Biomass burning is an important source of many atmospheric trace gases and aerosol particles. Quantitative characterization of biomass burning emissions is critical for modeling atmospheric chemistry and assessing the impact of fires on air quality, tropospheric ozone chemistry, and global climate. However, advancement in quantifying the emissions from fires spatially and temporally has been limited. We have developed a method for quantifying near real-time smoke emissions in a 1-km x 1-km resolution in North America based on MODIS data. The Fire Sciences Laboratory has been equipped with a satellite receiving station to retrieve, process, and archive real-time MODIS data. Our MODIS images cover most of the continental U.S., Alaska, Canada and northern Mexico. The MODIS algorithms of active fire detection have been validated by comparing the MODIS detected fire locations with the ground survey data in the 2002 National Fire Occurrence Database. MODIS detected about half of the fires smaller than 6 sq. km and detected about 80% of the fires larger than 6 sq. km, which account for 99% of the total area burned in the continental U.S. The fire detection by MODIS is significantly more accurate than the NOAA AVHRR satellite, especially in the grassland region. In addition to validating the MODIS fire detection algorithms, we have implemented a set of algorithms using 1.24μm and 2.13 μm spectral bands to map burned areas in a resolution of 500m x 500m. The algorithms were validated by comparing the MODIS derived burned areas with the fire perimeters mapped by the Forest Service's airborne IR radiometers. Combining the real-time active fire locations and burned areas from MODIS with a static fuel map and a fuel consumption model FOFEM, we quantified the CO and PM2.5 emissions every six hours from the I-90 Fire in Missoula, Montana from August 4 to 23, 2005. We also examined the impact of the I-90 Fire on regional air quality.