ATMS Contributions to Sounding Products

1
ATMS Contributions to Sounding Products
David H. Staelin and Philip W. Rosenkranz Resear
ch Laboratory of Electronics Massachusetts
Institute of Technology Cambridge,
Massachusetts Presented at the NPP Science
Team Meeting Annapolis, Maryland November 4-6,
2003
2
ATMS Center Frequencies (GHz)
Early Concept
2200-km swath width (no gaps)
3
Contributions to DeliverablesRole of PIs

• Participate in NPP Science Team meetings give
  reports
• Review sensor and algorithm documents
• Perform data simulation studies and technology
  transfer
• Support document preparation by NPP staff
• Algorithm analysis and improvement report
• Science operations concept document
• NPP calibration-validation plan
• PI/Co-PI
• Have participated in most microwave sounding
  satellite programs
• Are familiar with instrumentation, retrieval, and
  user issues
• Will share in all contributions to deliverables

4
Approach to EDR/CDR Contributions

• AIRS/AMSU/HSB heritage, also NOAA-15/16/17
• First-guess temperature, humidity, cloud liquid
  water
• Microwave transmittances, radiative transfer,
  surface emissivity (Aqua and NOAA-15/16/17
  data)
• ATMS precipitation products rain and snowfall
  rates and frequencies (Aqua and NOAA-15/16/17
  data)
• Optimal interpolation of ATMS data registration
  with IR pixel and beam sharpening for 23/31-GHz
  data (simulations)
• Field experiments and instrument issues (as
  appropriate)
• Stochastic CrIMSS cloud clearing (as feasible)

5
Global Precipitation NOAA-16 AMSU-A/B,15-km
resolution
Chen Staelin, Oct 2003
6
NOAA-15/16/17 AMSU Rain Frequencies0.5 0.5,
1-7 July 2003
1
0.8
0.6
Rain Frequency
0.4
0.2
0
Chen Staelin, Oct 2003
7
Preliminary Stochastic Cloud Detection Algorithm
Step 1
284 15-micron channels AIRS AIRS
15 Principal Components of ? (cloud sensitive)
PCA
cloud cleared TB
284 15-micron AIRS channels
Step 2
PCA
15 Principal Components for ? cloud
Neural Net
3 AMSU 54-GHz channels
gt?? ? cloud
Training to estimate
385 15-micron channels AIRS AIRS
cloud cleared TB
Cho Staelin, Oct 2003
8
Detection of Cloud Perturbations gt 1.5K

Estimated cloud-clearing TB
Estimated cloud-clearing TB
AIRS TB
229
ROC curve for detecting erroneous pixels
20
228
1
? 0.5
0.9
40
227
0.8
60
226
0.7
? 1
80
225
0.6
PD
100
224
? detector threshold
0.5
0.4
120
223
? 1.5
0.3
140
222
? 2
0.2
160
221
0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
180
220
PF
10
20
30
10
20
30
10
20
30
15-micron channel weighting function peaking at
7.2 km

• 1.5K
• 12 flagged
• PD 44
• PF 3

• 1.0K
• 30 flagged
• PD 76
• PF 18

Probability of False Alarm
Cho Staelin, Oct 2003