infrared and microwave remote sensing of sea surface temperature
DESCRIPTION
Infrared and Microwave Remote Sensing of Sea Surface Temperature. Gary A. Wick NOAA Environmental Technology Laboratory January 14, 2004. Outline. Motivation Basic SST Retrieval Methods Current Multi-Sensor Merging Efforts. Why SST?. Boundary Condition Weather Models - PowerPoint PPT PresentationTRANSCRIPT
Infrared and Microwave Remote Sensing of Sea Surface Temperature
Gary A. Wick
NOAA Environmental Technology Laboratory
January 14, 2004
Outline
Motivation Basic SST Retrieval Methods Current Multi-Sensor Merging Efforts
Why SST?
Boundary Condition
– Weather Models
– Estimation of Heat Content and Heat Flux Climate Monitoring and Change Detection Naval Operations
Climate Anomalies
Courtesy: NOAA Climate Diagnostics Center
Why Satellites?
Courtesy: R. Reynolds, NOAA NCDC
Desired Accuracy
WCRP (1985) - Tropics
– 0.3 K on 2° grid every 15 days Robinson et al. (1984) - Global SST Monitoring
– 0.05 K on 5° grid every 15 days
NPOESS SST EDR Objectives
– 0.1 K uncertainty at ~4 km resolution
Definition of SST
Interface SST Skin SST Sub-skin SST Near-Surface SST
or SSTDepth
Radiative Transfer Equation
Methods for SST Retrieval
Thermal Infrared Passive Microwave
Infrared Retrievals
Strengths
– High Accuracy
– High Resolution
– Long Heritage (over 20 years) Weaknesses
– Obscured by Clouds
– Atmospheric Corrections Required
Microwave Retrievals
Strengths
– Clouds Transparent
– Relatively Insensitive to Atmospheric Effects Weaknesses
– Sensitive to Surface Roughness
– Poorer Accuracy (?)
– Poorer Resolution
Spatial Coverage Differences
Infrared Retrieval Technique
Cloud Detection
Atmospheric Correction Multi-Channel SST
– TS = T1 + (T1 - T2)
– Multi-Frequency
– Multiple View
Algorithm Refinements
Additional path length term NLSST Use of multiple frequencies AND multiple view angles Independent estimate of water vapor content Iterative solution for both SST and
Microwave Retrieval Technique
Environmental Scenes42,195 Radiosondes
5 Cloud ModelsSST Randomly Varied for 0 to 30 C
Wind Speed Randomly Varied from 0 to 20 m/sWind Direction Randomly Varied from 0 to 360
Complete Radiative Transfer Model
Simulated AMSR TB's
Truth: Ts, W, V, L
Gaussian Noise Added
Derive Coefficients for Multiple Linear Regression Algorithm
Withheld Data Set
Algorithm Coefficients
Run Algorithm
Evalulate Algorithm Peformance
Retrieved values for Ts, W, V, L
Performance and Cross Talk Statistics
Courtesy: Remote Sensing Systems
Infrared Sensors
AVHRR ATSR GOES Imager MODIS
Others
– GMS
– SEVIRI
– VIRS
Microwave Sensors
TMI AMSR WindSat
Multi-Sensor Blended SST
Current Projects Key Issues Sample Results
GODAE High-Resolution SST Pilot Project
Provide rapidly and regularly distributed, global, multi-sensor, high-quality SST products at a fine spatial and temporal resolution
– Most promising solution to combine complementary infrared and passive microwave satellite measurements with quality controlled in situ observations from ships and buoys
www.ghrsst-pp.org
Next Generation SST
Created by Hiroshi Kawamura, Tohoku University, Japan http://www.ocean.caos.tohoku.ac.jp/~adeos/sst/
Blended SST Issues
Different product resolutions Different sensor error characteristics Different sampling times and effective depths Merging techniques
Error Characteristics – Overall Accuracy
Observed Differences Between Infrared and Microwave Products
Comparisons between the products show complex spatial and temporal differences
Sources of Product Differences
Diurnal Warming Effects
Skin Layer Effects
Courtesy: P. Minnett, U. Miami
Courtesy: S. Castro, U. Colorado
NOAA Environmental Technology Laboratory
Blended Infrared andMicrowave SST
Using derived corrections, the infrared and microwave SST products can be more accurately merged into a new enhanced product.
Diurnal warming effects are aliased into the product if not corrected.
Strong winds off Somalia cause perceived overcooling and large swath edge effects are visible.
Bias(K)
RMS(K)
w/ Adj -0.01 0.61
w/o Adj 0.15 0.67
Accuracy of MergedProduct vs. Buoys
Analyzed SST Product
Daily global (40 N – 40 S) 0.25 degree
Referenced to nighttime predawn value
Based on Reynolds and Smith Optimal Interpolation
Relative product uncertainties derived from difference analyses
Analysis Characteristics
Analyzed Product Accuracy Summary
Product Bias (K) RMS (K)
Full Analysis 0.13 0.68
Night Obs Only -0.08 0.58
AVHRR Obs Only -0.01 0.56
TMI Obs Only 0.22 0.74
Refined diurnal corrections are the most needed improvement
Summary
Complementary infrared and microwave SST products provide the opportunity for cross-validation and improved SST
Multiple sensor-related and geophysical effects lead to complex differences between the products
Optimal blending of the products requires careful treatment of the differences
Is blending correct?