7.14 DIFFERENCES BETWEEN INFRARED AND MICROWAVE SEA SURFACE TEMPERATURE PRODUCTS
Gary A. Wick*1 and Donna J. Scott2 1NOAA Environmental Technology Laboratory, Boulder, CO 2CIRES, University of Colorado, Boulder, CO
1. INTRODUCTION SST products computed from the AVHRR and TMI sensors. The AVHRR-based data is the nonlinear Infrared and passive microwave satellite sea surface temperature (NLSST) product from sensors provide highly complementary information the NOAA-14 satellite produced operationally by for the creation of sea surface temperature (SST) the Naval Oceanographic Office (May et al., products. Infrared sensors provide high resolution 1998). The TMI SST product is produced and and high accuracy estimates of the SST but the distributed by Frank Wentz and Remote Sensing measurements are completely obscured by Systems, Inc. We did not apply additional clouds. Passive microwave sensors, in contrast, processing to these datasets with the exception of have poorer resolution but are able to obtain gridding the NLSST data to a 0.25° grid common measurements through non-precipitating clouds. to the TMI SST and separating the TMI SST into By combining the measurements from the two day and nighttime observations. All our initial sensors it is possible to produce an all-weather, comparisons were performed at the 0.25° high-resolution SST product. We are currently resolution of the TMI SST. developing a blended SST product using infrared Differences were computed between daily data from the Advanced Very High Resolution daytime and nighttime averages of the two Radiometer (AVHRR) and passive microwave products. These daily differences were then data from the Tropical Rainfall Mapping Mission further averaged over a month to provide a better (TRMM) Microwave Imager (TMI). spatial pattern of the differences for comparison The first important step in developing a with environmental conditions and the expected method for combining the two products is behavior of the skin layer. Example average understanding the error characteristics and differences for November, 2000 are shown in differences between the products. In this work we Figure 1. The mean daytime difference is -0.43 K examine the differences between the products in (TMI SST larger) and the mean nighttime detail. We devote special attention to the relation- difference is –0.55 K. RMS differences are on the ship between the differences and both diurnal order of 0.8 K. warming and the temperature gradient that These mean differences are generally commonly exists at the skin of the ocean. Infrared consistent with an assessment of the accuracy of sensors measure radiation emitted only from the each product relative to quality controlled drifting top 10 microns of the ocean while microwave and moored buoys. We independently radiometers measure radiation from depths down constructed matchups between each product and to approximately 1 mm. The skin layer of the buoy data that showed (for the same month) the ocean can lead to significant and variable TMI SST data to be biased warm by approximately differences between the temperatures at these 0.25 K and the NLSST to be biased cold by near depths. We examine whether differences between 0.1 K. the products are consistent with the presence of Significantly different mean differences were the skin layer and whether there are obtained if the NLSST and TMI products were circumstances where the skin layer complicates or composited over a week or month before being prevents the combination of the products. compared. This illustrates the importance of the contrasting SST sampling characteristics (e.g. 2. OBSERVATIONS OF DIFFERENCES clear sky only vs. clear and cloudy) of the sensors. In addition to the mean offsets, the differences We have analyzed over a year of separate show significant spatial variations. Understanding the source and nature of these variations is critical to developing a strategy to combine these Corresponding author address: Gary A. Wick, products. In our presentation we will provide a NOAA ETL, R/ET7, 325 Broadway, Boulder, CO, detailed analysis of these differences relative to 80305; e-mail: [email protected] such factors as sampling characteristics, environmental conditions, diurnal heating, and supported by the NOAA Climate Change Data and presence of a skin layer. Detection Program.
3. ACKNOWLEDGEMENTS 4. REFERENCE
We would like to thank both Frank Wentz of May, D. A., M. M. Parmeter, D. S. Olzewski, and Remote Sensing Systems and the Naval B. D. McKenzie, 1998: Operational Processing Oceanographic Office for providing the TMI and of satellite sea surface temperature retrievals NLSST data respectively. The buoy data was at the Naval Oceanographic Office, Bull. provided by both Dr. Richard Reynolds and the Amer. Meteor. Soc., 79, 397-407. Naval Oceanographic Office. This work is
NLSST - TMI SST Daytime Differences: Nov 1-30, 2000 40
20
0
-20
-40 0 60 120 180 -120 -60 0
-1.0 -0.6 -0.2 0.2 0.6 1.0 K
NLSST - TMI SST Nighttime Differences: Nov 1-30, 2000 40
20
0
-20
-40 0 60 120 180 -120 -60 0
-1.0 -0.6 -0.2 0.2 0.6 1.0 K
Figure 1. Average daily daytime (top) and nighttime (bottom) difference between the TMI SST and NLSST products.