- Biomedical Optics & Medical Imaging
- Defense & Security
- Electronic Imaging & Signal Processing
- Illumination & Displays
- Lasers & Sources
- Micro/Nano Lithography
- Optical Design & Engineering
- Optoelectronics & Communications
- Remote Sensing
- Sensing & Measurement
- Solar & Alternative Energy
- Sign up for Newsroom E-Alerts
- Information for:
Detecting smoke over land using imaging satellite data from space
A multispectral approach that relies on remote-sensing measurements by NASA's Moderate Resolution Imaging Spectroradiometer makes it possible to distinguish polluting smoke from clouds.
9 May 2010, SPIE Newsroom. DOI: 10.1117/2.1201004.002893
Smoke from biomass burning is a major pollutant that directly affects climate (a process known as forcing) by altering cloud microphysics and sunlight reflectance.1 These perturbations in turn influence atmospheric large-scale circulation and precipitation.2 In addition to its impact on climate, biomass smoke may also affect the environment, air quality, and local weather. For these reasons, tracking and monitoring smoke in a timely manner from space is both important and necessary. One effective tool for this purpose is satellite remote sensing.3–8 But most current implementations of the technique are unsuitable for bright surface areas or in regions where smoke is mixed with clouds.
We have developed a multispectral approach to detecting smoke using Moderate Resolution Imaging Spectroradiometer (MODIS) satellite remote-sensing measurements.9 We identified smoke by filtering out other types of scenes based on differences in their spectral features. A scaled index was applied to define the confidence level of smoke pixels and to characterize the smoke intensity. Our algorithm provides a quick means of automatically localizing the smoke in an entire MODIS swath. Table 1 lists the scene types, threshold tests, and thresholds used for smoke detection over land. In the table, B denotes the MODIS band, BT indicates brightness temperature, and BTD is the brightness temperature difference.
The threshold tests and thresholds used for detecting smoke over land.
|Scene type||Threshold test||Threshold|
| ||B26 and||0.035|
|Cloud||BTD (31, 22) and||−10 K|
| ||BT31||293 K|
|Vegetation and water||and B8||0.17|
| ||and B1||0.14|
The most important step in detecting smoke is to accurately separate it from clouds. In our algorithm, three simultaneous tests examine the BT of B31 (11μm), the BTD between B31 and B22 (3.9μm), and the reflectance of B26 (1.38μm).10 These bands are insensitive to smoke but sensitive to the presence of clouds. The normalized ratio of B3 (470nm) and B7 (2.13μm) (see Table 1), whose value is scaled to [0, 1], is applied to filter out soil and bright surface pixels. The reflectance of B8 (412nm) is designed to filter out water and vegetation pixels.
The average geometrical radii of smoke particles are small, within the range of 0.01–0.05μm.11,12 According to the Rayleigh scattering theory, the two blue bands (B3 and B8) are very sensitive to the existence of smoke. Thus, we propose that the normalized ratio of these bands—ratio (B3, B8)—be considered a key parameter in the algorithm. The ratio defines the confidence level of smoke or nonsmoke pixels and serves as an indicator of smoke intensity. A close linear relationship between ratio (B3, B8) and the reflectance of B3 indicates that the stronger the smoke, the larger the ratio value (it is generally larger than −0.15). Normalization expands the ratio from [−0.15, 0.05] to [0, 1]. The pixels with ratios larger than 0.05 are treated as heavy smoke pixels and given the highest possible confidence value of 1. A 5×5 pixel box is used to single out pixel outliers based on the continuity of the smoke.
Figure 1 shows a smoke event that occurred on 1 April 2007 in Georgia (US). The image at the top left is the MODIS true-color image of the whole swath. The top right image shows smoke detection with all confidence levels. The two bottom images are enlargements of the selected area. In the smoke image (bottom right), only smoke pixels with a confidence flag larger than 0.25 are displayed. Comparing smoke and true-color images shows that most of the smoke pixels—especially the smoke core—were detected successfully. The algorithm has been applied to other regions of the world, with similarly successful results. Figure 2 presents several smoke events at different locations during 2009. In a few cases, bright surface pixels were misclassified as smoke. Some pixels located at the edge of the smoke and characterized by weak magnitude were also missed. Accurate smoke identification remains a challenge that needs to be addressed and improved in future work.
Figure 1. The Moderate Resolution Imaging Spectroradiometer (MODIS) smoke-detection image and corresponding true-color image for 1 April 2007 in Georgia (US). The bottom images are enlargements of the inset.
Figure 2. Smoke detection results at different locations at 06:05 on 22 May and 03:15 on 21 July in 2009. The left column shows the MODIS true-color images and the right column, smoke images. Only pixels with a confidence level larger than 0.25 are plotted in the smoke-detection images.
The algorithm described here provides a rapid means of automatically detecting smoke over land from space using MODIS satellite remote-sensing measurements. We propose a normalized ratio of two blue bands to characterize the smoke intensity and to define the confidence level of the results. Smoke was detected successfully in many locations using this approach. Nonetheless, as a next step, we plan to improve the algorithm by integrating multisensor measurements to detect biomass smoke more effectively and accurately.
Yong Xie, John J. Qu
George Mason University
Yong Xie is a research associate at George Mason University working on detecting smoke and dust storms, and parameter retrieval using multisensor measurements. He is also investigating sensor calibration and characterization for the Earth Observing System Terra/Aqua MODIS with the MODIS Calibration Support Team, as well as cross-sensor calibration.
Biospheric Sciences Branch
NASA Goddard Space Flight Center
Climate and Radiation Branch
NASA Goddard Space Flight Center
2. Y. Zhang, R. Fu, H. Yu, Y. Qian, R. Dickinson, M. A. F. Silva Dias, P. L. da Silva Dias, K. Fernandes, Impact of biomass burning aerosol on the monsoon circulation transition over Amazonia, Geophys. Res. Lett. 36, pp. L10814, 2009. doi:10.1029/2009GL037180
6. N. Chrysoulakis, I. Herlin, P. Prastacos, H. Yahia, J. Grazzini, C. Cartalis, An improved algorithm for the detection of plumes caused by natural or technological hazards using AVHRR imagery, Remote Sens. Environ. 108, no. 4, pp. 393-406, 2007. doi:10.1016/j.rse.2006.11.024
7. Y. J. Kaufman, C. I. Choku, L. Giglio, S. Korontzi, D. A. Chu, W. M. Hao, R.-R. Li, C. O. Justice, Fire and smoke observed from the Earth Observing System MODIS instrument-products, validation, and operational use, Int'l J. Remote Sens. 24, no. 8, pp. 1765-1781, 2003. doi:10.1080/01431160210144741
9. Y. Xie, J. J. Qu, X. Xiong, X. Hao, N. Che, W. Sommers, Smoke plume detection in the eastern United States using MODIS, Int'l J. Remote Sens. 28, no. 10, pp. 2367-2374, 2007. doi:10.1080/01431160701236795
11. L. A. Remer, D. Tanre, Y. J. Kaufman, C. Ichoku, S. Mattoo, R. Levy, D. A. Chu, B. Holben, O. Dubovik, A. Smirnov, J. V. Martins, R. R. Li, Z. Ahmad, Validation of MODIS aerosol retrieval over ocean, Geophys. Res. Lett. 29, no. 12, pp. 8008, 2002. doi:10.1029/2001GL013204