Operational AIRS-MODIS Co-location System AIRS Spatial Response Function & MODIS IR channel Spectral Response Function Haibing Sun 2 , W. Wolf 2 , C. Barnet 1, Lihang Zhou 2 and M. Goldberg 1 1 NOAA/NESDIS/ORA, 5200 Auth Road, Camp Springs, MD 20746 USA 2 QSS Group Inc, Lanham, MD, USA
AIRS-MODIS Collocation AIRS-MODIS Collocation Processing in ORA Processing in ORA • Operational Collocation Algorithms: Co-register observations from AIRS and MODIS • AIRS spatial response function simulation • Algorithm validation analysis and results • MODIS IR channel spectral response function . • Summary
Apply AIRS EFOV Spatial Response Apply AIRS EFOV Spatial Response Function for Collocations Function for Collocations 40.4 0 . 9 0 . 8 The AIRS radiance is contributed by all 40.3 0 . 7 0 . 6 the points within the EFOV of the 0 . 5 40.2 0 . 4 sensor. 0 . 3 40.1 0 . 2 0 . 1 -30.9 -30.8 -30.7 -30.6 -30.5 -30.4 -30.3 -30.2 -30.1 -30 In one AIRS EFOV, hundreds of MODIS Simplized AIRS EFOV SRF .Fp=1 observations are collocated. 40.4 1 0 . 9 0 . 8 A more realistic AIRS EFOV Spatial 40.3 0 . 7 0 . 6 Response Function is used to: 40.2 0 . 5 0 . 4 0 . 3 40.1 0 . 2 0 . 1 -30.9 -30.8 -30.7 -30.6 -30.5 -30.4 -30.3 -30.2 -30.1 -30 1: Select collocated MODIS observations Simulated AIRS EFOV SRF Fp=1 40.4 2: Calculate the weights for each MODIS 40.3 FOV 40.2 40.1 -30.9 -30.8 -30.7 -30.6 -30.5 -30.4 -30.3 -30.2 -30.1 -30 Collocated MODIS observation
AIRS Spatial Response Function AIRS Spatial Response Function Simulation: Physical Model Simulation: Physical Model • The radiance for each AIRS footprint is a combination of 51 instantaneous fields of view (IFOV) • IFOV spatial response function: – Pre-launch measurements – circle, ellipse, rectangle • The spatial response function of the AIRS EFOV is the convolution product of the spatial response function of each IFOV and the integration time. Other factors that we need to take into account are: a: Time integration pattern • b: Scanning pattern / stepwise or continuous • c: Reflect mirror rotation / whether mirror is used • d: Satellite movement • d: Earth rotation • e: No spherical earth • AIRS Sampling model: AIRS scan speed:8/3; 119 Sampling; 90Earthview • 0.022 Integration Period; 51subsampling
AIRS Spatial Response Function AIRS Spatial Response Function Simulation Methodology Simulation Methodology • Time Integration – 0.022 second sampling period – Simulate sub sampling IFOV during the EFOV integration period – Project all sub sampling IFOV spatial response function on earth – Integrate IFOV spatial response function over EFOV sampling time • Sub sampling – 51 times – 51 sub scan angles – 51 sub satellite positions • Satellite orbit calculation – along track movement, 51 pointing vectors • Add Reflection mirror • Use Earth elevation model (WSR84), non spherical • AIRS EFOV spatial response function in earth surface coordinates: – 90 spatial response function for 90 scan angles.
AIRS EFOV Spatial Response Function AIRS EFOV Spatial Response Function Nadir View Nadir View 0.008 0.006 0.004 50 40 0.002 Along the orbit 30 0 20 -0.002 10 -0.004 10 20 30 40 50 60 70 80 90 100 -0.006 -0.008 -0.008 -0.006 -0.004 -0.002 0 0.002 0.004 0.006 0.008 Along the scan fp=45 Lat=10
AIRS EFOV Spatial Response Function AIRS EFOV Spatial Response Function Opposite Edge of Scan Opposite Edge of Scan 0.008 0.006 0.0009 50 0.00085 0.004 0.0008 0.00075 40 0.0007 0.002 0.00065 0.0006 Along the orbit 0.00055 30 0.0005 0 0.00045 0.0004 0.00035 20 -0.002 0.0003 0.00025 0.0002 10 0.00015 -0.004 0.0001 5E-005 0 10 20 30 40 50 60 70 80 90 100 -0.006 -0.008 -0.008 -0.006 -0.004 -0.002 0 0.002 0.004 0.006 0.008 Along the scan fp=1 lat=10
Algorithm Validation Results and Analysis •Validation Method •AIRS SRF application: AIRS pointing correction • The problem in validation: MODIS Spectral Response Function MODIS Spectral Response Function Retrieval • Algorithm Validation with real Data: • Ocean Case Land Case • Polar Case Coastline Case Desert Coastline Case
AIRS EFOV Spatial Response Function AIRS EFOV Spatial Response Function Application Application Operational AIRS-MODIS Colocation processing AIRS-MODIS Colocation processing with simplized AIRS SRF With Simulated AIRD EFOV SRF 7.5 7 .5 25.9 7 7 25.88 6 .5 6.5 6 25.86 6 25.84 5 .5 5.5 25.84 5 5 4 .5 25.82 4.5 4 4 3 .5 25.8 3.5 3 175.9 175.95 176 176.05 176.1 176.15 176.2 176.25 176.3 25.78 2 .5 3 MODIS radiance 176 176.02 176.04 176.06 176.08 176.1 176.12 176.14 176.16 176.18 Simulated AIRS EFOV SRF 2.5 MODIS Radiance No simulated AIRS EFOV SRF 1 0.9 25.9 0.95 0.85 0.9 0.8 0.85 25.9 0.75 25.88 0.8 0.7 0.75 0.65 0.7 25.86 0.6 0.65 25.85 0.55 0.6 0.55 0.5 25.84 0.5 0.45 0.45 0.4 25.8 0.4 25.82 0.35 0.35 0.3 0.3 0.25 25.8 0.25 0.2 25.75 0.2 0.15 0.15 25.78 0.1 0.1 175.9 175.95 176 176.05 176.1 176.15 176.2 176.25 176.3 0.05 0.05 0 0 176 176.02 176.04 176.06 176.08 176.1 176.12 176.14 176.16 176.18 Weight in Degrading Weight in Degrading Simulated AIRS EFOV SRF Simple AIRS SRF Simple SRF Realistic SRF
AIRS EFOV Spatial Response Function AIRS EFOV Spatial Response Function Application: Pointing Bias Correction Application: Pointing Bias Correction 1.00E-3 8.00E-4 Bias setting: x: 30 mili degree y: 100mili degree 6.00E-4 4.00E-4 2.00E-4 0.00E+0 0 10 20 30 40 50 60 70 80 90 Pointing direction bias from centroid: In scan direction In orbit direction
Low Latitude Region (Land) : 4/16/119 Ascending Low Latitude Region (Land) : 4/16/119 Ascending Realistic SRF Simple SRF Radiances
Coastline: 04/17/143 Coastline: 04/17/143 Simple SRF Radiance Realistic SRF 20 20 140 18 130 18 16 120 16 110 14 14 100 12 12 90 10 10 80 8 70 8 6 60 6 4 50 4 40 2 2 30 0 0 20 -2 10 -2 -20 -15 -10 -5 -20 -15 -10 -5 With Simulated AIRS SRF
Desert Coastline: 04/17/111 Desert Coastline: 04/17/111 Realistic SRF Simple SRF Radiance
MODIS Response Function Retrieval • Part II MODIS IR channel Spectral Response Function
MODIS Relative Spectral Response Function MODIS Relative Spectral Response Function (RSRF) Problem (RSRF) Problem For the MODIS infrared channel 20-36, there is no way to monitor the instrument spectral sensitivity. Recent researching about the MODIS observation data show that there may be different instrument spectral response function shifting in these 16 infrared channels. For AIRS-MODIS co-location processing, an observation data based algorithm is developed to retrieve on-orbit ‘Broad band’ MODIS spectral response function with co-located AIRS high spectral resolution observation data. The retrievaled result provide the information of the ‘real’ MODIS RSRF on the observation system level. Those information can be used to improve the integration quality of different instrument and provide inter-instrument calibrilication ability.
Basic Methodology • Basic equation: Individual MODIS IR channel Rad Rad W = � MODIS AIRS i i • Rad (MODIS): Low spectrum resolution MODIS observation • Rad (AIRS)i High spectrum resolution AIRS observation • W i Weighting defined by spectral response function • To resolve the W i , A equation group is need: A X M = ij j i • A (I,j): AIRS observation matrix • X (j) Weighting vector • M (ij MODIS observation vector • I > J is required • Key Assumption: All no-linear behavior is in-depend of wave length.
AIRS Inter-channel Sampling 120 1.00 SPEC35 MODIS Channel 35 Spectrum Response Function 0.80 100 MODIS IR Channel(35) RSF AIRS Radiance 0.60 80 0.40 60 0.20 40 0.00 700 705 710 715 720 725 730 735 740 MODIS channel 35 Spectrum
Deficient Matrix inverse Problem • The X can be solved by multiplying both sides by the pseudo-inverse of A. But A is invariably rank deficient. • Physically, X is continues with un-limited dimensionality. The A and M are tend to be low dimensionality. • With rank deficient A, M, Retrieval of X based on matrix inversion are very sensitive to noise
Retrieval with Simulated Data 1.000 Simulated Data with zero noise to valid algorithms 0.800 Pre-launch Measurement SRF Retrievaled with U=0.01 AVE:1 SVD:111 0.600 0.400 0.200 0.000 0 10 20 30 40 50 60 70 80 90 100 110 120 MODIS CHANNEL 36( AIRS channel) Noise Free Data
Retrieval with Simulated Data 0.018 1.20 Retrieval with Simulted AIRS -MODIS Data with Addtional Noise: 0.02*MODIS(ch)*[-1,1] 0.016 U = 0.00 SVD:111 Data with Noise NOISE = [-0.5,0.5] 0.80 Retrieval with Real Observated AIRS -MODIS Data 0.014 U = 0.00 SVD:111 0.012 0.40 0.010 0.008 0.00 0.006 0.004 -0.40 0.002 0.000 -0.80 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 105 110 115 Channel 36 (AIRS ch Point index)
Recommend
More recommend
