1% euphotic depth from Lee et al. (2013) algorithm from NASA (2022) L3 satellite products. m n_profiles_bpp Number of backscatter profiles n_profiles_oxy Number of oxygen profiles n_floats_bbp Number of floats with backscatter measurements n_floats_oxy Number of floats with oxygen measurements gop_do Gross oxygen productivity estimated from dissolved oxygen mol m-3 yr-1 gop_do_serr Standard error of gross oxygen productivity estimated from dissolved oxygen mol m-3 yr-1 gop_do_p p-value of curve fit to hourly oxygen data gop_do_r2 r-squared value of curve to hourly oxygen data oxy_sr The calculated sunrise time as a fraction of the day oxy_ss The calculated sunset time as a fraction of the day gpp_bbp Gross carbon productivity estimated from optical backscatter mol m-3 yr-1 gpp_bbp_serr Standard error of gross carbon productivity estimated from optical backscatter mol m-3 yr-1 gop_do_p p-value of curve fit to hourly particulate organic carbon data gop_do_r2 r-squared value of curve to hourly particulate organic carbon data gop_bbp Gross oxygen productivity calculated from gross carbon productivity (gpp_bbp) mol m-3 yr-1 gop_bbp_serr Standard error of gross oxygen productivity calculated from gross carbon productivity (gpp_bbp_serr) mol m-3 yr-1 npp_bbp Net primary productivity calculated from backscatter-based gross oxygen productivity (gop_bbp) mol m-3 yr-1 npp_bbp_serr Standard error of net primary productivity calculated from backscatter-based gross oxygen productivity (gop_bbp_serr) mol m-3 yr-1 npp_do Net primary productivity calculated from oxygen-based gross oxygen productivity (gop_do) mol m-3 yr-1 npp_do_serr Standard error of net primary productivity calculated from oxygen-based gross oxygen productivity (gop_do_serr) mol m-3 yr-1 Data for Fig. S1 | Description for number_of_bbp_profiles_in_each_year.csv and number_of_oxy_profiles_in_each_year.csv Variable Description Units year Year bbp470 Number of backscatter profiles oxygen_anom Number of oxygen profiles Data for Fig. S2 | Description for npp_db_mean.csv. Variable Description Units mid_depth Depth of NPP profile m mean Mean volumetric 14C-NPP at depth mmol m-3 yr-1 median Median volumetric 14C-NPP at depth mmol m-3 yr-1 min Minimum volumetric 14C-NPP at depth mmol m-3 yr-1 maximum Maximum volumetric 14C-NPP mmol m-3 yr-1 Data for Fig. S5 | Description for bootstrap_mean.csv and bootstrap_std.csv. In bootstrap_mean.csv, the variables below represent the mean of the 1000 iterations describe below, while for bootstrap_std.sv, the variables below represent the standard deviation of the 1000 iterations. Variable Description Units subset Number of profiles randomly sampled from the co-located dataset int_npp_do Euphotic-depth-integrated net primary productivity calculated from oxygen-based gross oxygen productivity mol m-2 y-1 int_npp_bbp Euphotic-depth-integrated net primary productivity calculated from backscatter-based gross oxygen productivity mol m-2 y-1 gop_do_r2 R-squared of the sinusoidal curve to the diel cycle of oxygen anomaly gpp_bbp_r2 R-squared of sinusoidal curve to the diel cycle of particulate organic carbon Data from ETOPO5 | The etopo5.nc file was used to create a 2000 m bathymetry mask (called bathymetry.shp), both located in the folder 'Topography Mask'. This data is from NOAA (1988). Variable Description Units ROSE Topographic (negative values are below sea level) m ETOPO05_Y Latitude degN ETOPO05_X Longitude degE Data from 14C-NPP Database | Description for data in the folder '14C-NPP SO Data'. This data is a subset of the databases from Marra et al. (2021) and Mattei and Scardi (2021) Variable Description Units database Database the data was extracted from Month Month of NPP measurement month of year npp_14c Net primary productivity estimated from the radiocarbon method mmol m-3 y-1 depth depth of 14C-NPP measurement m

### 数据集说明
本数据集包含处理后的BGC-Argo浮标（BGC-Argo float）数据、绘图数据、放射性碳生产力子集、自助抽样分析结果，以及用于基于光学后向散射与溶解氧日周期计算净初级生产力的配套Python/Matlab代码。

本研究使用的原始浮标数据可从法国布雷斯特、美国加利福尼亚州蒙特雷的Argo全球数据汇编中心（Argo Global Data Assembly Centers）获取，对应链接分别为：ftp://ftp.ifremer.fr/ifremer/argo/dac/coriolis 与 ftp://usgodae.org/pub/outgoing/argo/dac/coriolis。

原始MODIS卫星反演生产力数据可从俄勒冈州立大学海洋生产力网站获取：http://orca.science.oregonstate.edu/npp_products.php。

原始MODIS卫星反演真光层深度估算数据可从NASA L3浏览器获取：https://oceancolor.gsfc.nasa.gov/l3/。

本研究使用的原始船测净初级生产力估算数据可从Pangaea数据库（https://doi.pangaea.de/10.1594/PANGAEA.932417）与英国海洋学数据中心（https://www.bco-dmo.org/dataset/814803）获取。

引用格式如下：Stoer, A. 与 Fennel, K. 2022. 《基于剖面浮标碳生物量日周期估算海洋净初级生产力的处理方法与数据》. Zenodo. doi: 10.5281/zenodo.6977161。

---

### Python/Matlab软件说明
1. `dielFit_GOPeqCR.m`：该代码源自Johnson与Bif（2021）的研究，我们新增了线性模型、PvE模型的标准误差输出以及日出/日落时间参数。若需结合配套Python代码运行此脚本，需安装Matlab引擎，详情请参阅：https://www.mathworks.com/help/matlab/matlab-engine-for-python.html。
2. `argo_so_processing_20220815.py`：该代码是两套用于南大洋浮标净初级生产力估算软件中的第一部分。本程序从BGC-Argo数据库（Argo, 2021）获取数据并进行处理，包含基础数据质量控制、插值、生物地球化学计算与数据分箱操作。处理后的浮标数据存储于'Processed Argo Transects'文件夹中。
3. `argo_daily_npp_20220815.py`：该代码基于包含溶解氧与颗粒后向散射测量数据的处理后Argo浮标数据推断净初级生产力。代码会筛选出在其完整观测周期内覆盖当日所有当地时段的浮标数据，通过计算每小时观测值的中位数构建日周期序列，并调用Johnson与Bif（2021）的代码（该代码为Barone等（2019）方法的修改版本）。将颗粒后向散射转换为颗粒有机碳的算法源自Graff等（2015）的研究。本研究假设溶解初级生产力占总初级生产力的30%（Moran等，2022）。
4. `argo_daily_npp_bootstrap_20220815.py`：该代码基于包含同步匹配的溶解氧与颗粒后向散射测量数据的处理后Argo浮标数据推断净初级生产力。本代码与`argo_daily_npp_20220815.py`高度相似，但在计算净初级生产力前，会针对不同抽样规模随机抽取同步剖面子集，并针对每个抽样规模重复计算1000次生产力。分析结果存储于'Bootstrapped Results'文件夹中，详细说明请参阅代码本身。

---

### 数据说明
#### 'Processed Argo Transects'文件夹数据 | 各变量说明
| 变量 | 说明 | 单位 |
|------|------|------|
| depth | 深度箱平均深度 | m |
| mid_depth | 深度箱中心深度 | m |
| pressure | 深度箱平均压力 | dbar |
| profile_index | 剖面编号或索引 | - |
| profile_longitude | 剖面平均经度 | degE |
| profile_latitude | 剖面平均纬度 | degN |
| profile_time | 剖面平均UTC时间 | yyyy-mm-dd hh:mm:ss |
| profile_local_time | 剖面平均当地时间 | yyyy-mm-dd hh:mm:ss |
| profile_local_hour | 当地时间戳对应的小时数 | - |
| salinity | 海水盐度 | PSU |
| temperature | 海水温度 | degC |
| oxygen | 溶解氧浓度 | umol kg-1 |
| oxygen_saturation | 基于Garcia与Gordon（1992）公式计算的饱和溶解氧浓度 | umol kg-1 |
| oxygen_anom | 观测溶解氧浓度与饱和溶解氧浓度的差值 | umol kg-1 |
| bbp470 | 470 nm处的光学后向散射系数，颗粒有机碳计算详见`argo_daily_npp_20220815.py` | m-1 |

#### 图1所用数据 | bbp_profile_locations.csv与oxy_profile_locations.csv变量说明
| 变量 | 说明 | 单位 |
|------|------|------|
| wmo | 浮标的WMO编号 | - |
| profile_index | 浮标采集的剖面索引或编号 | - |
| profile_latitude | 剖面平均纬度 | degN |
| profile_longitude | 剖面平均经度 | degE |

#### 图2所用数据 | curve_fits.csv与hourly_means.csv变量说明
| 变量 | 说明 | 单位 |
|------|------|------|
| fod | 当日时段占比 | - |
| oxy | 溶解氧的正弦曲线拟合值 | mol m-3 |
| poc | 颗粒有机碳的正弦曲线拟合值 | mol m-3 |
| oxy_med | 每小时溶解氧中位数 | mol m-3 |
| oxy_sem | 溶解氧每小时标准误差 | mol m-3 |
| poc_med | 每小时颗粒有机碳中位数 | mol m-3 |
| poc_sem | 颗粒有机碳每小时标准误差 | mol m-3 |
| region | 数据子集名称（例如：30-40°N，同步匹配数据集） | - |

#### 图3-4及补充图S2-S4所用数据 | 相关csv文件及'Bootstrapped Results'文件夹文件变量说明
涉及文件包括：lat_npp_profiles.csv、seas_npp_profiles.csv、npp_profiles_14c_compare、lat_npp_profiles_coloc.csv、npp_profile_coloc.csv、npp_profile_coloc_30_70.csv，以及'Bootstrapped Results'文件夹内所有文件。
| 变量 | 说明 | 单位 |
|------|------|------|
| region | 数据子集名称（例如：30-40°N） | - |
| depth | 剖面深度 | m |
| zeu | 基于NASA（2022）L3卫星产品、Lee等（2013）算法计算的1%光强真光层深度 | m |
| n_profiles_bpp | 后向散射剖面数量 | - |
| n_profiles_oxy | 溶解氧剖面数量 | - |
| n_floats_bbp | 具备后向散射测量数据的浮标数量 | - |
| n_floats_oxy | 具备溶解氧测量数据的浮标数量 | - |
| gop_do | 基于溶解氧估算的总氧气生产力 | mol m-3 yr-1 |
| gop_do_serr | 基于溶解氧估算的总氧气生产力标准误差 | mol m-3 yr-1 |
| gop_poc_p | 逐小时颗粒有机碳数据曲线拟合的p值 | - |
| gop_poc_r2 | 逐小时颗粒有机碳数据曲线拟合的决定系数（R²） | - |
| oxy_sr | 计算得到的日出时间（以当日占比表示） | - |
| oxy_ss | 计算得到的日落时间（以当日占比表示） | - |
| gpp_bbp | 基于光学后向散射估算的总碳生产力 | mol m-3 yr-1 |
| gpp_bbp_serr | 基于光学后向散射估算的总碳生产力标准误差 | mol m-3 yr-1 |
| gop_bbp | 基于总碳生产力（gpp_bbp）计算得到的总氧气生产力 | mol m-3 yr-1 |
| gop_bbp_serr | 基于总碳生产力（gpp_bbp）计算得到的总氧气生产力标准误差 | mol m-3 yr-1 |
| npp_bbp | 基于后向散射总氧气生产力（gop_bbp）计算得到的净初级生产力 | mol m-3 yr-1 |
| npp_bbp_serr | 基于后向散射总氧气生产力（gop_bbp）计算得到的净初级生产力标准误差 | mol m-3 yr-1 |
| npp_do | 基于溶解氧总氧气生产力（gop_do）计算得到的净初级生产力 | mol m-3 yr-1 |
| npp_do_serr | 基于溶解氧总氧气生产力（gop_do）计算得到的净初级生产力标准误差 | mol m-3 yr-1 |

#### 补充图S1所用数据 | number_of_bbp_profiles_in_each_year.csv与number_of_oxy_profiles_in_each_year.csv变量说明
| 变量 | 说明 | 单位 |
|------|------|------|
| year | 年份 | - |
| bbp470 | 后向散射剖面数量 | - |
| oxygen_anom | 溶解氧异常剖面数量 | - |

#### 补充图S2所用数据 | npp_db_mean.csv变量说明
| 变量 | 说明 | 单位 |
|------|------|------|
| mid_depth | 净初级生产力剖面的深度 | m |
| mean | 该深度下的体积加权14C-NPP均值 | mmol m-3 yr-1 |
| median | 该深度下的体积加权14C-NPP中位数 | mmol m-3 yr-1 |
| min | 该深度下的体积加权14C-NPP最小值 | mmol m-3 yr-1 |
| maximum | 该深度下的体积加权14C-NPP最大值 | mmol m-3 yr-1 |

#### 补充图S5所用数据 | bootstrap_mean.csv与bootstrap_std.csv变量说明
其中bootstrap_mean.csv中的变量代表1000次迭代结果的均值，bootstrap_std.csv中的变量代表1000次迭代结果的标准差。
| 变量 | 说明 | 单位 |
|------|------|------|
| subset | 从同步匹配数据集中随机抽取的剖面数量 | - |
| int_npp_do | 基于溶解氧总氧气生产力计算得到的真光层积分净初级生产力 | mol m-2 y-1 |
| int_npp_bbp | 基于后向散射总氧气生产力计算得到的真光层积分净初级生产力 | mol m-2 y-1 |
| gop_do_r2 | 溶解氧异常日周期正弦曲线拟合的决定系数（R²） | - |
| gpp_bbp_r2 | 颗粒有机碳日周期正弦曲线拟合的决定系数（R²） | - |

#### ETOPO5地形数据
本数据集使用etopo5.nc文件生成了2000米等深线地形掩膜文件（bathymetry.shp），二者均存储于'Topography Mask'文件夹中。该数据源自美国国家海洋和大气管理局（NOAA, 1988）。
| 变量 | 说明 | 单位 |
|------|------|------|
| ROSE | 地形高程（负值代表海平面以下） | m |
| ETOPO05_Y | 纬度 | degN |
| ETOPO05_X | 经度 | degE |

#### 14C-NPP数据库数据 | '14C-NPP SO Data'文件夹数据说明
该数据集为Marra等（2021）与Mattei及Scardi（2021）数据库的子集。
| 变量 | 说明 | 单位 |
|------|------|------|
| database | 数据提取来源的数据库名称 | - |
| Month | 净初级生产力测量的月份 | 月 |
| npp_14c | 基于放射性碳法估算的净初级生产力 | mmol m-3 y-1 |
| depth | 14C-NPP测量的深度 | m |