Abstract: The amount of water in the soil is a critical determinant in many complex processes of the Earth system. Model-simulated soil moisture has been widely used to understand these processes attributed to its large spatial and long temporal coverage at any desirable location and time. However, it is known that land surface models are strongly limited in their ability to reproduce observed soil moisture, often with biases in the mean, dynamic range, and time variability. In this study, we presented a cost-effective application of variance-based sensitivity analysis to quantify the relative contribution of different parameters and their interactions to the overall uncertainty in the modeled surface and root zone soil moisture from the Community Land Model 5.0 (CLM5). We focus on four parameters associated with the hydraulic property of mineral soil (saturated hydraulic conductivity, porosity, saturated soil matric potential, and shape-parameter) and organic matter fraction. A Gaussian process emulator is used to estimate the soil moisture across the five-dimensional parameter uncertainty space, based on a small number of CLM5 simulations at combinations of parameter values sampled with Maximin Latin hypercube. The procedure is exemplified for four seasons (DJF, MAM, JJA, and SON) across various sites of distinct soil and vegetation types in the continental US. Our results have shown that the emulator captures well the behavior of CLM5 across the entire parameter uncertainty space for different soil textures and seasons, with high correlations and low RMSEs between the emulator-predicted and CLM5-simulated soil moisture as well as small emulator uncertainty. We found that the large portion of the variances of both surface and root zone soil moisture is described by uncertainty in five parameters (excluding their interactions) and is dominated by the uncertainty in porosity and shape parameter for almost all the sites and seasons. Generally, the lower the fraction of sand is, the stronger (weaker) the individual parameter effects (the interaction effects) are. However, the relative importance of porosity versus shape parameter varies strongly with variables (surface versus root zone), soil textures (sites), and seasons. Over the majority of sites, the variance in surface soil moisture is attributed distinctly more to the uncertainty in shape parameter, while the uncertainty in porosity is more important in the variance of root zone soil moisture. Also, both individual parameter effects and interaction effects for root zone soil moisture demonstrate less variability across different soil textures and seasons than for surface soil moisture. These sensitivity results clearly indicate which parameters should be focused on to improve the model simulations of surface versus root zone soil moisture for different soil textures and seasons, which serves as a useful guidance to achieve improved modeling of soil moisture on a large scale.

Summary (MIT News): Historically, the oceans have done much of the planet’s heavy lifting when it comes to sequestering carbon dioxide from the atmosphere. Microscopic organisms known collectively as phytoplankton, which grow throughout the sunlit surface oceans and absorb carbon dioxide through photosynthesis, are a key player. To help stem escalating carbon dioxide emissions produced by the burning of fossil fuels, some scientists have proposed seeding the oceans with iron — an essential ingredient that can stimulate phytoplankton growth. Such “iron fertilization” would cultivate vast new fields of phytoplankton, particularly in areas normally bereft of marine life. A new MIT study suggests that iron ferilization may not have a significant impact on phytoplankton growth, at least on a global scale.