Ice Cover and Sea and Lake Ice Concentration with GOES-R ABI

Size: px

Start display at page:

Download "Ice Cover and Sea and Lake Ice Concentration with GOES-R ABI"

Caroline Stanley
5 years ago
Views:

1 GOES-R AWG Cryosphere Team Ice Cover and Sea and Lake Ice Concentration with GOES-R ABI Presented by Yinghui Liu 1 Team Members: Yinghui Liu 1, Jeffrey R Key 2, and Xuanji Wang 1 1 UW-Madison CIMSS 2 NOAA/NESDIS/STAR 1

2 The Cryosphere Team Members AWG Cryospere Team Chair: Jeff Key Cryosphere Application Team Jeff Key (Lead; STAR/ASPB)» Peter Romanov (UMD/STAR)» Don Cline (NWS/NOHRSC)» Marouane Temimi (CREST) Snow Depth Peter Romanov (Lead; UMD) Cezar Kongoli (QSS) Fractional Snow Cover Don Cline (Lead; NWS/NOHRSC) Tom Painter (U. Utah) Andy Rost (NWS/NOHRSC) Ice Cover and Concentration Yinghui Liu (Lead; CIMSS) Xuanji Wang (CIMSS) Jeff Key (STAR) Marouane Temimi (CREST) Hosni Ghedira (CREST) Rouzbeh Nazari (CREST) Kim Smith (CREST) Ice Motion Yinghui Liu (Lead; CIMSS) Jeff Key (STAR) Xuanji Wang (CIMSS) Ice Age/Thickness Xuanji Wang (Lead; CIMSS) Jeff Key (STAR) 2 Yinghui Liu (CIMSS)

3 Outline Executive Summary Algorithm Description Examples of Product Output Validation Approach Validation Results Next Steps to Reach 100% Conclusion and Perspectives 3

4 Executive Summary The GOES-R R Ice Cover and Sea and Lake Ice Concentration are Option 2 products. Software Version 3 was delivered in September The algorithm has been developed and tested using MODIS data, and validated with the Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) products. Validation studies indicate spec compliance 4

5 Executive Summary Product Measurement Precision Vendor Allocated Ground Latency Product Refresh Rate/Coverage Measurement Time Accuracy Measurement Range Mapping Accuracy Horizontal Resolution Vertical Resolution Geographic Coverage (G, H, C, M) User & Priority Name N/A 77,756 sec 180 min 85% correct detection FD N/A 2 km 1 km Binary yes/no detection Ice Cover GOES- R 5 FD Full Disk

6 Executive Summary Product Measurement Precision Vendor Allocated Ground Latency Product Refresh Rate/Coverage Time Measurement Accuracy Measurement Range Mapping Accuracy Horizontal Resolution Vertical Resolution Geographic Coverage (G, H, C, M) User & Priority Name Sea & Lake Ice: Concentration GOES-R C: Regional & Great Lakes and US costal waters containing sea ice hazards to navigation Ice Surface 3 km < = 1.5 km Ice concentration 1/10 to 10/10 Ice concentratio n 10% 180 min 3236 sec 30% Sea & Lake Ice: Concentration GOES-R FD: Sea ice covered waters in N & S Hemisphere s Ice Surface 10 km < = 5 km Ice concentration 1/10 to 10/10 Ice concentratio n 10% 6 hr 9716 sec 30% C CONUS FD Full Disk 6

7 Algorithm Description Ice Cover: A binary yes/no detection of ice cover over ocean, inland lakes, and rivers. Ice Concentration: The fraction (in tenths) of the sea or lake surface covered by ice. MODIS true color image over Great Lakes on February

8 Future GOES Imager (ABI) Band Ice Cover and Sea and Lake Ice Concentration with Nominal Wavelength Range ( m) Algorithm Description GOES-R ABI channel data used in the algorithm Nominal Central Wavelength ( m) Nominal Central Wavenumber (cm-1) Nominal sub-satellite satellite IGFOV (km) Sample Use ice concentration and extent ice concentration and extent ice concentration and extent ice concentration and extent ice concentration and extent

9 Algorithm Description Ice cover Detection: Ice or snow covered ice show high NSDI value and high reflectance at visible and near-infrared bands. NDSI=(R vis -R swir ) /(R vis +R swir ) NDSI value: Water: ~ 0.5 New ice: ~ 1.0 Snow ice: ~ 1.0 Cloud: ~ 0.0 8% 9 Reflectances of ice, clouds, and water (Riggs et al. 1999).

10 Algorithm Description Pixels with both NDSI and reflectance larger than set thresholds are ice. Top of the atmosphere NDSI in this algorithm is defined as NDSI=(R 1 -R 2 )/(R 1 +R 2 ) R 1 = reflectance at GOES-R ABI ch 3, μm R 2 = reflectance at GOES-R ABI ch 5, 1.61 μm Threshold : 0.6 Reflectance at near infrared channel, μm Threshold

11 Algorithm Description Ice Concentration: Ice reflectance (temperature) change linearly with ice concentration Pure ice Reflectance Pure water Temperature 11

12 Algorithm Description Tie-point algorithm: In a search window, the reflectance (temperature) e) of pure ice is determined as the reflectance (temperature) with the maximum relative frequency. Figure: Reflectance probability density distribution at 0.64 μm for ice cover over Lake Erie on Feb 24, 2008 (left), and over Barents and Kara Seas on Mar 31, 2008 (right). The ice tie points 12(0.6, 0.65) are indicated.

13 Algorithm Description Ice/snow surface temperature is retrieved by the following equation (Key et al. 1997). Ts = a + b T11 + ct12 + d [(T11-T12)(secθ-1)] Ts = the estimated surface temperature (K) T11 = the brightness temperatures (K) at 11 um T12 = the brightness temperatures (K) at 12 um θ = sensor scan angle a, b, c, d = coefficients, derived for the following temperature ranges: T11 < 240K, 240K < T11 < 260K, T11 > 260K. The 13

14 Algorithm Description Fractional ice concentration for each pixel (F p ) in a search window is then calculated as F p = (B p -B water ) / (B ice B water ) B water = the reflectance/temperature (K) of a pure water pixel B ice B p = the reflectance/temperature (K) of a pure ice pixel = the observed reflectance/temperature (K) of the pixel. In this algorithm, reflectance at GOES-R ABI channel 2 (0.64 μm) is used during the day, and surface temperature is used at night. The spatial resolution is 0.5 km at 0.64 μm channel, and 2.0 km for surface temperature at sub-satellite FOV. 14

15 Algorithm Description High Level Flowchart of the ice concentration and extent algorithm Ice concentration and extent algorithm begin ABI channel radiance, satellite viewing angles, cloud mask, land/water mask step1 Group-criteria detection Ice step2 Tie point algorithm Ice concentration step3 Ice extent Ice concentration and extent algorithm end 15

Examples of Product Output Comparisons with AMSR-E product show good agreement Sea ice concentration (SIC) (%) retrieved from (a) MODIS Sea Ice Temperature (SIT), (b)

16 Examples of Product Output Comparisons with AMSR-E product show good agreement Sea ice concentration (SIC) (%) retrieved from (a) MODIS Sea Ice Temperature (SIT), (b) MODIS visible band reflectance, and (c) from Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) Level-3 gridded daily mean from NSIDC on March 31,

17 Examples of Product Output Comparisons with true color image show good agreement Lake ice concentration (%)retrieved from (a) SEVIRI Surface Ice Temperature (SIT), (b) SEVIRI visible band reflectance (0.64 μm, an (c) satellite true color image over Caspian Sea on January 27,

data (left), MODIS true color image (middle), and from

18 Examples of Product Output Good performance over the Great Lakes Lake ice concentration (%) with MODIS Aqua data (left), MODIS true color image (middle), and from AMSR-E (right) over Great Lakes on February

19 Validation Performance of Ice Cover product Case number Total pairs: Sea/Lake ice cover determined from AMSR-E Water determined from AMSR-E Sea/Lake ice cover Water Correct detection ratio = ( )/ = 87.6% The product measurement accuracy is higher than required 85% correct detection. 19

20 Validation Ice Cover and Sea and Lake Ice Concentration retrieved from MODIS data as proxy were validated with ice cover and concentration from the AMSR-E E product as truth. AMSR-E Level 3 product provides ice concentration over the Great Lakes and the Arctic Ocean. Ice cover is assigned when ice concentration is larger than 15%. Ice cover and sea and lake ice concentration retrievals were from MODIS data using the GOES-R ABI algorithm. Two datasets are matched daily by averaging the retrieved product in the footprint of AMSR-E product. Validation statistics are computed based on 1,576,298 matched pairs covering different seasons. 20

21 Validation Performance of Ice Concentration product Ice concentration difference of AMSRE product and MODIS Mean bias (%) Standard deviation (%) Over Arctic Ocean Over Great Lakes Required measurement accuracy 10 Required measurement precision 30 The product measurement accuracy and precision meet the requirement. 21

22 Validation Performance of ice concentration product Frequency distribution of ice concentration differences between the AMSR-E product and retrievals using the ABI algorithm based on selected 41 clear day MODIS data in four seasons in 2007 over the Arctic Ocean. 22

23 Next Steps to Reach 100% This algorithm will be constantly updated for improvement. Algorithm Improvements: Optimize the thresholds in the group thresholds test. In the tie-point algorithm, size of the search window will be test to get the best results. Current size of the search window is 50 km by 50 km The tie point reflectance (temperature) is being improved, including how to deal with the effect of different ice types in one search window, to get the best result. Validation Improvement: More quantitative validation of the products with AMSRE and other available products, including ice chart, and data from LANDSAT are underway. Other Improvements: The instrument noise on the final retrieval results is being investigated. 23

Sea ice monitoring over the Caspian Sea using geostationary satellite data at NOAA CREST The average percentage of cloud

Daily maps of ice distribution and concentration with minimal cloud coverage were produced.

latitude-longitude grid on 23 January 2007 at 10:15 AM UTC.

true-colour image for the same day (left) The obtained correlation coefficients with IMS charts for 2007 and a 2008 were 0.

The technique has been proposed as one of candidate ice mapping techniques for the future GOES-R R ABI instrument. TEMIMI, M.

24 Sea ice monitoring over the Caspian Sea using geostationary satellite data at NOAA CREST The average percentage of cloud reduction because of the daily compositing ranged from 22% to 25%. Daily maps of ice distribution and concentration with minimal cloud coverage were produced. MSG SEVIRI full disk false color composited image and the portion of the image over Caspian Sea reprojected to latitude-longitude grid on 23 January 2007 at 10:15 AM UTC. SEVIRI-based sea ice map over the northern part of the Caspian Sea on 28 February 2007 at 11h15 AM UTC (right) and the MODIS true-colour image for the same day (left) The obtained correlation coefficients with IMS charts for 2007 and a 2008 were 0.92 and 0.83 respectively. The technique has been proposed as one of candidate ice mapping techniques for the future GOES-R R ABI instrument. TEMIMI, M., ROMANOV, P., GHEDIRA, H., KHANBILVARDI, R. & SMITH, K. (2010) Sea ice monitoring over the Caspian Sea using geostationary satellite data. International Journal of Remote Sensing, Accepted. Instantaneous ice maps (left column) and original MSG SEVIRI images on 23 January False color images in the right column are constructed with Ch.3 reflectance (red), HRV reflectance (green) and inverted infrared brightness temperature (blue)

25 Conclusion and Perspectives ABI allows us to monitor ice conditions at high temporal and spatial resolution. The ice cover product is generated using group threshold tests. Comparison of ice cover products using MODIS as proxy with AMSR-E shows that products exceed the required detection accuracy of 85%. The ice concentration product is produced using a tie-point algorithm. Validation of this product with the AMSR-E product shows that products meet the required measurement accuracy and precision. Further tuning of the algorithm, including test thresholds in the group threshold test, and parameters in tie-point algorithm are needed. More quantitative validation of the products with AMSRE and other available products, including ice chart, and data from LANDSAT are underway. 25

26 Ice in North America 26

DEVELOPMENT AND VALIDATION OF A BRDF MODEL FOR ICE MAPPING FOR THE FUTURE GOES-R ADVANCED BASELINE IMAGER (ABI) USING ARTIFICIAL NEURAL NETWORK

P1.10 DEVELOPMENT AND VALIDATION OF A BRDF MODEL FOR ICE MAPPING FOR THE FUTURE GOES-R ADVANCED BASELINE IMAGER (ABI) USING ARTIFICIAL NEURAL NETWORK Hosni Ghedira, Marouane Temimi*, Rouzbeh Nazari and