Simulating summer mixing heights in California's San Joaquin Valley using the WRF meteorological model with three land surface modules

Abstract

During the 1990 SARMAP/AUSPEX study a network of rawinsondes was flown at 3- and 6-h intervals to evaluate the meteorological conditions conducive high ozone concentrations in the Central Valley of California. Vertical profiles of air temperature, humidity, and pressure from 9 of the sites were used to determine maximum daily mixing heights in California’s San Joaquin Valley (SJV) during the period August 3–6, 1990 and to evaluate their spatial distribution. The analysis showed a general pattern of increasing mixing heights from north to south within the SJV. But, it also showed that mixing heights were lower in areas in proximity to irrigated agriculture. The WRF prognostic meteorological model was used to simulate mixing heights during this period using three land surface modules (LSM): thermal diffusion (TD), Noah, and Pleim-Xiu (PXU). The maximum daily mixing heights simulated using the TD LSM were generally negatively biased; while those simulated using the Noah and PXU LSMs were positively biased. The differences in simulated mixing heights were attributed to differences in how crop land irrigation water is represented within each LSM. The TD LSM effectively assumes a fixed soil water content, but of sufficient magnitude to reflect latent heat fluxes associated with crop land irrigation. The PXU and Noah LSMs include comprehensive descriptions of the soil water budget. However, soil water may be depleted in the arid summer climate and these LSMs provide no direct means of accounting for water added through irrigation. The LSMs were also sensitive to estimates of ground cover in irrigated areas. For all 9 sites, the root mean square error (RMSE) for maximum daily mixing heights for the TD LSM, the Noah LSM, and the PXU LSM was 269, 350, and 366 m, respectively; and the biases for the 9 sites were − 161, 355, and 437 m, respectively. Using the assumption of well-managed irrigation and leaf area index (LAI) estimates from AVHRR remote sensing data sets, adjustments to the WRF input files were made to update soil water contents and ground cover for irrigated crop land use. Using the Noah LSM with the adjusted files, site-specific mixing height biases were reduced up to a factor of 4 and, for all sites, the RMSE for maximum daily mixing heights were reduced from 350 to 260 m; for the PXU LSM RMSEs were reduced from 366 to 238 m; and for the TD LSM from 269 to 231 m. This analysis illustrates the importance of accounting for the effects of applied irrigation water and ground cover in the simulation of mixing heights during the arid summer within the SJV. Irrigation can influence actual mixing heights and their distribution within a domain. It also illustrates that the choice of LSM is of less importance than setting LSM parametrizations to values appropriate for the modeling domain.