We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Machine Learning Accelerates Parameter Optimization and Uncertainty Assessment of a Land Surface Model.
- Authors
Sawada, Yohei
- Abstract
The performance of land surface models (LSMs) significantly affects the understanding of atmospheric and related processes. Many of the LSMs' soil and vegetation parameters were highly uncertain so that it is crucially important to efficiently optimize them. Here I present a globally applicable and computationally efficient method for parameter optimization and uncertainty assessment of the LSM by combining Markov Chain Monte Carlo (MCMC) with machine learning. First, I performed the long‐term (decadal‐scale) ensemble simulation of the LSM, in which each ensemble member has different parameters' values, and calculated the gap between simulation and observation, or the cost function, for each ensemble member. Second, I developed the statistical machine learning‐based surrogate model, which is computationally cheap but accurately mimics the relationship between parameters and the cost function, by applying the Gaussian process regression to learn the model simulation. Third, I applied MCMC by repeatedly evaluating the surrogate model to get the posterior probabilistic distribution of parameters. Using satellite passive microwave brightness temperature observations, both synthetic and real‐data experiments in the Sahel region of west Africa were performed to optimize unknown soil and vegetation parameters of the LSM. The primary findings are (1) the proposed method is 50,000 times as fast as the direct application of MCMC to the full LSM; (2) the skill of the LSM to simulate both soil moisture and vegetation dynamics can be improved; and (3) I successfully quantify the characteristics of equifinality by obtaining the full nonparametric probabilistic distribution of parameters. Key Points: Machine learning realizes a globally applicable method for parameter optimization and uncertainty assessment of the LSMsNonparametric probabilistic distribution of parameters can be obtained by assimilating satellite observation
- Subjects
MACHINE learning; UNCERTAINTY; HYDROLOGICAL databases; PROBABILISTIC databases; NONPARAMETRIC statistics
- Publication
Journal of Geophysical Research. Atmospheres, 2020, Vol 125, Issue 20, p1
- ISSN
2169-897X
- Publication type
Article
- DOI
10.1029/2020JD032688