200 likes | 292 Views
OUTLINE. Background: Automated monitoring of ice cover at NOAA NESDIS, sensors, techniques, products. Other factors affecting/contributing to ice mapping with optical imagery NPP VIIRS for ice monitoring Path forward. Automated satellite-based mapping of ice cover at NOAA NESDIS
E N D
OUTLINE Background: Automated monitoring of ice cover at NOAA NESDIS, sensors, techniques, products. Other factors affecting/contributing to ice mapping with optical imagery NPP VIIRS for ice monitoring Path forward
Automated satellite-based mapping of ice cover at NOAA NESDIS • Since early 1990s: Ice cover maps from satellite microwave data (Grody, Ferraro). • Daily, global, 30 km resolution • Small water bodies were not resolved • Since early 2000s: Ice and snow maps from combined observations by vis/IR and microwave sensors • Daily, global, 4 km nominal resolution • Improved ice mapping over lakes and along the coastal line
Automated multisensor snow and ice mapping system Operational since 2006. Daily, 4 km nominal resolution. Provides spatially continuous maps of snow and ice cover. Current configuration includes METOP AVHRR, GOES-E and -W Imager, MSG SEVIRI, SSMIS DMSP F-16,-17,-18. Focus is on daily and seasonal change in the global and continental scale ice and snow extent. Primary application is in NWP models and climate studies.
Snow and Ice retrievals from optical sensors data METOP AVHRR • Polar and geo satellites • 2-4 km spatial resolution • Gaps due to clouds and darkness GOES-E and -W Imager
Ice cover: Caspian Sea METOP AVHRR Feb 20, 2013 RosHydroMet, Interactive ice chart Feb 20, 2013 Cloud gap-filled snow/ice Caspian Sea, 4 km
Ice break up on small lakes 2007 spring season • Derived from GOES daily snow/ice maps • Large lakes and seas (Great Lakes, Hudson Bay, etc.) excluded • Ice fraction is aggregated within 10 lat x 20 lon grid cells • Ice break-up is assumed when the ice fraction becomes less than 0.8
Northern Hemisphere Ice Extent North America Snow Extent Eurasia Snow Extent
Clouds are the principle issue in the automated ice mapping • Our cloud identification algorithms include • Spectral-based analysis of scene response • Spatial consistency check • Climatology consistency check METOP AVHRR • Availability of multiple observations per day alleviates the problem: • Cloud-clear image compositing reduces cloud obscuration • 10% less clouds if Terra and Aqua MODIS imagery is combined • Analysis of temporal variations of scene response improves cloud detection
Polar-orbiting satellites: Repeated observations in polar areas Repeated observations help to (1) Reduce the area obscured by clouds through clear-sky compositing (2) Better identify clouds by assessing temporal variation (or stability) of the scene reflectance and temperature (3) Better assess the ice motion Number of daily views from NPP VIIRS, March 30 2013 1 2 3 4 5 6 7
Reflectance Anisotropy Satellite-observed reflectance of the land surface depends on particular viewing and illumination geometry of observations. It has to be accounted for when trying to identify and map ice cover with optical sensors data. Diurnal change of the observed reflectance of cloud-clear ice-covered surface. MSG SEVIRI data. SEVIRI: imaging instrument onboard European geostationary platform
MSG SEVIRI Ice Reflectance Anisotropy: Caspian Sea Ice Water Feb 20, 2012 Ice Open Water Reflectance anisotropy changes with surface type and with the wavelength
Correcting for Reflectance Anisotropy (1) Physical approach: Look-up tables of simulated ice and water reflectance based on physical RT models (DISORT, 6S) (2) Empirical approach: Corrective factors are determined from accumulated statistics of satellite observations Empirical kernel-driven models (tested with GOES, AVHRR, MODIS) - Actively used in land surface studies - Good for cases when observation angular range is limited R(ƟS, ƟV, φ) = C0 + C1 f1 (ƟS, ƟV, φ)+C2 f2(ƟS, ƟV, φ) f1 =( tanƟS + tan ƟV ) : to reproduce reflectance change with solar and view angles ƟS and ƟV f2=( cosφ + 1 )2 (tanƟS tanƟV)1/2 : to reproduce reflectance azimutal change
Reflectance Model Fit Feb 20, 2012 Solid: Observed Reflectance Dashed: Model Fit MSG SEVIRI data over Caspian Sea R1 (vis), R2 (near-IR), R3 (SWIR)
Visible Infrared Imaging Radiometer Suite (VIIRS) Launched in October 2012 onboard NPP 22 bands (14 solar, 7 thermal 1 Day-night band) I-bands (375m FOV) and M-bands (750m) 12-bit quantization Swath width: 3040 km (compare to 2330 km of MODIS and 2900 km of AVHRR) 24 EDRs to be routinely produced Ice products: - Ice concentration - Ice Surface Temperature - Ice Age - @750 m spatial resolution - Will be decclared “beta” in late April 2013
VIIRS vs MODIS • MODIS: optical bands @ 250-500m, thermal bands @1000m • VIIRS: all imaging bands @375m MODIS TERRA 1030UTC NPP VIIRS 1330 UTC Bothnian Bay March 27, 2013 10 km Potentials and ways to conduct routine ice mapping at 250 and 375m with MODIS and VIIRS data are being investigated.
Project plan Develop the new advanced ice identification algorithm Develop an automated system to routinely map ice cover using NPP-Aqua-Terra data. Focus on small-scale details in the ice cover distribution Validate ice retrievals with interactive ice charts, active microwave imagery • Primary output: Daily maps of ice cover at 0.5 km resolution. This is 1.5-2 times higher spatial resolution than the resolution of current MODIS and VIIRS ice products.
Study Area 550 N - 850 N 1800 W - 1400 W ice Seasonal min ice extent (mid-September) ice • Includes Chukchi, Beaufort and Bering Seas • Fully covers seasonal variations of the ice edge position in the region Seasonal max ice extent (mid-March)
Caveats • Limited potentials for mapping ice in the study area during October to December time period • - “Polar night” conditions advance northward earlier than the ice edge. • No retrievals in cloudy conditions • - Delayed identification of changes in the ice extent
Validation • Interactive Charts: • Analysts may interpret the same imagery differently • Small details in the ice distribution may be omitted • Will also rely on the original satellite imagery analysis Source: RosHydroMet Data used: MODIS, OSCAT, VIIRS Source: National Ice Centert Data used: MODIS, Radarsat