Supporting Dataset and Source Code for "Exploring Marine Biogeochemical Prediction with Linear Dynamical Frameworks"

This repository contains the codes and datasets used to reproduce the analyses presented in the paper "Exploring Marine Biogeochemical Prediction with Linear Dynamical Frameworks", submitted to [Geophysical Research Letters]. Contents Observation datasets (LIM_input_obs_*.nc):Spatially averaged observations over the study region for four physical variables — mixed layer depth (HBLT), temperature (TEMP), sea surface height (SSH), and vertically integrated primary production (photoC_TOT_zint_100m) — covering the period 1998–2024. Model datasets (LIM_input_LE_*.nc):CESM2 Large Ensemble (CESM2-LE) data averaged over the same region, including the above four variables plus nutrients (NO₃, Fe). The dataset includes 50 ensemble members, spanning 1850–2024. Code notebooks (.ipynb): LIM_extract_data.ipynb: Extracts regional data from CESM2-LE raw outputs. LIM_Climatology.ipynb: Computes and visualizes climatological means. LIM_timeseries.ipynb: Produces time-series plots and diagnostic figures used in the study. Usage These datasets and scripts allow users to reproduce the preprocessing steps and core analyses of the Linear Inverse Model (LIM) experiments described in the paper. Temporal coverage Observations: 1998–2024 CESM2-LE model: 1850–2024🔸 The full LIM code package and additional supporting data will be released in version 2 (v2) upon journal publication.