Global Sensitivity Analysis of the LPJ Model for Larix olgensis Henry Forests NPP in Jilin Province, China

Parameter sensitivity analysis can determine the influence of input parameters on model output. Identification and calibration critical are crucial points process optimization. Based Extended Fourier Amplitude Sensitivity Test (EFAST) Morris method, this paper analyzes compares parameter annual mean net primary productivity (NPP) Larix olgensis Henry forests in Jilin Province simulated by Lund–Potsdam–Jena dynamic global vegetation (LPJ model) 2009–2014 2000–2019, deeply examines two methods to each their respective model’s Moreover, it optimizes some selected re-simulates NPP from 2010 2019. The conclusions following: (1) For LPJ model, sensitive non-influential could be identified, which guide optimization order was valuable for area applications. (2) results were similar but not identical. significantly correlated (p < 0.05); krp most parameter, followed αm, αa gm. These mainly found photosynthesis, water balance, allometric growth modules. (3) EFAST method had a higher precision than calculate quantitatively contribution rate variances results; however, involved fewer running times efficiency. (4) relative error (MRE) absolute (MAE) value after decreases. optimized 2019 580 g C m−2 a−1, with 2.13%, exhibiting fluctuating trend. MAE