This repo contains the data and code used in the paper "Preferential Phosphorus Retention in Lakes Exacerbates the Imbalance of Global Nutrient Cycles"

NLA2012_data.csv contains the data used in NLA2012 model.

Global_data.csv is a dataset of 5,622 lakes as an intersection of Chlorophyll and Water Chemistry database and HydroLAKES database.

NLAmodel.stan is the stan code for NLA2012 model.

NLAmodel.ipynb runs the stan code in NLAmodel.stan using data in NLA2012_data.csv and generate model output in NLAmodel_output.csv.

global_data_clean.R is used for the data cleaning. Cleaned data is in folder global results/ named as global_data_clean.csv.

glm_N.R and glm_P.R are code for GLM models to relate EN and DE of N and P with multiple explanatory variables including water residential time, surface water temperature, water depth, area and trophic index. Statistics are generated in the folder fitting statistics/. Then cleaned global data is used to predict the EN and DE of N and P of the global lakes and placed in the folder global results/, namely global_N.csv and global_P.csv respectively.

upscaling.ipynb contains the code for Monte Carlo approach to predict the internal nutrient cycling fluxes of global lakes from the distribution of the nutrient budget model results.