Satellite Asian Dust Monitoring of...
Transcript of Satellite Asian Dust Monitoring of...
KMA
KMA
Satellite Asian Dust Monitoring of KMA
Jae-Dong Jang NMSC
SCOPE-Nowcasting EP-1
KMA
DUST 10.8㎛ 12㎛ CLOUD 10.8㎛ 12㎛
BT10.8- BT12 : negative absorption 10.8㎛>12㎛
BT10.8- BT12 : positive absorption 10.8㎛<12㎛
Absorption coefficient of radiance depending on atmospheric particle
BTD(Brightness Temperature Difference) = 10.8㎛ -12㎛
Dust detection of COMS
Factors influencing BTD Aerosol types Satellite zenith angle Surface emissivity Surface temperature Surface reflectance Dust altitude
use BTV
Specific period COMS L1B data
Cloud Screening (BTD > 0.5K)
Extract Maximum BT11㎛ (extract maximum 11㎛ BTV
at same time, same pixel )
11㎛ – 12㎛ =BTV When 11㎛ has maximum value
BTV
Usage of BTV(Background Threshold Value) - BTD expected to be clear sky - To Minimize effects generated by surface characteristics
when using BTD
BTD(11㎛-12㎛) BTV BTD-BTV
2015.02.22. 00:00UTC
KMA
Area: East Asia Spatial res.: 4km Temporal res.: 15 min. NMSC is currently operated using Static BTV made of a month data
(specific period:2011.04).
<BTV>
COMS AI Algorithm
<COMS AI Algorithm>
KMA
False detection in land Weakened signal in ocean
Increase of radiance emitted from surface by heating of surface
Decrease of 12㎛ radiance by water vapor effect in ocean
Problems on Dust detection of COMS
2012.03.31 06:00UTC
2012.03.31 06:16UTC
GOCI DUST WEATHER CHART
2015.02.22 03:00UTC
KMA Use of Dynamic BTV
Static BTV applied uniformly cause error because BTV pixels change depending on time and space.
Change of Static BTV as time passes
DAY
NIGHT
→ Use of Dynamic BTV considering spatial-temporal change
KMA New COMS AI algorithm
Usage of dynamic BTV(10 days from the day before observation) Application of BTV condition depending on BTV threshold [consideration of influences on Surface characteristic(surface emissivity, temperature)]
⁃ Land : BTV<0: BTD-BTV, BTV>0: BTD ⁃ Ocean : BTD-BTV
Adjustment of threshold in ocean: AI≤-0.5 →AI≤-0.2
KMA
Previous algorithm new algorithm MODIS RGB IMAGE Dust weather chart
Application of new COMS AI algorithm
2015.02.22. 03:00 UTC False detection over land
Previous algorithm new algorithm HIMAWARI dust RGB LIDAR
This image cannot currently be displayed.
Discontinuity between land and ocean 2016.04.13. 16:00 UTC
KMA COMS AI Algorithm applied to Himawari-8(1/3)
Channel
AHI(Himawari-8) MI(COMS)
Center wave length(㎛)
Spatial resolution(km)
Center wave length(㎛)
Spatial resolution(km)
1(VIS) 0.46 1
2(VIS) 0.51 1
3(VIS) 0.64 0.5 0.67 1
4(NIR) 0.86 1
5(NIR) 1.6 2
6(NIR) 2.3 2
7(IR) 3.9 2 3.7 4
8(WV) 6.2 2
9(WV) 7.0 2 6.7 4
10(WV) 7.3 2
11(IR) 8.6 2
12(IR) 9.6 2
13(IR) 10.4 2 10.8 4
14(IR) 11.2 2
15(IR) 12.3 2 12.0 4
16(IR) 13.3 2
<COMS AI Algorithm applied to Himawari-8 data>
CH
AHI MI
C.W. (㎛)
Spatial res. (km)
Application C.W. (㎛)
Spatial res. (km)
13 10.4 2 Surface, cloud, dust 10.8 4
14 11.2 2 SST, cloud, precipitation
15 12.3 2 SST, Dust, Ash, TPW 12.0 4
<Spectral Response Function>
Response function vary with wave length of satellite channels.
#13-#15(10.4㎛-12.3㎛) exhibits the greatest similarity to that of MTSAT-2(IR1-IR2(10.8㎛-12.0㎛))
-Hidehiko et al., 2015
Ch13(10.4㎛) and Ch15(12.3㎛) were used considering similarity to center wave length between COMS and MTSAT2.
KMA
Application of equation to through channel analysis between COMS and Himawari-8
EQ_CH13 = 0.9198*CH13 +22.7638 EQ_CH15 = 0.9430*CH15 +16.8116
AI ≤-0.5K(COMS AI algorithm) → AI ≤-0.85K, AI ≤-1.0K
Adjust AI threshold 30days dynamic BTV
regression coefficient to correct difference of Satellite channel
COMS AI Algorithm applied to Himawari-8(2/3)
2016.05.05. 0300UTC
COMS AI No Equation Equation
CLOUD THRE. BTD1315 > 0.5
BTV
AI Dust Signal
White: cloud area
Weaken dust signal
KMA
COMS AI≤-0.5 AI≤-0.85 AI≤-1.0
2016.03.06. 04:00UTC
2016.04.21. 10:00UTC
2016.05.05. 03:00UTC
2017.01.26. 04:00UTC
Most similar to COMS AI
COMS AI Algorithm applied to Himawari-8(3/3)
KMA Test operation of COMS AI applied to Himawari-8
Test operation of COMS AI algorithm based on Himawari-8 using equation (2017.2.20. 0210UTC)
• 30days Dynamic BTV • Application of Equation • Cloud threshold(BTD, BTV) BTD1315 > 0.5 • AI threshold AI ≤ -0.5 AI ≤ -0.85 AI ≤ -1.0
KMA
Example Products Resolution
D*-parameter • Dust detection Using 3 bands(8.6㎛,11㎛,12㎛) of Aqua/Terra
MODIS (Moderate Resolution Imaging Spectradiometer) • D*-parameter > 1 : Dust
1X 1 (km)
Polarized Optical Depth Index • AQUA data use only • polarized method(Fresnel Equation) use to improve detection
problem based on IR(discontinuous dust signal between land and sea)
• Large Nr(dust index) is dust.
Nighttime AOD • Production using retrieved weight from AOD and Artificial
Neural Network(ANN) model to improve detection problem based on IR (discontinuous dust signal between day and night)
• Dust area in D*-parameter use only
AOD/ Dust Height • Production for vertical dust information using AIRS
(Atmospheric Infrared Sounder) of AQUA • AOD and dust Height are produced using 111 BTs and
retrieved weight from BTD thresholds and ANN model
2.3X2.3 (km)
Dust Detection from Low orbit satellite
Productions : D*-parameter,
Polarized Optical Depth Index, Nighttime AOD (MODIS)/AOD, dust height(AIRS)
Period : 2016. 11. 27. ~
Example of test operation
KMA D*-parameter Algorithm
(D*>1: dust , C= -0.5, E= 15) D*= 𝑒𝑒𝑒𝑒𝑒𝑒(𝐵𝐵𝐵𝐵𝐵𝐵 11−12 −𝐶𝐶
𝐵𝐵𝐵𝐵𝐵𝐵 8−11 −𝐸𝐸)
C(BTD11-12 thermal offset) : related to dust area offset C can vary as BTD11-12 is sensitive to dust composition, surface type, and column water vapor E(BTD8-11 thermal offset) : related to dust strength Cirrus cloud(ice) particle absorption: 11㎛>8.6㎛ (BTD8-11: positive) Silicate mineral(dust) absorption : 8.6㎛>11㎛ (BTD8-11: negative) These offsets are adjustable based on analyses of dust-laden MODIS scenes and simulations to enhance the dust signal against a cloud-filled background.
(Hansell et al., 2007)
KMA D*-parameter applied to Himawari-8(1/2)
channel
center wave length
AHI (Himawari-8)
MODIS (Aqua/Terra)
11(IR) 8.6 8.529(B29)
12(IR) 9.6 9.734(B30)
13(IR) 10.4 B30+B31
14(IR) 11.2 11.019(B31)
15(IR) 12.3 12.032(B32)
Use L1B MOD02(1km), MYD02(1km)
MOD02,MYD02 product: 16 bands of Level1B-Calibrated collocated Radiances
1 B20 9 B29(8.40-8.70㎛)
2 B21 10 B30(9.580-9.880㎛)
3 B22 11 B31(10.780-11.280㎛)
4 B23 12 B32(11.770-12.270㎛)
5 B24 13 B33
6 B25 14 B34
7 B27 15 B35
8 B28 16 B36
MODIS(y) AHI(x) Equation
B29(8.529) CH11(8.6) f(x) = 1.0252060391x – 5.6846872174
B31(11.019) CH13(10.4) f(x) =1.0401442194x – 11.530109417
B31(11.019) CH14(11.2) f(x) =1.0286364624x – 8.363347252
B32(12.032) CH15(12.0) f(x) = 1.1336973173x – 36.3084260175
regression coefficient to correct difference of Satellite channel
KMA D*-parameter applied to Himawari-8(2/2)
2017. 05. 05. 0600UTC
MODIS D*_parameter Himawari dust RGB COMS AI HIMAWARI AI (AI≤-1.0)
No Equation BTD13-15(C=-0.85)
Equation BTD13-15(C=-0.85)
No Equation BTD14-15(C=-0.85)
Equation BTD14-15(C=-0.85)
Application of equation to through channel analysis between COMS and Himawari-8 Adjust C threshold(C = -0.85)
Most similar to COMS AI
Detect dust signal
KMA Validation of Himawari-8 D*-parameter with CALIPSO AOD
- AREA Lat: 30.0N - 50.0N, Lon: 80.0E - 135.0E
- DATA CALIPSO: L2 Profile : Column_Optical_Depth_Tropospheric_Aerosols_532 Himawari-8: D*_parameter Index(No equation, C=-0.85) Nearest Himawari-8 pixel within 2km at CALIPSO Lat/Lon
2016.5.1. 06:18UTC
CALIPSO orbit
Dust weather chart
Himawari-8 D*-parameter
result
KMA Summary
Current status of COMS AI - Using BTD and BTV with dynamic threshold, adjust AI to improve the
detection
Apply COMS algorithm to Himawari-8 - Using spectral matching equation and use different BTV
Algorithms for LEO
- D*-parameter, PODI, Nighttime AOD, AOD/Dust Height from LEO
KMA
THANK YOU!