Fast spatio-temporal interpolation modules for photochemical grid based air quality models
S. Isukapalli, V. Vyas, P.G. Georgopoulos
Environmental and Occupational Health Sciences Institute, UMDNJ - R.W. Johnson Medical School and Rutgers University, Piscataway, NJ
Ambient air quality information from photochemical air quality models is often needed at resolutions higher than provided by regional scale models, such as the Models-3/Community Multiscale Air Quality (CMAQ). This is especially true for exposure assessment modeling, which requires ambient concentrations at census tract or census block levels, while the air quality models provide estimates at grid cells larger than 2 km x 2 km. Techniques such as the Spatio-Temporal Random Field (STRF) and the Bayesian Maximum Entropy (BME) have been recently used for subgrid interpolation of photochemical model estimates. These techniques are general in nature, and can be applied to both monitor data and model outputs, while not requiring a regular grid for the interpolation. However, there are significant performance limitations in using the current implementations of STRF and BME for the interpolation of massive datasets, such as the CMAQ outputs from an annual simulation; often the computational time for interpolation is of the same order of magnitude as the CMAQ simulation itself. Furthermore, there are several pre-processing steps involved in the application of the STRF and BME: (a) the CMAQ outputs have to be processed and transformed, (b) the outputs need to be manually split into several large data sets, as the current implementations cannot handle such large data sets. This work presents recent improvements of the STRF modules available in the MENTOR (Modeling ENvironment for TOtal Risk studies) system, and provides modules for fast spatio-temporal interpolation on regular grids. These modules are designed to operate directly upon the standard formats of model outputs from several photochemical grid models, and are demonstrated by interpolating CMAQ model outputs at different scales.
This work is funded in part by a University Partnership Agreement between USEPA and EOHSI. Viewpoints expressed here do not necessarily reflect the views of USEPA or its contractors.