Vis/NIR Data Quality, Cloud Detection and IR co-registration - - PowerPoint PPT Presentation

▶
vis nir data quality cloud detection and ir co
SMART_READER_LITE
LIVE PREVIEW

Vis/NIR Data Quality, Cloud Detection and IR co-registration - - PowerPoint PPT Presentation

Feb 16, 2024 415 likes •918 views

Vis/NIR Data Quality, Cloud Detection and IR co-registration Catherine Gautier and Yang Shiren Institute of Computational Earth System Science UCSB Gautier - Yang AIRS Meeting 9/19/02 1 Overview Data Quality Stripes

slide-1
SLIDE 1

Gautier - Yang AIRS Meeting 9/19/02 1

Vis/NIR Data Quality, Cloud Detection and IR co-registration

Catherine Gautier and Yang Shiren Institute of Computational Earth System Science UCSB

slide-2
SLIDE 2

Gautier - Yang AIRS Meeting 9/19/02 2

Overview

  • Data Quality
  • Stripes
  • Co-alignment between Vis/NIR and AIRS
  • Effect Registration between Vis/NIR and

1993 NDVI on cloud detection

  • Cloud Detection
  • Future Work
slide-3
SLIDE 3

Gautier - Yang AIRS Meeting 9/19/02 3

Stripes

  • Stripes caused by difference of gain and
  • ffset between 9 Vis/NIR elements
  • In-between element adjustment + vicarious

calibration used to remove stripes

  • Remaining small stripes in different

conditions

  • Detailed investigation of stripes

characteristics

slide-4
SLIDE 4

Gautier - Yang AIRS Meeting 9/19/02 4

Gain computed over bright surface

slide-5
SLIDE 5

Gautier - Yang AIRS Meeting 9/19/02 5

Stripes in Sunglint area

slide-6
SLIDE 6

Gautier - Yang AIRS Meeting 9/19/02 6

slide-7
SLIDE 7

Gautier - Yang AIRS Meeting 9/19/02 7

slide-8
SLIDE 8

Gautier - Yang AIRS Meeting 9/19/02 8

slide-9
SLIDE 9

Gautier - Yang AIRS Meeting 9/19/02 9

slide-10
SLIDE 10

Gautier - Yang AIRS Meeting 9/19/02 10

slide-11
SLIDE 11

Gautier - Yang AIRS Meeting 9/19/02 11

Stripes Analysis: Preliminary Results

  • Suggestion of a non-linear gain and/or

BRDF effects due to scanning geometry

slide-12
SLIDE 12

Gautier - Yang AIRS Meeting 9/19/02 12

Registration between Vis/NIR and IR

  • Use transition targets to evaluate quality
  • f Vis/IR alignment/co-registration
  • Investigated E-W and N-S alignment
slide-13
SLIDE 13

Gautier - Yang AIRS Meeting 9/19/02 13

slide-14
SLIDE 14

Gautier - Yang AIRS Meeting 9/19/02 14

slide-15
SLIDE 15

Gautier - Yang AIRS Meeting 9/19/02 15

slide-16
SLIDE 16

Gautier - Yang AIRS Meeting 9/19/02 16

slide-17
SLIDE 17

Gautier - Yang AIRS Meeting 9/19/02 17

slide-18
SLIDE 18

Gautier - Yang AIRS Meeting 9/19/02 18

slide-19
SLIDE 19

Gautier - Yang AIRS Meeting 9/19/02 19

slide-20
SLIDE 20

Gautier - Yang AIRS Meeting 9/19/02 20

slide-21
SLIDE 21

Gautier - Yang AIRS Meeting 9/19/02 21

Preliminary Results

  • Very good within accuracy of approach
  • Need longer data set over same

transition targets to quantitatively evaluate alignment

slide-22
SLIDE 22

Gautier - Yang AIRS Meeting 9/19/02 22

Registration between Vis/NIR and 1993 NDVI

  • 1993 NDVI used in computation of cloud

threshold as a surface type (reflectance) indicator

  • Preliminary analysis=>mis-alignment

between NDVI 1993 and AIRS Vis/NIR NDVI

  • Use transition targets to evaluate Vis/IR

alignment/co-registration

  • Investigate E-W and N-S alignment
slide-23
SLIDE 23

Gautier - Yang AIRS Meeting 9/19/02 23

Selection of Transition Targets

slide-24
SLIDE 24

Gautier - Yang AIRS Meeting 9/19/02 24

slide-25
SLIDE 25

Gautier - Yang AIRS Meeting 9/19/02 25

slide-26
SLIDE 26

Gautier - Yang AIRS Meeting 9/19/02 26

Selection of Transition Targets

slide-27
SLIDE 27

Gautier - Yang AIRS Meeting 9/19/02 27

slide-28
SLIDE 28

Gautier - Yang AIRS Meeting 9/19/02 28

Impact on Cloud Detection

slide-29
SLIDE 29

Gautier - Yang AIRS Meeting 9/19/02 29

Registration between Vis/NIR and 1993 NDVI : Preliminary Results

  • Very good N-S alignment
  • 1-2 Vis/NIR pixel E-W misalignment
  • Found some changes related to time

difference between 2002 and 1993

  • Will be developing AIRS NDVI soon
slide-30
SLIDE 30

Gautier - Yang AIRS Meeting 9/19/02 30

Vis/NIR Cloud Detection

  • Based on cloud threshold
  • Cloud threshold: parameters adjusted to

preliminary calibration

– Visual assessment of cloud detection very good

  • Some issues due to 1993 NDVI
  • Comparison with spatial inhomogeneity

approach results

slide-31
SLIDE 31

Gautier - Yang AIRS Meeting 9/19/02 31

Ocean Conditions: Sunglint

slide-32
SLIDE 32

Gautier - Yang AIRS Meeting 9/19/02 32

Ocean Conditions: scene selection

slide-33
SLIDE 33

Gautier - Yang AIRS Meeting 9/19/02 33

Vis/IR cloud mask vs. spatial homo

slide-34
SLIDE 34

Gautier - Yang AIRS Meeting 9/19/02 34

Vis/IR cloud mask vs. spatial homo

slide-35
SLIDE 35

Gautier - Yang AIRS Meeting 9/19/02 35

Vis/IR cloud mask vs. spatial homo

slide-36
SLIDE 36

Gautier - Yang AIRS Meeting 9/19/02 36

Ocean conditions: hurricane

slide-37
SLIDE 37

Gautier - Yang AIRS Meeting 9/19/02 37

Ocean Conditions: scene selection

slide-38
SLIDE 38

Gautier - Yang AIRS Meeting 9/19/02 38

Vis/IR cloud mask vs. spatial homo

slide-39
SLIDE 39

Gautier - Yang AIRS Meeting 9/19/02 39

Vis/IR cloud mask vs. spatial homo

slide-40
SLIDE 40

Gautier - Yang AIRS Meeting 9/19/02 40

Vis/IR cloud mask vs. spatial homo

slide-41
SLIDE 41

Gautier - Yang AIRS Meeting 9/19/02 41

Land Conditions

slide-42
SLIDE 42

Gautier - Yang AIRS Meeting 9/19/02 42

Land Conditions: scene selection

slide-43
SLIDE 43

Gautier - Yang AIRS Meeting 9/19/02 43

Vis/IR cloud mask vs. spatial homo

slide-44
SLIDE 44

Gautier - Yang AIRS Meeting 9/19/02 44

Vis/IR cloud mask vs. spatial homo

slide-45
SLIDE 45

Gautier - Yang AIRS Meeting 9/19/02 45

Cloud Detection : Preliminary Results

  • Very good visual agreement
  • Impact of 1-2 Vis/NIR pixel E-W misalignment

and changes related to time difference between 2002 and 1993 will be addressed by AIRS NDVI soon

  • Improvements based on spatial homogeneity

approach

slide-46
SLIDE 46

Gautier - Yang AIRS Meeting 9/19/02 46

slide-47
SLIDE 47

Gautier - Yang AIRS Meeting 9/19/02 47

Near-term Activities

  • Cloud flag over land evaluation
  • Validation of cloud detection against

local meteorological data and ARM/CART data

  • Improved validation of geo-location and

co-registration

  • Vicarious calibration
slide-48
SLIDE 48

Gautier - Yang AIRS Meeting 9/19/02 48

Vicarious Calibration

  • Surface observation-based calibration

suggests some surface BRDF effects that need to be taken into account

  • Development of BRDF model based on

satellite observations

  • Adaptation of UCSB RTM BRDF model

to Railroad Valley surface conditions

slide-49
SLIDE 49

Gautier - Yang AIRS Meeting 9/19/02 49

Longer-term Activities

  • Use combined Vis/NIR and IR channels

for low cloud detection

  • IR Spatial Inhomogeneity vs. Vis/NIR

cloud threshold

  • Cloud fraction comparisons
  • Vis/NIR spectral response function
  • Surface radiation budget modeling
  • Land surface spectral emissivity