Peter romanov
Transcript of Peter romanov
Satellite-Based Monitoring of Snow Cover
at NOAA: Application to Himalayan-
Tibetan Region
Peter Romanov
NOAA-CREST, City University of New York
Center for Satellite Applications and Research, NOAA/NESDIS
NOAA: National Oceanic and Atmospheric Administration, USA
NESDIS: National Environmental Satellite Data and Information Service of NOAA
Outline
• NOAA/NESDIS snow mapping/monitoring activities
• Snow products: Application in mountainous regions
Snow cover: Needs and Requirements
Requirements to snow products
• Daily, Spatially continuous
• Continental or global scale coverage
• 1-4 km resolution
Surface observations are not enough, satellite data should be used
NOAA needs information on snow for
• Weather prediction
• Climate studies
• Hydrological forecasts, flood warnings
Snow cover: Techniques
Satellite-based snow mapping techniques used
• Interactive
• Automated
- Visible & infrared
- Passive microwave
- Combined visible-infrared-microwave
Focus on operational polar and geostationary satellites & sensors
NOAA, GOES, NPP, METOP, Meteosat, GCOM, MTSAT, DMSP
On the Web: http://www.natice.noaa.gov/ims/
Interactive snow mapping
- Based on visual analysis of
satellite imagery in optical bands
- Yes/no classification (snow/land,
water/ice)
- Maps available since early 1970s
Temporal sampling and spatial
resolution:
1972-1997 : weekly at 180km
1998-2003 : daily at 24 km
2004-2014 : daily at 4 km
2015 - : twice daily at 1 km NOAA Interactive Multisensor Snow and Ice
Mapping System (IMS)
Interactive snow/ice maps: example
Snow: white
Ice: yellow
Background:
elevation
- Interactive snow/ice maps are spatially continuous
- Cloudy areas: analysts make reasonable guess or use in situ data
- Effective spatial resolution may be coarser than nominal
Pamir-Tien Shan region
Snow cover from visible/infrared data
- Automated algorithms applied to AVHRR, VIIRS, SEVIRI sensor data
- Land/snow, water/ice, cloud categories
- High spatial resolution: 1 km and below
- High retrieval accuracy
- Requires daylight
- Gaps due to clouds (~40% of the area)
Metop AVHRR Snow: white
Ice: yellow
Clouds: gray
Background:
elevation
Mar, 15 Mar, 19 Mar, 22
Clouds
Snow temperature: snow melt identification
GOES-East Imager data
Snow-free land Red: Melting Snow
Based on visible and
infrared observations
Timely snow melt
identification may be
hampered by persistent
clouds
Snow retrievals from microwave
5 km
- Spatial resolution: 25-50 km
- Weather independent, mostly continuous coverage
- Problems: mountains, melting and shallow snow
- Sensitive to snow depth and snow water equivalent
- But retrieval errors are 50-100%
Snow water equivalent from
AMSR-E Aqua
On the Web: http://www.orbit.nesdis.noaa.gov/smcd/emb/snow/HTML/multisensor_global_snow_ice.html
Combined visible-infrared-microwave
- Automated algorithm, uses strengths of both techniques
- Spatially-continuous maps of snow and ice cover
- Nominal spatial resolution : 4 km
- Effective resolution varies depending on the source of data used.
Most satellite products agree to surface observations of snow
at the rate of over 90%.
Vis/IR: 93-98% agreement (but only for cloud-clear scenes)
Interactive: 90-95%
Combined: 90-95%
Microwave: 80-90%
How well satellite snow extent maps agree to surface observations ?
Snow cover duration
Based on daily snow extent maps
Used in climate change studies
2008-2009 Derived from IMS data
Snow extent change: 1972-2013
Estimated yearly mean snow extent decrease rate in NH is ~1.7% per decade Largest decrease occurred in 1980s
1970 1980 1990 2000 2010
Year
0
5
10
15
20
25
30
Are
a,
mln
sq
km
Northern Hemisphere
Eurasia
North America
Feb 22, 2012
VIIRS snow cover product at 0.5 km spatial Specific problems of snow mapping in mountains
- Topographical shadowing
- Geo-registration may be less accurate
- Lack of ground-truth for validation
- Microwave snow products are not reliable
Different snow products over Tibet
AVHRR False Color AVHRR (Automated)
IMS (Interactive)
Closer look reveals some
differences between products…
Interactive analysts tend to
overestimate snow extent
Clouds: gray Snow: white
Dec 25, 2014
Feb 22, 2012
VIIRS snow cover product at 0.5 km spatial Scaling issue
- Algorithms and analysts typically map pixel with any
amount of snow in it as “snow covered”
- As a result, snow extent in finer spatial resolution products
is smaller than in the coarser resolution products
- This effect is most pronounced in the mountains
Snow occurrence from interactive maps
1972-1997 week 33
180 km
1998-2003 week 33
24 km
2004-2014 week 33
4 km
Snow occurrence on week 33
(Aug 13-19)
Snow occurrence calculated at 180
km spatial resolution
Trend in snow occurrence in the
mountains areas is mostly spurious
Its is caused by different spatial
resolution of base snow maps
All products are available through NOAA web sites
Most reliable/accurate retrievals are of snow extent. Limited
ability to derive snow depth or SWE.
All products have some “inertia”: None of the sensors can “see”
through precipitation clouds
Differences in snow products are due to different techniques,
data sources, spatial resolution, time of observation.
All NOAA snow mapping techniques are continental/global. For
regional studies use global products with care
- Regionally-tuned algorithms may produce better results
Final notes
Links
NESDIS Automated snow remote sensing page:
http://www.star.nesdis.noaa.gov/smcd/emb/snow/HTML/snow.htm
NOAA Interactive snow charts:
http://www.natice.noaa.gov/ims/
NESDIS Microwave remote sensing page:
http://www.star.nesdis.noaa.gov/corp/scsb/mspps/
NOAA Satellite Data Archive:
http://www.class.noaa.gov
THANK YOU
Snow and Ice cover 2013-2014
Based on daily maps generated with NOAA Global Multisensor Automated Snow and Ice Mapping System (GMASI)