Surface Roughness Correction and Rain Effects
The effect of rain-induced stratification and surface roughness on Aquarius SSS retrieval
Contact author: Wenqing Tang, <email@example.com>
Wenqing Tang, Jet Propulsion Laboratory
Simon Yueh, Jet Propulsion Laboratory
Alex Fore, Jet Propulsion Laboratory
Akiko Hayashi, Jet Propulsion Laboratory
The objective of this study is to develop a rain roughness correction to reduce the uncertainty of Aquarius SSS retrieved under rainy conditions. The calibration reference chosen by the Aquarius project calibration/validation team is the SSS from HYCOM (SSSHYCOM), with Geophysical Model Function (GMF) built using rain-free data. Under rainy conditions, surface salinity stratification associated with rain freshwater inputs may cause large discrepancy in salinity measured by Aquarius at 1-2 cm (the penetration depth of L-band radiometer) and SSSHYCOM a few meters below the surface. The sensitivity of model function to rain accumulation is tested by comparing radiometer GMF built using data under currently rain free condition (within 1 hour) with those built using only data free of rain in the past 24 hours, with rain history extracted from NOAA CMORPH data at 30-minute intervals. A method is proposed to use the rain impact model (RIM) [Asher et al., 2014; Santos-Garcia et al., 2014] to adjust SSSHYCOM to reflect near surface stratification caused by freshwater inputs accumulated from rain events occurred over the past 24 hours prior Aquarius overpassing. It is hypothesized that when calibrated with RIM adjusted SSSHYCOM (named SSSRIMHYCOM), the residuals, i.e. the difference between measured and model predicted TB under rainy conditions, represent the rain-induced roughness. The JET PROPULSION LABORATORY Combined Active Passive (CAP) algorithm is used to retrieve SSS in parallel with (SSSCAP_RC) or without (SSSCAP) rain roughness correction. ∆SSS (SSSCAP_RC-SSSCAP), which highly correlate with rain pattern as expected, reaches about 0.2 PSU in wet areas such as ITCZ. Validation is performed with salinity measured at 50 cm by drifters (S50cm) and at 1m and 10m by mooring buoys. The difference between satellite retrieval (both SSSCAP and SSSCAP_RC) and in situ salinity measurements increases with rain rate, as well as the difference in depth. With rain-induced roughness accounted for, SSSCAP_RC likely represents the surface stratification effect in terms of the negative slope of SSSCAP_RC-Sin_situ(depth) vs. rain rate, which decreases with depth (towards the surface).
RAIN IMPACT MODEL (RIM) FOR AQUARIUS
Contact author: Maria Jacob, <firstname.lastname@example.org>
Maria Jacob, CONAE
Andrea Santos-Garcia, CFRSL
Maria Jacob, CONAE
Linwood Jones, CFRSL
William Asher, University of Washington Applied Physics Laboratory
This paper presents results of a recent empirical investigation into the impact of rain on the Aquarius (AQ) Sea Surface Salinity (SSS) measurements. Results demonstrate that AQ SSS measurements are realistic characterizations of a transient dilution of the surface salinity, but they are NOT representative of the bulk salinity at 5 m depth given by HYCOM. We believe that, during recent rain events, careful interpretation of AQ Level-2 (L-2) data is required, and as a result, the Rain Impact Model (RIM) product has been developed. It is available to AQ science users to promote the understanding of the relationship between precipitation and the corresponding AQ SSS measurement. This paper presents the description of RIM and comparisons between RIM and AQ L-2 SSS are presented for a number of rain events along the Pacific ITCZ (Inter-tropical Convergence Zone). Results demonstrate high correlation between RIM and AQ SSS for moderate to strong rain events that occurred within a few hours of the AQ observation time.
The RIM model is based on the temporal superposition of rain events (integrated rain history for the last 24 hours) using a one-dimensional stratification model and HYCOM as initialization. It estimates SSS in a quarter degree spatial resolution and integrates over the AQ IFOV (100 km) using a weighted average based on the antenna beam efficiency. Thus, the RIM predicts the modeled surface salinity that can be compared to the observed SSS as an overly to the AQ L-2 data product. In addition, the RIM provides the corresponding rain beam-fill fraction and the probability of salinity stratification. This latter parameter can be used as a “rain impact” quality flag to identify SSS that are affected by near surface stratification.
Assessment of rain impact on the Aquarius salinity retrievals
Contact author: Thomas Meissner, <email@example.com>
Thomas Meissner, Remote Sensing Systems
Frank Wentz, Remote Sensing Systems
Joel Scott, Remote Sensing Systems
Kyle Hilburn, Remote Sensing Systems
Fresh biases up to 0.4 psu are observed when comparing Aquarius salinities with ARGO or HYCOM in the tropical and subtropical regions. These biases are partly due to inaccuracies in the geophysical model function that correlate with sea surface temperature and partly due to salinity stratification in the upper ocean layer. Aquarius measures salinity within a few centimeters of the surface, which is fresher during and after rain effects than the 5 m depth layer, to which ARGO and HYCOM refer to. In order to make improvements to the salinity retrieval algorithm and avoid overcorrecting it is essential to separate these two effects. Our presentation shows how to do that and addresses the following issues:
1. We have collocated rain rate measurements from various sources with Aquarius: the CONAE MWR, SSMIS, WindSat, TMI, GMI, and the CMORPH data set. That allows us not only to filter out rain at the Aquarius observation but also quantify the salinity stratification in the upper ocean layer as function of rain accumulation. Salinity measurements from the PMEL moored buoy array at 1m depth help further to assess the stratification effect.
2. We find that the rain freshening amounts to only a third of the observed fresh biases in the tropics and thus conclude that the bulk of the freshening is spurious and due to inaccuracies in the geophysical model function. We have derived a mitigation for these biases which is to be implemented in the upcoming V4.0 release.
3. We have studied the effect of rain splashing on the ocean surface. It is possible to find combinations between the Aquarius V-pol and H-pol channels that are insensitive to changes in the surface salinity but sensitive to surface roughness. Analyzing these brightness temperature differences shows little to no sensitivity to rain at the surface if scatterometer derived wind speed is used in the surface roughness correction. We conclude that the scatterometer wind speed itself is an appropriate proxy for surface roughness including rain splash effects and no splash correction should be done in addition.
4. We have also studied the impact of rain and stratification on the sensor calibration. The calibration of Aquarius V3.0 is based on matching the Aquarius with the HYCOM reference globally. No rain filtering is done for that matching and the argument has been made that, as a consequence, the global Aquarius salinity average is too high. We have computed the impact of rain filtering on the calibration and found that filtering rain changes the global average by only about 0.03 psu and thus the rain freshening in the tropics has very little impact on the sensor calibration.
Observations and Modeling of Rain-Induced Near Surface Salinity Anomalies
Contact author: William Asher, <firstname.lastname@example.org>
William Asher, University of Washington Applied Physics Laborabory
Andrew Jessup, University of Washington Applied Physics Laborabory
Kyla Drushka, University of Washington Applied Physics Laborabory
Precipitation on the ocean forms salinity and temperature gradients in the top few meters of the ocean surface. If present, these gradients will complicate comparing salinity measured by ARGO drifters at typical depths of five meters to salinities retrieved using L-band microwave radiometers such as Aquarius and SMOS at depths on order of 0.01 m. Therefore, understanding the spatial scales and temporal persistence of these gradients, and the conditions under which they form, will be important in calibrating satellite measurements of sea surface salinity. A towed, surface-following profiler was deployed from the R/V Kilo Moana in December of 2011 in the central tropical Pacific Ocean. The profiler measured temperature and conductivity at depths of 10 cm, 20 cm, and 100 cm, and 200 cm. Profiles of temperature and conductivity in the presence of rain show that rain both freshens and cools the ocean surface. This provides opposite forcing for the density gradient at the surface, and depending on the rain rate the density profile can be either stable (freshening dominated) or unstable (cooling dominated). The dynamics of this are investigated using a one-dimensional model of ocean surface turbulence.
A MWR Ocean Roughness Correction Algorithm for the Aquarius SSS Retrieval
Contact author: Linwood Jones, <email@example.com>
Linwood Jones, University of Central Florida (UCF)
Yazan Hejazin, UCF
Linwood Jones, UCF
The AQ retrieval of sea surface salinity (SSS) is based upon the smooth ocean surface brightness temperature (Tb). Since AQ measurements are at the top of the ionosphere, there are many corrections applied to achieve the desired quantity. Of the various corrections, the uncertainty in ocean Tb due to surface roughness (caused by surface wind speed) is the worst. The AQ baseline approach to provide the roughness correction using the AQ scatterometer ocean radar backscatter to infer excess ocean emissivity. This poster presents a second approach, which is derived from independent coincident Tb measurements from the CONAE MicroWave Radiometer (MWR).
In this poster, a MWR derived sea surface roughness correction algorithm is presented that uses a new semi-empirical microwave Radiative Transfer Model (RTM) to estimate the excess ocean emissivity using ancillary data such as sea surface temperature (SST) and ocean surface wind vector. This RTM has been tuned using 2-years of observed AQ and MWR Tb’s and corresponding atmospheric and oceanic environmental conditions from numerical weather models (GDAS). The ocean roughness correction algorithm uses this RTM and collocated MWR Ka-band Tb’s and available ancillary data (e.g., sea surface temperature, surface wind vector, and HYCOM SSS); and the outputs are the corresponding roughness corrections for each AQ footprint.
Results of independent comparisons (not used in the RTM tuning process) are presented between the baseline AQ scatterometer derived ocean roughness correction and the MWR roughness correction algorithm. Also SSS retrievals using these two independent approaches will be compared to the Hybrid Coordinate Ocean Model (HYCOM) salinity and collocated AQ Validation Data System (AVDS) buoy SSS measurements. Results suggest advantages of combining both roughness corrections (AQ SCAT and MWR) to achieve improved SSS retrievals.