Sea surface effects on the sea surface temperature estimation by remote sensing—Part 2

By referring to the sea surface temperature profiler buoy (SSTPB) data observed in Mutsu Bay, this study showed under calm and strong sunshine conditions, the vertical water temperature profile near the sea surface bends abruptly, and sea surface temperature detected by satellite remote sensing is not necessarily coincident to the bulk sea surface temperature. Besides the atmospheric effect, this effect causes another error in the estimation of sea surface temperature by remote sensing data known as the sea surface effect (SSE). As a sequel to a former paper, this paper is concerned with the investigation of the conditions which occur in the apparent SSE. Statistical analyses were directed to the total data set of SSTPB. The amount of SSE was evaluated by the water temperature difference between the uppermost surface and 1 m depth, and apparent SSE was identified to when the absolute value of the difference is larger than 0·5 °C. Apparent SSE was observed in the season from May to September. Its occurrence rate in May and June was about 40 per cent of the total days, and about 20 per cent in July. SSE grew when wind speed was less than 2 m s^?1 and the solar zenith angle is smaller than 30°. If either of those two conditions were violated, SSE easily transferred into reducing phase.     

Near-surface thermal structure and surface diurnal warming in the Adriatic Sea using satellite and drifter data

An extensive set of in situ temperature data collected by surface drifters is combined with satellite-derived sea surface temperature images to study the difference between the pseudo-bulk and bulk temperatures (ΔT_pb-b) in the Adriatic Sea for the period 21 September 2002-31 December 2003. The variations of this temperature difference are described as a function of local wind speed and incoming solar radiation provided by a local area atmospheric model. The daily sea surface temperature variability is also assessed by computing the temperature difference between the daily maximal and minimal values (ΔT_day-night). The data show that the smaller the wind speed and the larger the solar radiation, the larger ΔT_pb-b. The temperature difference reached the highest value (∼5 °C) on a hot day (more than 600 W/m²) of May 2003 in weak wind condition (around 3 m/s). For strong winds (speed > 6 m/s) the dependence on both the wind and solar radiations vanishes as the temperature difference approaches zero because the near-surface water becomes thermally homogenous due to the wind-induced vertical mixing. Strong diurnal warming of the sea surface, as derived by the pseudo-bulk estimates, and a strong near-surface stratification were found during the spring/summer season. Monthly mean statistics show that the diurnal cycle of the pseudo-bulk and bulk temperature starts to become significant already in February and March. Subsequently (from April to August) both the diurnal warming and the stratification are maximal (monthly means of ΔT_day-night ∼1-2 °C and of ΔT_pb-b ∼0.5 °C ), while in fall and early winter the ΔT_pb-b values are quite small (monthly means near 0 °C) and the ΔT_day-night monthly means are bounded by 0.5-1.5 °C. Maximal amplitudes of the diurnal cycle can exceed 4 °C (mostly in spring-summer) for both the pseudo-bulk and bulk temperatures. However, the monthly means of ΔT_day-night is generally twice as large for the pseudo-bulk estimates (∼2 °C) with respect to the bulk layer (∼1 °C). The diurnal warming of the sea surface, as derived by the pseudo-bulk temperature, occurs at about 14:30 local time, that is more than 2 h after the maximal sun elevation and an hour earlier than the bulk temperature maximum at 20-40 cm depth.     

The effect of monomolecular surface films on the microwave brightness temperature of the sea surface

Abstract

Airborne microwave radiometer measurements at 1·43 and 2·65 GHz over a sea surface covered with a monomolecular oleyl alcohol surface film and over adjacent slick sea surfaces are presented. The measurements show that at 2·65 GHz the brightness temperature T _B is not affected by the slick, while at 1·43 GHz it drops from 93 K to a minimum value of almost O K. This implies that at 1·43 GHz the emissivity of the slick-covered sea surface is extremely small, similar to a metallic layer, and that this resonant-type phenomenon is confined to a narrow frequency band of width δ?/ ?<0·6.

The theoretical implications of these experimental findings are discussed in the framework of the Debye relaxation theory of polar liquids. It is conjectured that a thin layer of water molecules polarized by the surface film gives rise to an anomalous dispersion, which causes the large decrease in brightness temperature at 1·43 GHz.

The modulus of the relative dielectric constant ε? is estimated to be ≥ 5·2 × 10^?4 and the thickness of the emitting layer ≤1·9 × 10^?4 m for 1·43 GHz. Furthermore, the film-induced surface activation energy is calculated to be 9·18 × 10^?21 J. These values seem reasonable in the light of the theories on the physicochemical structure of surface layers.     

Infrared sensing of sea surface temperature from space

Error sources in infrared remote sensing of sea surface temperature are discussed, e.g., imperfect transmittance models, uncertain or unknown atmospheric pressure-temperature-humidity vertical profiles, temperature discontinuities at the air-sea interface, temperature differences between surface and bulk water, and neglect of surface emissivity and reflectance. Some of these are analyzed using a simplified version of the transmittance function of Prabhakara et al. (1974). The rms error in conventional sea surface temperature retrievals, in which computers are used to integrate the equation of radiative transfer over many atmospheric layers, has thus far been reduced to about ±1 K (Maul, 1980). This error is for optimum conditions, and seems irreducible. Unless the accuracy can be improved it seems impractical to spend so much effort on lengthy computer retrievals. Prabhakara et al. (1974) have devised a much simpler retrieval method using three infrared bands, which yields an rms error of ±1.1 K. A very simple method yielding ±1.0 K with two infrared bands is described here.     

Spatial and temporal variability of satellite-derived sea surface temperature in the Barents Sea

The Barents Sea (BS) is an important region for studying climate change. This sea is located on the main pathway of the heat transported from low to high latitudes. Since oceanic conditions in the BS may influence vast areas of the Arctic Ocean, it is important to continue to monitor this region and analyse the available oceanographic data sets. One of the important quantities that can be used to track climate change is the sea surface temperature (SST). In this study, we have analysed the 32 years, (1982–2013) National Oceanic and Atmospheric Administration (NOAA) Optimum Interpolation SST Version 2 data for the BS. Our results indicate that the regionally averaged SST trend in the BS (about 0.03°C year^–1) is greater than the global trend. This trend varies spatially with the lowest values north from 76° N and the highest values (about 0.06°C year^–1) in proximity of Svalbard and in coastal regions near the White Sea. The SST and 2 m air temperature (AT) trends are high in winter months in the open BS region located west from Novaya Zemlya. Such trends can be linked to a significant retreat of sea ice in this area in recent years. In this article, we also documented spatial patterns in the annual cycle of SST in the BS. We have shown that the interannual variability of SST is similar in different regions of the BS and well correlated with the interannual patterns in AT variability.     

The impact of spectral emissivity on the measurement of land surface temperature from a satellite

Abstract

The split-window method is successfully used to infer sea surface temperature from satellite radiances, principally because sea surface temperature is not very different from the air temperature near the surface and because the emissivity of the sea is constant over large areas and is not very different from one in the spectral channels of interest. This is not true for land surfaces and the split-window method has to be re-examined for such a case. This is the aim of this paper. In order to relate land surface temperature to the two brightness temperatures measured from space in the two channels of interest (namely, AVHRR 4 and AVHRR 5), several formulae are derived and their accuracies are discussed. Assuming that the emissivities ε₁ and ε₂ in the two channels considered, and therefore their average $ are unity, it is shown that the error ΔT generated on the land surface temperature by correcting atmospheric effects using the split-window method in most situations studied is of the order of

$

This error may be quite significant, except for the sea surface where it is shown to be negligible. In order to infer land surface temperature from space, it is therefore necessary to know the surface spectral emissivity to good accuracy. Possible methods to determine it are then proposed and discussed.     

Size estimates of the factors controlling sea surface temperature with AVHRR data

NOAA AVHRR thermal infrared images, meteorological measurements and model calculations have been used to estimate the magnitude of the processes controlling the sea surface temperature variability in the Bay of Biscay. The estimates are based on the equation of temperature conservation and the equation for the total derivative. The results show that local time change, advection, horizontal diffusion and vertical turbulent flux vary with time and space. The sea surface temperature variability is mainly controlled by vertical turbulent flux and advection. For an average of five-day estimates, the order of the relative importance of the processes is vertical turbulence (38 per cent), advection (32 per cent), local time change (22 per cent) and horizontal diffusion (8 per cent).     

Development of polarization ratio model for sea surface wind field retrieval from TerraSAR-X HH polarization data

In this article, the polarization ratio (PR) of TerraSAR-X (TS-X) vertical–vertical (VV) and horizontal–horizontal (HH) polarization data acquired over the ocean is investigated. Similar to the PR of C-band synthetic aperture radar (SAR), the PR of X-band SAR data also shows significant dependence on incidence angle. The normalized radar cross-section (NRCS) in VV polarization data is generally larger than that in HH polarization for incidence angles above 23°. Based on the analysis, two PR models proposed for C-band SAR were retuned using TS-X dual-polarization data. A new PR model, called X-PR hereafter, is proposed as well to convert the NRCS of TS-X in HH polarization to that in VV polarization. By using the developed geophysical model functions of XMOD1 and XMOD2 and the tuned PR models, the sea surface field is retrieved from the TS-X data in HH polarization. The comparisons with in situ buoy measurements show that the combination of XMOD2 and X-PR models yields a good retrieval with a root mean square error (RMSE) of 2.03 m s^–1 and scatter index (SI) of 22.4%. A further comparison with a high-resolution analysis wind model in the North Sea is also presented, which shows better agreement with RMSE of 1.76 m s^–1 and SI of 20.3%. We also find that the difference between the fitting of the X-PR model and the PR derived from TS-X dual-polarization data is close to a constant. By adding the constant to the X-PR model, the accuracy of HH polarization sea surface wind speed is further improved with the bias reduced by 0.3 m s^–1. A case acquired at the offshore wind farm in the East China Sea further demonstrates that the improvement tends to be more effective for incidence angles above 40°.     

In situ angular measurements of thermal infrared sea surface emissivity—Validation of models

In this paper, sea surface emissivity (SSE) measurements obtained from thermal infrared radiance data are presented. These measurements were carried out from a fixed oilrig under open sea conditions in the Mediterranean Sea during the WInd and Salinity Experiment 2000 (WISE 2000). The SSE retrieval methodology uses quasi-simultaneous measurements of the radiance coming from the sea surface and the downwelling sky radiance, in addition to the sea surface temperature (SST). The radiometric data were acquired by a CIMEL ELECTRONIQUE CE 312 radiometer, with four channels placed in the 8-14 μm region. The sea temperature was measured with high-precision thermal probes located on oceanographic buoys, which is not exactly equal to the required SST. A study of the skin effect during the radiometric measurements used in this work showed that a constant bulk-skin temperature difference of 0.05±0.06 K was present for wind speeds larger than 5 m/s. Our study is limited to these conditions. Thus, SST used as a reference for SSE retrieval was obtained as the temperature measured by the contact thermometers placed on the buoys at 20-cm depth minus this bulk-skin temperature difference.SSE was obtained under several observation angles and surface wind speed conditions, allowing us to study both the angular and the sea surface roughness dependence. Our results were compared with SSE models, showing the validity of the model of Masuda et al. [Masuda, K., Takashima, T., & Takayama, Y. (1988) Emissivity of pure seawaters for the model sea surface in the infrared window regions. Remote Sensing of Environment, 24, 313-329.] for observation angles up to 50°. For larger angles, the effect of double or multiple reflections on the sea surface produces discrepancies between measured and theoretical SSEs, and more complex models should be used to get accurate SSE values, such as the model of Wu and Smith [Wu, X., & Smith, W.L. (1997). Emissivity of rough sea surface for 8-13 μm: modelling and verification. Applied Optics, 36, 2609-2619.].     

Mining geophysical parameters through decision-tree analysis to determine correlation with tropical cyclone development

Correlations between geophysical parameters and tropical cyclones are essential in understanding and predicting the formation of tropical cyclones. Previous studies show that sea surface temperature and vertical wind shear significantly influence the formation and frequent changes of tropical cyclones. This paper presents the utilization of a new approach, data mining, to discover the collective contributions to tropical cyclones from sea surface temperature, atmospheric water vapor, vertical wind shear, and zonal stretching deformation. A decision tree using the C4.5 algorithm was generated to illustrate the influence of geophysical parameters on the formation of tropical cyclone in weighted correlations. From the decision tree, we also induced decision rules to reveal the quantitative regularities and co-effects of [sea surface temperature, vertical wind shear], [atmospheric water vapor, vertical wind shear], [sea surface temperature, atmospheric water vapor, zonal stretching deformation], [sea surface temperature, vertical wind shear, atmospheric water vapor, zonal stretching deformation], and other combinations to tropical cyclone formation. The research improved previous findings in (1) preparing more precise criteria for future tropical cyclone prediction, and (2) applying data mining algorithms in studying tropical cyclones.