OS33A-01
MOBY Normalized Water-Leaving Radiance Time-series Uncertainty Reduction for Improved Multi-platform Satellite Sensor Vicarious Calibration
The Marine Optical Buoy (MOBY), a radiometric buoy stationed in the waters off Lanai, Hawaii, is the primary ocean observatory for vicarious calibration of satellite ocean color sensors. Since late 1996, MOBY has been the primary basis for the on-orbit vicarious calibrations of the USA Sea-viewing Wide Field-of-view Sensor (SeaWiFS), the Japanese Ocean Color and Temperature Sensor (OCTS) and Global Imager (GLI), the French Polarization Detection Environmental Radiometer (POLDER), the USA Moderate Resolution Imaging Spectrometers (MODIS, Terra and Aqua), the Japanese Global Imager (GLI), and the European Medium Resolution Imaging Spectrometer (MERIS). The MOBY vicarious calibration of these sensors supports the international effort to develop a global, multi-year time series of consistently calibrated ocean color data products. A longstanding goal of the Ocean Color Science Teams is to determine satellite-derived normalized water-leaving radiance (LWN) with a combined standard uncertainty of 5 percent. A critical component of this approach is to reduce uncertainties in MOBY in situ LWN data. As has been the case since the first MOBY deployment, these improvements are achieved incrementally and from a variety of system aspects. We will discuss these efforts and present results relating to the radiometric calibration, instrument stability during deployments, sensitivity to temperature, stray light corrections, data acquisition protocols, and instrument self shading.
OS33A-02
Reanalysis of GODAE High Resolution SSTs and Critical Baseline Datasets
The GODAE High Resolution SST (GHRSST) project is delivering a large and growing number of forward-mode
Level 2 SST data streams from individual sensors like AVHRR, MODIS, GOES, AATSR, SEVIRI, and AMSR-E. In
addition, more and more Level 4 gap-free analysis products are being produced by various centers around the
world and archived at NODC's GHRSST Long Term Stewardship and Reanalysis Facility. Different groups are
also producing critical baseline datasets, created by reprocessing individual satellite datasets using consistent
algorithms in a retrospective mode. One example of these baselines is the AVHRR Pathfinder dataset, widely
used in many applications. Their relative longevity, greater accuracy, and improved consistency make these
baselines datasets fundamentally important for the proper production of merged, multi-sensor climate data
records for SST. The current state of these critical baselines will be reviewed and presented in the context of the
GHRSST Reanalysis program.
http:ghrsst.nodc.noaa.gov
OS33A-03
Reconstruction of past decades sea level using thermal expansion, tide gauge and satellite altimetry data
Several studies have developed methods for reconstructing past 2-D time series of oceanographic fields (e.g., sea surface temperature, sea surface height, surface pressure) by combining 2-D data of limited temporal coverage (in general available from satellite observations over the last 2 decades or less) with historical (several decades-long), sparse 1-D records. EOF spatial modes of the gridded data are fitted to the 1-D records to provide reconstructed long-term 2-D fields. An implicit assumption of the method is that spatial patterns of the EOF leading modes computed from the short-term 2-D fields are stationary. The objective of the present study is to test the influence of the temporal coverage of the 2-D fields on the reconstructed signal. For that purpose we use global grids of steric sea level data, available over the last 53 years. Global sea level trends computed over that period are used as reference. The 50-year long reconstructed steric sea level signal is computed using different time spans (in the range 10-50 years) for the EOF decomposition. Different geographical distributions for the 1-D steric 53-year long steric sea level time series (interpolated at specific locations from the 2-D grids) are also considered. Then we determine thresholds (in terms of 2-D fields temporal coverage and 1-D records spatial distribution) for which the reconstructed 2-D fields are correctly recovered. In a second step, we apply the method for reconstructing 2-D sea level data over the past 53 years, combining sparse tide gauge records with 2-D steric sea level EOFs computed over the last 53 years. In effect, several recent studies have shown that the spatial patterns of sea level trends recovered by satellite altimetry are very similar as those of the steric sea level. We also reconstruct past sea level over 1950-2003 using 2-D grids of satellite altimetry sea level over 1993-2003. We focus on spatial patterns of reconstructed sea level trends as well as on the global mean sea level. Comparison of reconstructed sea level with tide gauge records is also performed. The two methods of past sea level reconstruction(ie, spatial covariances based on thermal expansion grids and satellite altimetry)are compared and discussed.
OS33A-04
Seasonal to Interannual Variability of Sea Surface Temperature in the Gulf of Panama
Satellite derived sea surface temperatures (SSTs), in-situ data, and output from a Regional Ocean Modeling System (ROMS) are used to study the seasonal-to-interannual variability in the Gulf of Panama. Monthly Climatological data from SSTs clearly show a drop of 8 degrees Celsius during the late winter-early spring surface cooling. The cooling is significantly correlated with local wind data and confirms its relationship to the panama wind jets. This confirms that surface cooling is due to local wind induced upwelling. On the interannual time scale, however, SST shows no correlation with the local wind forcing. Both satellite derived SST data and in-situ data indicate the interannual variability is dominated by offshore warming of greater than 2 degrees Celsius during El Nino years. From these in situ measurements, we hypothesized the seasonal scale is driven by local wind events, while the interannual scale is driven by remote forcing associated with the El Nino-La Nina oscillation. Comparisons with ocean model simulations that were carried out on the Columbia computer managed through the NASA Advanced Supercomputing division indicate that to properly simulate the annual cooling, high-resolution QuikSCAT 25km winds are necessary, as opposed to 2 degree NCEP winds. We concluded that the high- resolution25km winds are necessary to resolve the mountain gaps associated with the Panama Jets.
OS33A-05
Radiative Transfer Modeling of AVHRR Brightness Temperatures for Improved Sea Surface Temperature Retrievals: Initial Results
Operational Sea Surface Temperature (SST) products at NOAA/NESDIS have been derived from the Advanced Very High Resolution Radiometers (AVHRR) onboard NOAA satellites since the early 1980s. The two major SST algorithms, the multi-channel and non-linear SST (MC/NLSST), were initially introduced for the AVHRR in the mid- 1980s and mid-1990s, respectively. Both algorithms are based on solid physical principles but also include some empirical elements (such as for instance treatment of the view angle dependence in the SST equations and tuning their coefficients against in-situ SST), to account for approximations and assumptions in their derivation. The simple MC/NLSST formulations proved to be accurate and robust, and are still in use with the newer sensors, such as the Moderate Resolution Imaging Spectro-Radiometer (MODIS). They are also considered for the future sensors such as the Visible Infrared Imager and Radiometer Suite (VIIRS) onboard NPOESS, and the Advanced Baseline Imager (ABI) onboard GOES-R. Despite the apparent success and good accuracies, improved SST formulations should be explored based on improved and accurate Radiative Transfer Models (RTM). In order to be useful for inverse problem simulation (i.e., SST retrieval), the RTM should be able to adequately reproduce the top-of-the-atmosphere brightness temperatures. In this study, we use MODTRAN 4.2 model, coupled with two surface reflection models (black body and Fresnel), to simulate TOA brightness temperatures in the three thermal infrared AVHRR bands. Atmospheric profiles and SST come from the National Centers for Environmental Prediction (NCEP) Global Data Analysis System (GDAS) data as input, assuming aerosol- and cloud-free conditions. The simulation results are then convoluted with the respective relative spectral response functions of the individual sensors, and compared with collocated TOA brightness temperatures measured from NOAA-16, -17, and -18 satellites during nighttime. Initial results suggest, model brightness temperatures are biased high with respect to measured, in all bands. Bias is smallest in AVHRR channel 4 (11 μm; a few tenths of Kelvin), largest in channel 5 (12 μm; more than 1K), with channel 3B (3.7 μm) falling in between. In all bands, RTM slightly underestimates angular dependence. Including surface reflectance improves agreement between RTM and measurements, in all bands, but still measurable differences exist. Agreement also improves in areas which are more densely populated with satellite data, suggesting that AVHRR SSTs may be subject to residual cloud. Possible causes of the differences between RTM simulations and AVHRR measurements, and ways to reconcile them are discussed.
OS33A-06
Sea Surface Global Climate Datasets With Compatible High Resolutions
Present day global ocean observing system consists of multiple satellites and in-situ platforms. Blending of these observations has made it possible to produce gridded global climate datasets with increasingly higher resolutions that are demanded by the research and operational forecast communities. However, caution must be exercised when producing and utilizing global high resolution products: under-sampling could result in significant alias errors for variables with higher frequency variability. The resolutions of the blended products have to be compatible with the available observational data density or frequency. In this paper we present a case study, taking sea surface wind speed as an example. Sea surface wind speed has been observed from multiple satellites and in-situ instruments. These long-term satellites ranged from one DMSP (the Defense Meteorological Satellite Program) satellite (F08) in mid 1987 to the present six or more satellites since June 2002. We shall show that on a global 0.25° grid, blended products with temporal resolutions of 6-hours, 12-hours and daily have become feasible since mid 2002, mid 1995 and January 1991, respectively (with greater than 75 percent time coverage and greater than 90 percent spatial coverage between 65°S-65°N). Thus, for a uniform long-term climate product on a global 0.25° grid and over the whole time period (July 1987 to present), a near Gaussian 3-D (x, y, t) interpolation was used with the spatial and time windows of 125 km and 12-hours. To take advantage of the high data density of the later years (since mid 2002), 4 times per day snapshots have been generated. Documentation of the feasibility study, data production, data visualization, sub-setting and downloading can be obtained at: http:www.ncdc.noaa.gov/oa/satellite.html; http:nomads.ncdc.noaa.gov:8085/las/servlets/dataset; ftp:eclipse.ncdc.noaa.gov/pub. Our analysis shows that the unique sampling times of the AMSR-E are largely responsible for the feasibility of producing 6-hourly products, filling the gaps between the DMSP satellites that tend to cluster together. Even spacing should be addressed in future satellite missions, as has been done in NPOESS to some extent.
OS33A-07
Gap filling in satellite datasets for Southern Ocean
Variability of sea-level wind and sea-surface temperature (SST) are strongly related to each other in regions of enhanced SST gradients. Among such regions of the World Ocean, the Southern Ocean stands out due, in part, to its special role in linking the Atlantic, Pacific, and Indian oceans. Quantitative description of the sea-level-wind--SST feedbacks in the Southern Ocean is especially challenging due to the complex nature of boundary-layer processes, as well as their coupling to the intrinsic dynamics of the fluids on both sides of the ocean--atmosphere interface. Recent high-resolution records from remote sensing over the Southern Ocean are crucial to understand these coupling effects. However such data are full of gaps which naturally arise from the limited satellite coverage. Depending on instrumentation, these measurements are usually hampered by clouds, aerosols, or heavy precipitation. We utilize the data for SST and sea-level wind over the Southern Ocean from the NASA satellite-data products, such as AVHRR (SST) and QuikSCAT (surface winds), to construct time-space continuous data records. We apply singular spectrum analysis to fill the missing data with smooth information from an iteratively inferred "signal" that represents coherent spatio-temporal structures (Kondrashov and Ghil, 2006).
OS33A-08
The Multi-sensor Improved SST (MISST) Project
The MISST project focuses on (1) producing an improved global, daily, high-resolution, sea surface temperature
(SST) product through the combination of observations from complementary infrared and microwave sensors and
(2) demonstrating the impact of improved multi-sensor SST products on operational ocean models, numerical
weather prediction, and tropical cyclone intensity forecasting. Producing a high-quality multi-sensor SST requires
careful inter-calibration of different satellite sensors, calculation of sensor-specific observation errors that
consider environmental variables, location of observation, time of day, and sensor calibration problems;
development of techniques for relating and combining measurements at different depths, spatial resolutions, and
times of the day; and implementation of data fusion methodologies.
A number of new SST data sets and several blended global SSTs are currently being produced by this project.
These new SST data sets have ancillary information important for research and climate studies. The impact of
the high-resolution blended global SSTs on operational weather prediction, hurricane intensity prediction, and
radiance assimilation is being evaluated. These new products and the impact studies will be discussed in
detail. This project will make a direct US contribution to the Global Ocean Data Assimilation Experiment (GODAE)
by working within the GODAE High-Resolution SST Pilot Project (GHRSST-PP), initiated by the international
GODAE steering team, to coordinate the production of a new high-resolution SST.
http:www.misst.org
OS33A-09
Tracing a River Plume Using Satellite Derived Surface Chlorophyll
Based on the analysis of satellite derived surface chlorophyll-a (CSAT) and concomitant in-situ surface salinity (S) data we show that CSAT estimated by the OC4v4 SeaWiFS retrieval algorithm is a good indicator of surface salinity over a continental shelf dominated by freshwater discharge. The CSAT 'S relation is illustrated for the case of Rio de la Plata, which discharges about 23,000 m3/s into the western South Atlantic near 35°S. During two high resolution surveys conducted in 2003 and 2004 CSAT and concomitant S observations present high correlations: 0.95 and 0.84, respectively. The log (CSAT) distribution over the shelf presents three modes, each associated to distinct water masses based on their S ranges. Across the offshore edge of the Plata plume the log (CSAT) 0.4 to 0.8 range is associated with a sharp surface salinity transition from 28.5 to 32.5. CSAT data from the period 1998-2006 are then used to characterize the space time variability of the Plata plume from seasonal to interannual time scales. The seasonal plume variations derived from the CSAT analysis are in excellent agreement with S variations based on historical hydrographic data. In austral winter CSAT maxima extend northeastward from the Plata estuary and along the southern Brazil shelf beyond 30°S, while in summer the high CSAT waters retreat to 32°S and extend south of the estuary to about 37.5°S, only exceeding this latitude during extraordinary events. Satellite derived winds and river outflow data are used to assess their effects on the plume spreading over the continental shelf. The seasonal CSAT variations northeast of the estuary are primarily controlled by reversals of the along-shore wind stress and surface currents. These CSAT variations over the shelf are opposite to what would be expected for phytoplankton blooms associated with wind induced coastal upwelling. At interannual time scales the Plata plume penetrations in winter range between 1200 and 650 km from the estuary, associated with changes in the intensity and persistency of the northeastward wind stress. Intense southwestward plume extensions, beyond 38°S, are dominated by interannual time scales and appear to be related to the magnitude of the river outflow. The plume response to the large river outflow fluctuations observed at interannual time scales is moderate, except offshore from the estuary mouth, where outflow variations lead CSAT variations by about two months.
OS33A-10
Changing Requirements for Archiving Climate Data Records Derived From Remotely Sensed Data
With the arrival of long term sets of measurements of remotely sensed data it becomes important to improve the standard practices associated with archival of information needed to allow creation of climate data records, CDRs, from individual sets of measurements. Several aspects of the production of CDRs suggest that there should be changes in standard best practices for archival. A fundamental requirement for understanding long- term trends in climate data is that changes with time shown by the data reflect changes in actual geophysical parameters rather than changes in the measurement system. Even well developed and validated data sets from remotely sensed measurements contain artifacts. If the nature of the measurement and the algorithm is consistent over time, these artifacts may have little impact on trends derived from the data. However data sets derived with different algorithms created with different assumptions are likely to introduce non-physical changes in trend data. Yet technology for making measurements and analyzing data improves with time and this must be accounted for. To do this for an ongoing long term data set based on multiple instruments it is important to understand exactly how the preceding data was produced. But we are reaching the point where the scientists and engineers that developed the initial measurements and algorithms are no longer available to explain and assist in adapting today's systems for use with future measurement systems. In an era where tens to hundreds of man years are involved in calibrating an instrument and producing and validating a set of geophysical measurements from the calibrated data we have long passed the time when it was reasonable to say "just give me the basic measurement and a bright graduate student and I can produce anything I need in a year." Examples of problems encountered and alternative solutions will be provided based on developing and reprocessing data sets from long term measurements of atmospheric, land surface and ocean measurements covering, in one case, a series of fifteen instruments currently scheduled to continue from 1978 through several decades into the 2030s. Possible changes in approach for developers of instruments and processing algorithms, archival centers, funding organizations and the climate science community will be suggested. There is a cost in both time and money associated with most of these changes and the hope is that this presentation will prompt further discussion on what should be done.