SPIE Digital Library Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
    Advertisers
SPIE Photonics West 2017 | Register Today

SPIE Defense + Commercial Sensing 2017 | Call for Papers

Journal of Medical Imaging | Learn more

SPIE PRESS




Print PageEmail PageView PDF

Electronic Imaging & Signal Processing

Monitoring global precipitation using satellites

A multisatellite image-processing system can improve near-real-time precipitation measurement in remote, ungauged areas.
19 October 2012, SPIE Newsroom. DOI: 10.1117/2.1201210.004475

Floods caused by extreme precipitation are one of the most frequent and widespread natural hazards. They are more costly and dangerous than ever, as population in urban areas increases and the global climate becomes more extreme and variable. Data shows that each year there are more than a hundred million people affected by flood events with a cost of more than $100 billion.1 While accurate precipitation monitoring is a key element for improving flood forecasting, traditional means of precipitation observation, such as ground-based gauges and radars, are limited in their spatial coverage. Recent advances in satellite remote sensing techniques have enabled precipitation observation in remote and ungauged regions to help hydrologists better forecast floods and manage water resources.

Satellite-based precipitation retrieval algorithms use information from visible to longwave-IR images from Geostationary Earth Orbital (GEO) satellites and passive/active microwave images from Low Earth Orbital (LEO) satellites. GEO satellites provide more frequent observations at every 15–30 minutes, but their information is only indirectly related to surface rainfall. Passive microwave (PMW) sensors on LEO satellites give more direct sensing of rain clouds. Their low sampling frequency, however, limits the effectiveness of the rainfall measurement in high temporal and spatial scales. Effective integration of LEO and GEO satellite information is critical for improving rainfall measurement at short-time scales.

Our algorithm, Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN), provides global precipitation estimates using combined GEO-IR and PMW precipitation data from multiple GEO and LEO satellites.3–6 The algorithm uses an artificial neural network to extract cold cloud pixels and their neighboring textures from the GEO-IR imagery, and associates variations in each pixel's brightness with temperature to estimate the pixel's surface rain rate. GEO-IR-based precipitation estimates are further adjusted by the PMW precipitation estimates, using the data from LEO satellites. A feedback process using the PMW estimates as the reference surface precipitation for the adjustment may further improve estimate quality.

We are currently working to improve the data quality and resolution of the multisatellite precipitation estimation algorithm. These improved features include multispectral images for cloud classification, precipitation tracking/advection, and near-real-time adjustment of estimates.6, 7 The recently developed PERSIANN Cloud Classification System (PERSIANN-CCS), for example, implements image processing and pattern classification techniques based on analysis of GEO-IR cloud images.8–10 Rainfall estimation from the PERSIANN-CCS is shown in Figure 1. First, image segmentation is used to separate cloud patches from their image background using a watershed delineation process. This is followed by extraction and classification, in which cloud patches are treated as independent objects and described by object features such as patch coldness, size, shape, and texture. Classification of cloud-patch objects is achieved through an unsupervised self-organizing feature-map clustering scheme. When patch rainfall is assigned to a classified patch group, it establishes an interpretative relationship between the cloud-patch property and rainfall. The classified patch group's rainfall distribution is described by a set of GEO-IR brightness temperature and rainfall rate (Tb-R) functions. Parameters of the nonlinear Tb-R functions are calibrated from the spatial and temporal co-located satellite image and radar/PMW rainfall images.


Figure 1. The cloud image segmentation, feature extraction, classification, and rainfall estimation of the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Cloud Classification System (PERSIANN-CCS) algorithm. Geostationary satellites used include Geostationary Operational Environment Satellites (GOES) 8 and 10, Geostationary Meteorological Satellite (GMS), and Meteosat 6 and 7. Polar and near-polar instruments contributing data include Precipitation Radar (PR), Tropical Rainfall Measurement Mission (TRMM), TRMM Microwave Imager (TMI), Special Sensor Microwave Imager (SSMI), the Advanced Microwave Scanning Radiometer (AMSR), the NASA Earth Observation System (NASA-EOS), the AMSR for EOS (AMSR/E), the EOS Aqua satellite, weather satellites 15, 16 and 17 run by the National Oceanic and Atmospheric Administration (NOAA) and weather satellites F13, F14 and F15 run by the Defense Meteorological Satellite program (DMSP). Tb-P: Brightness-temperature and rainfall rate. ANN: Artificial neural network.

The Center for Hydrometeorology and Remote Sensing (CHRS) at the University of California, Irvine—with support from NASA, the National Oceanic and Atmospheric Administration (NOAA), the US Army Research Office, and the United Nations Educational Scientific and Cultural Organization's International Hydrological Program (IHP)—has been building capacity for forecast and mitigation of hydrologic disasters. A near-real-time global precipitation server2 provides visualization and mapping of satellite-based precipitation estimates at 0.04° × 0.04° latitude-longitude spatial resolution using real-time implementation of the PERSIANN-CCS (see Figure 2). The interface to PERSIANN-CCS data allows for ease of access, rapid image rendering, interface simplicity, hydrologically relevant functionalities, and portability to other mirror sites. To assist the user's interaction with the site, several multilanguage tutorials have been made available on the YouTube11 and Hydis12 websites.


Figure 2. University of California, Irvine, Center for Hydrometeorology and Remote Sensing (CHRS) Global Network for Water and Development Information for Arid Lands server for monitoring near-real-time global precipitation distribution.2

PERSIANN-CCS produces near-real time precipitation data, which is useful for operational flood forecasting and hydrologic data assimilations. PERSIANN products have been applied to a number of hydrologic studies, including validation of daily rainfall and diurnal rainfall patterns against observations provided by the field campaign measurement over the Amazon region; comparison of mean areal precipitation with rain gauge and Nexrad estimates; evaluation of the MM5 (Fifth-Generation Penn State/NCAR Mesoscale Model) numerical weather-forecast model estimates over the southwest US, Mexico, and adjacent oceanic regions; assimilation into a Regional Atmospheric Modeling System model for the southwestern US to investigate land-surface hydrologic processes including soil moisture, merging gauge and PERSIANN data over Mexico; and application of PERSIANN rainfall to the runoff forecasting of several basins in the US and after severe storm events such as hurricane Katrina (see Figure 3).


Figure 3. Three-hour accumulated rainfall observed from PERSIANN-CCS during the time of Hurricane Katrina's landfall along the coastlines of Louisiana, Mississippi, and Alabama.

PERSIANN-CCS is effective in providing near-real-time precipitation estimation at 0.04° × 0.04° latitude-longitude resolution covering the area of 60°S to 60°N. Rainfall measurements are critical for a variety of applications, especially for flood forecasting. Our future work will extend satellite-based algorithms' capabilities to capture warmer rainfall, to employ multispectral images, and to include recent developments in cloud development modeling, satellite cloud-feature extraction, cloud image tracking, geostatistics and sequential filtering theory for rainfall measurement.7–10, 13

Funding was provided by NASA Precipitation Measurement Mission grant NNX10AK07G, NOAA Climate Change Data & Detection grant NA10OAR4310122, and US Army Research Office grant W911NF-11-1-0422.


Kuolin Hsu, Soroosh Sorooshian, Xiaogang Gao, Dan Braithwaite, Amir AghaKouchak
University of California, Irvine
Irvine, California

Kuolin Hsu is an associate professor in the Department of Civil and Environmental Engineering and CHRS. His research interests include remote sensing of precipitation, hydrology modeling, and data assimilation of hydrologic systems.

Soroosh Sorooshian is a distinguished professor in the departments of Civil and Environmental Engineering and Earth System Science, and director of CHRS. His areas of expertise are hydrometeorology and climatology, water resources engineering, hydrologic modeling, and application of remote sensing to earth science problems.

Xiaogang Gao is an adjunct professor in the Department of Civil and Environmental Engineering. His research interests include environmental remote sensing, climate and hydrology, and water resource management.

Dan Braithwaite is a programmer analyst in the Department of Civil and Environmental Engineering and CHRS. His interests include developing online data visualization, manipulation, and acquisition tools along with computer systems administration.

Amir AghaKouchak is assistant professor of Civil and Environmental Engineering. His principal research interests include stochastic modeling, remote sensing, extreme value analysis, and hydroclimatology.


References:
1. D. Singh, Unplanned urbanization increasing flood impacts. UNISDR news brief, 9 August 2012. http://www.unisdr.org/archive/27965
2. http://hydis.eng.uci.edu/gwadi/ CHRS GWADI server. Accessed 3 October 2012.
3. K. Hsu, X. Gao, S. Sorooshian, H. V. Gupta, Precipitation estimation from remotely sensed information using artificial neural networks, J. Appl. Meteorol. 36(9), p. 1176-1190, 1997.
4. K. Hsu, H. V. Gupta, X. Gao, S. Sorooshian, Estimation of physical variables from multiple channel remotely sensed imagery using a neural network: application to rainfall estimation, Water Resour. Res. 35(5), p. 1605-1618, 1999.
5. S. Sorooshian, K. Hsu, X. Gao, H. V. Gupta, B. Imam, Dan Braithwaite, Evaluation of PERSIANN system satellite-based estimates of tropical rainfall, Bull. Am. Meteorol. Soc. 81(9), p. 2035-2046, 2000.
6. Y. Hong, K. Hsu, X. Gao, S. Sorooshian, Precipitation estimation from remotely sensed information using an artificial neural network—cloud classification system, J. Appl. Meteorol. 43, p. 1834-1852, 2004. doi:10.1175/JAM2173.1
7. A. Behrangi, K. Hsu, B. Imam, S. Sorooshian, R. J. Kuligowski, PERSIANN-MSA: A precipitation estimation method from satellite-based multispectral analysis, J. Hydrometeorol. 10(6), p. 1414-1429, 2009. doi:10.1175/2009JHM1139.1
8. T. Bellerby, K. Hsu, S. Sorooshian, LMODEL: A satellite precipitation algorithm using cloud development modeling and model updating. Part I: model development and calibration, J. Hydrometeorol. 10(5), p. 1081-1095, 2009. doi:10.1175/2009JHM1091.1
9. K. Hsu, T. Bellerby, S. Sorooshian, LMODEL: A satellite precipitation algorithm using cloud development modeling and model updating. Part II: model updating, J. Hydrometeorol. 10(5), p. 1096-1108, 2009. doi:10.1175/2009JHM1092.1
10. A. Behrangi, B. Imam, K. Hsu, S. Sorooshian, T. J. Bellerby, G. J. Huffman, Rain estimation using forward adjusted-advection of microwave estimates (REFAME), J. Hydrometeorol. 11, p. 1305-1321, 2010. doi:10.1175/2010JHM1248.1
11. http://www.youtube.com/user/CHRSUCIrvine YouTube GWADI tutorial. Accessed 3 October 2012.
12. http://persiann.eng.uci.edu/gwadi_tutorial_videos.html Hydis GWADI tutorial. Accessed 3 October 2012.
13. S. Sorooshian, A. AghaKouchak, P. Arkin, J. Eylander, E. Foufoula-Georgiou, R. Harmon, J. Hendrickx, Advanced concepts on remote sensing of precipitation at multiple scales, Bull. Am. Meteorol. Soc. 92(10), p. 1353-1357, 2011. doi:10.1175/2011BAMS3158.1