The GOESR AWG Cloud Mask Product Andrew Heidinger

  • Slides: 16
Download presentation
The GOES-R AWG Cloud Mask Product Andrew Heidinger (NOAA/NESDIS) William Straka (UW-CIMSS)

The GOES-R AWG Cloud Mask Product Andrew Heidinger (NOAA/NESDIS) William Straka (UW-CIMSS)

What You Should Know by the End of Training • Product interpretation through examples

What You Should Know by the End of Training • Product interpretation through examples of interest. • Strengths and limitations of the cloud mask product • Future improvements for the cloud mask algorithm Images courtesy of NASA ISS and STS-107

Introduction • End user products viewable in AWIPS • Cloud Mask (4 level) •

Introduction • End user products viewable in AWIPS • Cloud Mask (4 level) • AVHRR SST is generated using the AWG Cloud Mask • All GOES Cloud Products generated from the Mask • This product is made from the visible reflectance (during the day) infrared and water vapor observations. In addition, information from an NWP model such as the GFS and radiative transfer model are used. • The GOES-R products will be made with information from 8 of the 16 bands available, including all 5 GOES-11 bands. GOES-13/14/15 utilizes information from all but the 13. 3 mm channel.

What Is Currently Available • Infrared and visible satellite imagery • ASOS cloud cover

What Is Currently Available • Infrared and visible satellite imagery • ASOS cloud cover observations • Model Forecasts of cloud cover Key: While there is information on cloud cover, it is extremely limited.

What is wrong with the current approachs? ASOS stations • Limited in scope and

What is wrong with the current approachs? ASOS stations • Limited in scope and widely spaced. • Amount of sky that is taken into account gets smaller as the clouds get closer to the instrument • Clouds above 12 km are missed. Satellite imagery (reflectances, BTs and BTDs): • Optically thin clouds are difficult to see in single-band images. • At night, low level clouds can be missed when looking at a single channel of data. • Channel differences, while they offer more information than just a single channel of imagery, can result in false cloud detection over various surfaces. • Hard to quantify for sky-cover estimates. Numerical Models • Cloud physics are often inadequate for realistic cloud forecasts.

AWG Cloud Mask (ACM) • AWG Cloud Mask (ACM) uses 9 out of the

AWG Cloud Mask (ACM) • AWG Cloud Mask (ACM) uses 9 out of the 16 ABI spectral bands, including all but the 13. 3 micron channel on current GOES • Makes extensive of NWP and Radiative Transfer Model (RTM) data, to generate the expected clear-sky state for the spectral and temporal tests. • Thresholds for cloud mask tests based off of training with CALIPSO (a lidar) • Adaptable to any current imager (AVHRR, GOES-11, 12, 13, MODIS and VIIRS), simply be turning of un-used tests Validation against CALIPSO performed on AQUA/MODIS – an Advanced Imager from NASA with all the channel combinations we use. ) www. nasa. gov

ABI Cloud Mask Test Description The ABI Cloud Mask is a series of tests

ABI Cloud Mask Test Description The ABI Cloud Mask is a series of tests • 14 tests that look for specific signatures of the presence of cloud (cloud, bright, non-uniform, …) • 2 tests look for non-uniformity to filter clear pixels the in the case of missed cloud. • 1 restoral test to prevent reclassification of clear pixels in regions where absolutely no cloud was detected. • Some tests are new, some are borrowed and most are modified versions of well-known tests

Known Strengths • Products are generated within minutes of receiving satellite data. ACM runs

Known Strengths • Products are generated within minutes of receiving satellite data. ACM runs on the current GOES and POES Imagers. • ACM is flexible. Tests can be turned off to optimize the results for specific applications. We are open to optimizing the ACM based on NWS feedback. • Our analysis shows we meet the GOES-R specification on current GOES though GOES-R ABI should be much better. • Probably clear and probably cloud are correlated to sky cover. These categories may aid in sky-cover forecasts and verification.

GOES-R Cloud Mask Product: What is Provided

GOES-R Cloud Mask Product: What is Provided

ACM IR Window Visible Reflectance ACM in AWIPS (upper left image). Colors are white

ACM IR Window Visible Reflectance ACM in AWIPS (upper left image). Colors are white (cloudy), red (probably cloudy), cyan (probably clear), black (clear). H 2 O IR

ACM in AWIPS (upper left image). Colors are white (cloudy), red (probably cloudy), cyan

ACM in AWIPS (upper left image). Colors are white (cloudy), red (probably cloudy), cyan (probably clear), black (clear). ACM Visible Reflectance IR WINDOW

ACM in AWIPS (upper left image). Colors are white (cloudy), red (probably cloudy), cyan

ACM in AWIPS (upper left image). Colors are white (cloudy), red (probably cloudy), cyan (probably clear), black (clear). ACM Cloud Height Cloud Type Cloud Temperature

Cloud Mask Applications • The Cloud mask can be used to determine the extent

Cloud Mask Applications • The Cloud mask can be used to determine the extent of cloudiness Ø What is currently happening, what is coming towards a location in terms of if a pixel is cloudy or not. • Can be used with cloud amount forecast product Ø Potential to use as an input to sky-cover forecast model Ø Can be used as a verification of cloud sky cover forecast.

Near-Term Improvements • End user feedback is essential and will be used to guide

Near-Term Improvements • End user feedback is essential and will be used to guide future improvements. • Known issues include Ø Missing low cloud at night (fog detection can be used to help remedy this). Ø Terminator discontinuities. Ø Sensitivity to NWP. Large errors in NWP surface temperature can impact ACM.

What You Should Know by the End of Training • Which GOES-R cloud Mask

What You Should Know by the End of Training • Which GOES-R cloud Mask products are available within AWIPS • Product interpretation through examples of interest in CONUS and AK • Strengths and limitations of the cloud mask product • Future improvements for the cloud Mask algorithm

Conclusions • Feedback is greatly appreciated. • Thank you for viewing this training. •

Conclusions • Feedback is greatly appreciated. • Thank you for viewing this training. • Contacts: Ø Andrew Heidinger (Andrew. [email protected] gov) Ø William Straka ([email protected] wisc. edu)