Sensitivity of Numerical Weather Forecasts to Initial Soil Moisture Variations in CFSv2

Abstract

When initial soil moisture is perturbed among ensemble members in the operational NWS global forecast model, surface latent and sensible fluxes are immediately affected much more strongly, systematically, and over a greater area than conventional land–atmosphere coupling metrics suggest. Flux perturbations are likewise transmitted to the atmospheric boundary layer more formidably than climatology-based metrics would indicate. Impacts are not limited to the traditional land–atmosphere coupling hot spots, but extend over nearly all ice-free land areas of the globe. Key to isolating this effect is that initial atmospheric states are identical among quantities correlated, pinpointing soil moisture and snow cover. A consequence of this high sensitivity is that significant positive impacts of realistic land surface initialization on the skill of deterministic near-surface temperature and humidity forecasts are also immediate and nearly universal during boreal spring and summer (the period investigated) and persist for at least 3 days over most land areas. Land surface initialization may be more broadly important for weather forecasts than previously realized, as the research focus historically has been on subseasonal-to-seasonal time scales. This study attempts to bridge the gap between climate studies with their associated coupling assessments and weather forecast time scales. Furthermore, errors in land surface initialization and shortcomings in the parameterization of atmospheric processes sensitive to surface fluxes may have greater consequences than previously recognized, the latter exemplified by the lack of impact on precipitation forecasts even though the simulation of boundary layer development is shown to be greatly improved with realistic soil moisture initialization.

On the Harvest of Predictability From Land States in a Global Forecast Model

“…Precipitation is conspicuously insensitive, as has been noted before (cf. Dirmeyer, ; Dirmeyer & Halder, , ). This result is examined further below and motivates inclusion of the HCF and NoHCF simulations in the analysis in section .…”

“…Dirmeyer and Halder (, ) focused on one operational model, the National Centers for Environmental Prediction Coupled Forecast System version 2 (CFSv2; Saha et al, ). They examined how realistic land surface initialization affected skill in a range of variables, both of practical interest (e.g., near‐surface temperature and precipitation) and that lie in the chain of processes (e.g., land surface fluxes and boundary layer height) that link land surface states to important meteorological phenomena like precipitation.…”

Simulation of surface energy fluxes and meteorological variables using the Regional Atmospheric Modeling System (RAMS): Evaluating the impact of land-atmosphere coupling on short-term forecasts

“…Considering all these interactions and feedbacks, different studies have been performed with the aim of examining and studying the land and atmosphere conditions and interactions (see e.g. De Goncalves et al, 2006;Koster et al, 2009;Betts, 2009;Seneviratne et al, 2010;Dirmeyer et al, 2012;Dirmeyer et al, 2013;Ferguson et al, 2012;Santanello et al, 2013;Zaitchik et al, 2013;Meng and Shen, 2014;Pathirana et al, 2014;Hirsch et al, 2014;Daniels et al, 2015;Dirmeyer and Halder, 2016). All these studies have highlighted the significant importance of soil parameters, specially focused on the relationship between the SM field and sensible weather parameters.…”

Improved meteorology and surface energy fluxes in mesoscale modelling using adjusted initial vertical soil moisture profiles

“…This crucial significance of the soil moisture conditions on weather forecasts has been stated in a number of previous studies, both at short and medium range, and using different Numerical Weather Prediction (NWP) modelling environments (LeMone et al, 2007;van den Hurk et al, 2008;Hong et al, 2009;Angevine et al, 2014;de Rosnay et al, 2014;Daniels et al, 2015;Daniels et al, 2016;Gómez et al, 2015b;Lin and Cheng, 2016;Dillon et al, 2016;Dirmeyer and Halder, 2016;Gómez et al, 2016b;Kalverla et al, 2016).…”