3rd IPWG Workshop on Precipitation Measurements - Melbourne, 23-27 October, 2006 Do cloud microphysical parameters derived from daytime multi-spectral satellite observations correlate with rainfall estimates? Elsa Cattani, Francesca Torricella, and Vincenzo Levizzani ISAC CNR (Italy) Institute of Atmospheric Sciences and Climate National Research Council - Bologna
22
Embed
3rd IPWG Workshop on Precipitation Measurements - Melbourne, 23-27 October, 2006 Do cloud microphysical parameters derived from daytime multi-spectral.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Statistical comparison of TMI and PR surface rain data
The original 2A12 and 2A25 rain data are mapped to 0.1° latitude –longitude grid. On average 20 PR and 27 TMI observations in each grid mesh Daytime data: 40 orbits for TMI (34,462 pixels in the common area of the swaths) and 35 PR orbits (108,406 pixels) over the area during the 10 days The rain information from PR and TMI are compared after re-projection onto this grid The whole WA is divided in macro-areasFor each macro-area the analysis is applied separately The PR 2A25 rain data are considered as rain truth data
Definitions from: WWRP/WGNE Joint Working Group on Verification Forecast Verification - Issues
Probability of detection (hit rate)* Answers the question: What fraction of the observed "yes" events were correctly forecast? Range: 0 to 1. Perfect score: 1.
False alarm ratio Answers the question: What fraction of the predicted "yes" events actually did not occur (i.e., were false alarms)? Range: 0 to 1. Perfect score: 0.
Heidke skill score (Cohen's ) where
Answers the question: What was the accuracy of the forecast relative to that of random chance? Range: minus infinity to 1, 0 indicates no skill. Perfect score: 1.
* We used both POD0 i.e. having a zero rain/no-rain boundary and POD1, having 1 mm h -1 rain/no-rain boundary
The RGB technique provides a relatively simple rendering of multispectral satellite information for the meteorological scenario interpretation
The optimum coloring of the RGB image composites rely on the proper selection of the channels and the enhancement of the individual colors
The channel selection must be driven by the particular phenomenon (low or high clouds, dust, smoke etc.) to be emphasized in the satellite image
The proper color enhancement requires the conversion from radiances to brightness temperatures (IR channels) or reflectances (VIS/NIR channels), selection of the display mode (inverted or not inverted), stretching of the dynamic range of the satellite data, and gamma correction
The adopted scheme, so-called day microphysical is particularly recommended for cloud analysis (optical thickness, effective size and top height) and for emphasizing the presence of severe convection
From the SEVIRI images interpretation guidelines*, we selected the so-called day-microphysical scheme, developed to interpret and analyze the following components/scenarios:Cloud – Convection – Fog – Snow – FiresAdapted to VIRS, it combines channels this way:R = Channel 01 (VIS0.6)G = Channel 03 (NIR3.7 only reflected solar component)B = Channel 04 (IR10.8)
deep precip cloud
thin Cirrussmall ice particle
severe conv.
thin Cirruslarge ice particle
veget. land
ocean
June 1th
* By, and with the contribution of: H. P. Roesli, J. Kerkmann, D. Rosenfeld, M. KönigAvailable at http://oiswww.eumetsat.org/WEBOPS/msg_interpretation/index.html
The cloud mask can be described as a cascade of tests involving VIRS channels. It is applied separately over land and sea pixels.Cloudy pixel in the cloud swath are identified by using very conservative tests, already set to identify precipitating clouds by the CERES science team.This kind of selection identify optically thick, ice clouds such as towering cumulonimbus.
Over the landif [TB(CH4) < 257 K] and [R(CH1) > 0.38] it’s a cloud and also [TB(CH3) - TB(CH4) > 20 K] and [TB(CH4) < 237 K] and [R(CH1) > 0.45] it’s raining (maybe!)
Over the sea First avoid sun glint: θscatt > 36° then if [TB(CH4) < 257 K] and [R(CH1) > 0.065] it’s a cloud and also [TB(CH3) - TB(CH4) > 20 K] and [TB(CH4) < 237 K] and [R(CH1) > 0.45] it’s raining (maybe!)
Microphysical properties retrieval from VIRS: CAPCOM (Comprehensive Analysis Program for Cloud Optical Measurement)
It allows for the retrieval of cloud optical thickness (), effective radius (Re)
and top temperature (Ttop) from satellite measurements in the VIS, NIR and IR
channels. The cloud phase is supposed known.
Undesirable radiation components (ground reflected and thermal emitted radiation, NIR thermal contribution) are subtracted from the satellite radiance to derive the cloud signal.
The retrieval is performed by means of comparison between the modelled cloud radiances and the corresponding satellite radiance measurements.
The LUTs are built for a grid of selected values of , Re, Ttop, water vapour
amount above and below the cloud layer, solar zenith, satellite zenith and relative azimuth angles. Separate LUTs for ice and water clouds are computed.
Nakajima, T. Y., and T. Nakajima, 1995, J. Atmos. Sci.,52, 4043 – 4059Kawamoto, K., T. Nakajima, and T. Y. Nakajima, 2001, J. Climate, 14, 2054 – 2068
CAPCOM is available at http://www.ccsr.u-tokyo.ac.jp/~clastr/
Based on TRMM data, WA daytime convection is analysed in several channels (from VIS to MW) to derive maximum information on rain processes
Cloud mask tests perform very well in this geographic area having such large thermal contrast between cloud top and surface
Cloud structures delineated in the cloud mask nicely agree with RGB images
Rain data derived from TMI and PR show a limited agreement depending also on the climatic sub-area, due to differences in algorithms, resolution, observing geometry, frequency, etc.
The microphysical characterization of the cloud tops (especially in terms of Re) does not add significant information about the underlying precipitation layer: the thick frozen layer detected from the scattering in TMI high frequency channels is detected in multispectral VIRS data as well, but even the greater Re values do not correlate with surface rain
Moreover, due to the input set-up of the microphysical retrieval code, in large areas corresponding to the colder and higher (overshooting) cloud structure, the algorithm is not able to retrieve meaningful information, probably due to the very low temperatures, not represented in the input data and the vanishing sensitivity of the 3.7 µm reflectances to Re
The cloud mask from VIRS channels, can be considered a useful input to the screening procedure of PMW based rain algorithm