Algorithm and validation of sea surface temperature observation using MODIS sensors aboard terra and aqua in the western North Pacific

A regional algorithm to estimate SST fields in the western North Pacific, where small oceanographic disturbance are often found, has been developed using Moderate Resolution Imaging Spectroradiometers (MODIS) aboard Terra and Aqua. Its associated algorithm, which includes cloud screening and SST estimation, is based on an algorithm for the Global Imager (GLI) aboard Advanced Earth Observing Satellite-II (ADEOS-II) and is tuned for MODIS sensors. For atmospheric correction, we compare Multi-Channel SST (MCSST), Nonlinear SST (NLSST), Water Vapor SST (WVSST) and Quadratic SST (QDSST) techniques. For NLSST, four first-guess SSTs are investigated, including the values for MCSST, climatology with two different spatial resolutions, and near-real-time objective analysis. The results show that the NLSST method using high-resolution climatological SST as a first-guess has both good quality and high efficiency. The differences of root-mean-square error (RMSE) between the NLSST models using low-resolution climatology and those using high-resolution climatology are up to 0.25 K. RMSEs of the new algorithm are 0.70 K/0.65 K for daytime (Aqua/Terra) and 0.65 K/0.66 K for nighttime, respectively. Diurnal warming and the stratification of the ocean surface layer under low wind are discussed.     

Hourly sea surface temperature retrieval using the Japanese Geostationary Satellite,Multi-Functional Transport Satellite (MTSAT)

An algorithm has been developed for retrieving sea surface temperature (SST) from hourly data transmitted from the Japanese Advanced Meteorological Imager (JAMI) aboard a Japanese geostationary satellite, Multi-functional Transport Satellite (MTSAT)-1R. Threshold tests screening cloudy pixels are empirically adjusted to cases of daytime with/without sun glitter, and nighttime. The Non-Linear SST (NLSST) equation, including several new additional terms, is used to calculate MTSAT SST. The estimated SST is compared with drifting and moored buoy measurements, with the result that the bias of the MTSAT SST is nearly 0.0°K. The root mean square (rms) error is about 0.8°K, and it is 0.7°K under the condition that the satellite zenith angle is less than 50°. It is demonstrated that the hourly MTSAT SST produced by the algorithm developed here captures diurnal SST variations in the equatorial sea in mid-November 2006.     

Difference characteristics of sea surface temperature observed by GLI and AMSR aboard ADEOS-II

This study compares infrared and microwave measurements of sea surface temperature (SST) obtained by a single satellite. The simultaneous observation from the Global Imager (GLI: infrared) and the Advanced Microwave Scanning Radiometer (AMSR: microwave) aboard the Advanced Earth Observing Satellite-II (ADEOS-II) provided an opportunity for the intercomparison. The GLI-and AMSR-derived SSTs from April to October 2003 are analyzed with other ancillary data including surface wind speed and water vapor retrieved by AMSR and SeaWinds on ADEOS-II. We found no measurable bias (defined as GLI minus AMSR), while the standard deviation of difference is less than 1°C. In low water vapor conditions, the GLI SST has a positive bias less than 0.2°C, and in high water vapor conditions, it has a negative (positive) bias during the daytime (nighttime). The low spatial resolution of AMSR is another factor underlying the geographical distribution of the differences. The cloud detection problem in the GLI algorithm also affects the difference. The large differences in high-latitude region during the nighttime might be due to the GLI cloud-detection algorithm. AMSR SST has a negative bias during the daytime with low wind speed (less than 7 ms⁻¹), which might be related to the correction for surface wind effects in the AMSR SST algorithm.     

A new set of MCSST equations for NOAA-9/AVHRR

A new set of multi-channel sea surface temperature (MCSST) equations for the Advanced Very High Resolution Radiometer (AVHRR) on NOAA-9 is derived from regression analyses between two-channel brightness temperatures andin situ SST obtained from moored buoys around Japan. Two equations are derived: one for daytime and the other for nighttime. They are linear split window type and both the equations contain a term dependent on satellite zenith angle, which has not been accounted for in the previous daytime split window equations for NOAA-9. It is shown that the new set of equation can give SSTs in much better precision than those without the zenith-angle-dependent terms. It is also found that the split window equation for NOAA-9 provided by the National Oceanographic and Atmospheric Administration/National Environmental Satellite, Data and Information Service (NOAA/NESDIS) considerably underestimates the daytime SSTs; sometimes nighttime SSTs are evenhigher than daytime SSTs. This is because the zenith angle effect to the radiation deficiet is neglected in the daytime equation by NOAA/NESDIS. By using the new MCSST equations, it is expected that the quality of satellite MCSST would be much improved, at least in regional applications around Japan, for the period of NOAA-9's operation.     

Retrieval of Sea Surface Temperature from TRMM VIRS

The Visible and Infrared Scanner (VIRS) aboard the Tropical Rainfall Measuring Mission (TRMM) is a five-channel radiometer with wavelength from 0.6 to 12 μm. Daily 0.125° sea surface temperature (SST) data from VIRS were first produced at the National Space Development Agency (NASDA) for comparison with SST from TRMM Microwave Imager (TMI). In order to obtain accurate high spatial resolution SST for the merging of SST from infrared and microwave measurements, new SST retrieval coefficients of the Multichannel SST (MCSST) algorithm were generated using the global matchups from VIRS brightness temperature (BT) and Global Telecommunications System (GTS) SST. Cloud detection was improved and striping noise was eliminated. One-year global VIRS level-1B data were reprocessed using the MCSST algorithm and the advanced cloud/noise treatments. The bias and standard deviation between VIRS split-window SST and in situ SST are 0.10°C and 0.63°C, and for triple-window SST, are 0.06°C and 0.48°C. The results indicate that the reprocessing algorithm is capable of retrieving high quality SST from VIRS data. This revised version was published online in July 2006 with corrections to the Cover Date.     

The sea surface temperature product algorithm of the Ocean Color and Temperature Scanner (OCTS) and its accuracy

The Ocean Color and Temperature Scanner (OCTS) aboard the Advanced Earth Observing Satellite (ADEOS) can observe ocean color and sea surface temperature (SST) simultaneously. This paper explains the algorithm for the OCTS SST product in the NASDA OCTS mission. In the development of the latest, third version (V3) algorithm, the OCTS match-up dataset plays an important role, especially when the coefficients required in the MCSST equation are derived and the equation form is adjusted. As a result of the validation using the OCTS match-up dataset, the algorithm has improved the root mean square (rms) error of the OCTS SST up to 0.698°C although some problems remain in the match-up dataset used in the present study.     

Accuracies of NOAA/NESDIS sea surface temperature estimation technique in the oceans around Japan

We have examined accuracies of nine nighttime National Oceanic and Atmospheric Administration/National Environmental Satellite, Data and Information Service (NOAA/NESDIS) equations for SST estimation using the Advanced Very High Resolution Radiometer (AVHRR)/NOAA-11 dataset produced by Sakaida and Kawamura (1992). Among the nine equations, the revised triple-window CPSST algorithm gives the smallest rms error, which is 0.38°C. The dual-window MCSST algorithm gives the largest rms error 0.56°C. Rms errors of the other algorithms are smaller than 0.5°C.     

Estimation of sea surface temperatures around Japan using the advanced very high resolution radiometer (AVHRR)/NOAA-11

The accuracy of sea surface temperatures (SSTs) derived from the Advanced Very High Resolution Radiometer (AVHRR)/NOAA-11 is examined by comparison with sea-truth SSTs obtained from ocean data buoys durings November 1988 through December 1989. We made a 122 point data set of buoy SSTs from the oceans around Japan and the corresponding brightness temperatures of channels 4 and 5 during cloud free periods. The satellite temperatures are corrected for atmospheric effects using the NOAA Multi-Channel SST (MCSST) and Cross Product SST (CPSST) algorithms. The two algorithms give similar results for our data set and result in biases of about –0.1°C with rms errors of about 0.6°C relative to buoy SSTs. It is found that MCSSTs and CPSSTs tend to be higher than SSTs from the buoy in the Japan Sea in summer. New coefficients for the MCSST equations suitable for our data set are determined and the resultant rms error is 0.49°C. If we eliminate the cluster of anomalous summer data in the Japan Sea, the rms error becomes 0.43°C.     

Validation of GLI and other satellite-derived sea surface temperatures using data from the Rottnest Island ferry, Western Australia

An accurate platinum resistance thermometer (PRT) has been installed on a commercial ferry that operates between Hillarys Marina, some 15 km north of Fremantle, and Rottnest Island off the Western Australian coast. The PRT is located in the engine intake system and provides continuous under-way measurements of the bulk sea surface temperature (SST) at a depth of 1 m. The “SeaFlyte” ferry makes the trip to Rottnest Island between 3 and 5 times daily and so a wealth of data is available for comparison with the SST derived using data from the GLI instrument on ADEOS-II. Analyses of the ferry and satellite data confirm the excellent quality of SST estimates from the GLI as well as four other satellite instruments—AVHRR on NOAA-16, AATSR on ENVISAT, and the MODIS instruments on TERRA and AQUA. All satellite instruments showed a comparison standard deviation of better than 0.6°C with GLI being better than 0.4°C. The number of ferry-satellite data coincidences used in this study demonstrates one of the advantages of installing measurement systems on commercial ships that operate regular passenger or freight services rather than infrequent deployments on research vessels. The analyses also demonstrate that satellite-derived SST estimates obtained under low surface wind conditions must be treated with care.     

Advanced sea surface temperature retrieval using the Japanese geostationary satellite,Himawari-6

Producing high-quality match-ups coupling the Japanese geostationary satellite, Himawari-6 (H6), and buoy SST observations, we have developed the new SST retrieval method. Kawamura et al. (2010) developed the previous version of SST product called MTSAT SST, which left several scientific/technical questions. For solving them, 6,711 algorithm tuning match-ups with precise navigation and 240,476 validation match-ups are generated for covering all seasons and wide ocean coverage. For discriminating the previous MTSAT SST, we call the new version of SST H6 SST. It is found that the SZA dependences of MTSAT SST algorithm are different from area to area of SZA > 40–50° N/S. The regionally different SZA dependences are treated by dividing the H6 disk coverage into five areas by the latitude lines of 40° N/S first and the longitude lines of 100° K and 180° K. Using the algorithm tuning match-ups, Nonlinear SST (NLSST) equations are derived for all of the five areas. Though the sun zenith angle dependency correction term is also examined, there is no significant regional difference. Therefore, this term is used in the H6 SST algorithm again. The new H6 SST equation is formed by the areal NLSST and the sun zenith angle dependency term for each area. The statistical evaluation of H6 SST using the validation match-ups show the small negative biases and the RMS errors of about 0.74° K for each area. For the full H6 disk, the bias is −0.1° K and the RMS error 0.74° K. The histogram of H6 SST minus the in situ SST for each area has a similar Gaussian shape with small negative skewness, and the monthly validation of H6 SST for each area is consistent with those for the whole period and the histograms     

Highers—The AVHRR-based higher spatial resolution sea surface temperature data set intended for studying the ocean south of Japan

For the study of the ocean south of Japan, the HIGHER spatial resolution Sea surface temperature (HIGHERS) data have been produced from the AVHRR/NOAA data. The grid size is 0.075 degree and is much smaller than that of the usual sea surface temperature (SST) analysis field, which is gridded at most in 1 degree. In order to calculate accurate SST, the split-window Multi-Channel SST method is applied to cloudless pixels, which are detected with a cloud detection algorithm. The accuracy of the HIGHERS data is checked in comparison with the in-situ observation data.     

Improved Estimates of Wide-Ranging Sea Surface Temperature from GMS S-VISSR Data

We have developed an algorithm to estimate the wide-ranging Sea Surface Temperature (SST) data from the GMS-5 (Geostationary Meteorological Satellite) S-VISSR (Stretched-Visible Infrared Spin Scan Radiometer). Better SST estimates are realized by averaging the temporal variation of the VISSR calibration table and decreasing noise of the split-window terms using a spatial filter. The effects of the satellite zenith angle were examined in detail for better estimates, and VISSR-derived SSTs with root-mean-square (rms) error of 0.8 K were achieved using a new algorithm. The accuracy of SST estimates has been improved by using the temporal-spatial average of the split-window terms. Using the new techniques, we demonstrate that the hourly wide-ranging SST image data can be used to study the daily variations of SSTs in the Northern and Southern Pacific Oceans.     

A-HIGHERS—The System to Produce the High Spatial Resolution Sea Surface Temperature Maps of the Western North Pacific Using the AVHRR/NOAA

To study on the oceanic variations in the western North Pacific, we developed a system to produce a high spatial resolution sea surface temperature (SST) map from the data obtained by the Advanced Very High Resolution Radiometer (AVHRR) aboard the National Oceanic and Atmospheric Administration (NOAA) satellites. As the system has been improved on the HIGHERS (Sakaida and Kawamura, 1996), it is called the Advanced-HIGHERS (A-HIGHERS). The A-HIGHERS has been developed on the super computer in the Tohoku University, which is favorable for handling of a large volume of data. Mainly because of improvements in the cloud detection algorithm, the A-HIGHERS can deal with the data obtained at dawn and dusk around the year, and at daytime in summer more effectively. The A-HIGHERS are used to produce SST maps spanning from (60°N, 120°E) to (20°N, 160°E) with a grid size of 0.01 degree.     

卫星遥感业务系统海表温度误差控制方法

提高卫星遥感海表温度的反演精度是各种反演模型追求的目标,也是遥感系统业务化应用的关键.据相关文献报道,在晴空无云的条件下遥感海表温度的精度达到了0.5℃,但考虑到影响海表温度反演精度的多种因素,在遥感业务系统真正实现SST精度在1℃以内是非常困难的.在北太平洋渔场速报制作系统中,对遥感海表温度与船测温度误差统计显示均方根误差达到5.71℃,匹配点误差分布显示存在大量较大的负误差值,最大的为-17.2℃,遥感温度图也反映出存在片状温度低值区,这些区域很可能被错误地当作冷涡或冷锋区,严重干扰渔情分析,这些异常的温度误差很难通过海表温度反演模式和云检测技术来消除.采用一种标准海表温度参考图用于温度误差控制技术,可有效地检测温度反演异常值,将均方根值从5.71℃降低到1.75℃,如果采用2℃阈值控制计算均方根值,则海表温度精度达到0.785℃.该方法基本消除了遥感海表温度的低值现象,明显提高了遥感海表温度的精度,并已成功地应用于北太平洋渔区的海况速报产品制作中.     

Diurnal sea surface temperature variation and its impact on the atmosphere and ocean: A review

The importance of the diurnal variability of sea surface temperature (SST) on air-sea interaction is now being increasingly recognized. This review synthesizes knowledge of the diurnal SST variation, mainly paying attention to its impact on the atmosphere or the ocean. Diurnal SST warming becomes evident when the surface wind is weak and insolation is strong. Recent observations using satellite data and advanced instruments have revealed that a large diurnal SST rise occurs over wide areas in a specific season, and in an extreme case the diurnal amplitude of SST exceeds 5 K. The large diurnal SST rise can lead to an increase in net surface heat flux from the ocean of 50–60 Wm⁻² in the daytime. The temporal mean of the increase exceeds 10 Wm⁻², which will be non-negligible for the atmosphere. A few numerical experiments have indicated that the diurnal SST variation can modify atmospheric properties over the Pacific warm pool or a coastal sea, but the processes underlying the modification have not yet been investigated in detail. Furthermore, it has been shown that the diurnal change of ocean mixing process near the surface must be considered correctly in order to reproduce SST variations on an intraseasonal scale in a numerical model. The variation of mixed-layer properties on the daily scale is nonlinearly related to the intraseasonal variability. The mixed-layer deepening/shoaling process on the daily scale will also be related to biological and material circulation processes.     

Estimating photosynthetically available radiation at the ocean surface from ADEOS-II global imager data

A simple, yet efficient and fairly accurate algorithm is presented to estimate photosynthetically available radiation (PAR) at the ocean surface from Global Imager (GLI) data. The algorithm utilizes plane-parallel radiation-transfer theory and separates the effects of the clear atmosphere and clouds, i.e., the planetary atmosphere is modeled as a clear atmosphere positioned above a cloud layer. PAR is computed as the difference between the incident 400–700 nm solar flux at the top of the atmosphere (known) and the solar flux reflected back to space by the atmosphere and surface (derived from GLI radiance), taking atmospheric absorption into account. Knowledge of pixel composition is not required, eliminating the need for cloud screening and arbitrary assumptions about sub-pixel cloudiness. For each GLI pixel, clear or cloudy, a daily PAR estimate is obtained. Diurnal changes in cloudiness are taken into account statistically, using a regional diurnal albedo climatology based on 5 years of Earth Radiation Budget Satellite (ERBS) data. The algorithm results are verified against other satellite estimates of PAR, the National Centers for Environmental Prediction (NCEP) reanalysis product, and in-situ measurements from fixed buoys. Agreement is generally good between GLI and Sea-viewing Wide Field-of-view Sensor (SeaWiFS) estimates, with root-mean-squared (rms) differences of 7.9 (22%), 4.6 (13%), and 2.7 (8%) Einstein/m²/day on daily, weekly, and monthly time scales, and a bias of only 0.8–0.9 (about 2%) Einstein/m²/day. The rms differences between GLI and Visible and Infrared Spin Scan Radiometer (VISSR) estimates and between GLI and NCEP estimates are smaller and larger, respectively, on monthly time scales, i.e., 3.0 (7%) and 5.0 (14%) Einstein/m²/day, and biases are 1.1 (2%) and −0.2 (−1%) Einstein/m²/day. The comparison with buoy data also shows good agreement, with rms inaccuracies of 10.2 (23%), 6.3 (14%), and 4.5 (10%) Einstein/m²/day on daily, weekly, and monthly time scales, and slightly higher GLI values by about 1.0 (2%) Einstein/m²/day. The good statistical performance makes the algorithm suitable for large-scale studies of aquatic photosynthesis.     

基于变分理论的AVHRR海表温度反演应用及效果评估

基于变分理论算法实现了METOP-A卫星AVHRR传感器探测数据的海洋表面温度变分反演,进行了连续1个月的海表温度反演试验,并分别从全球、分纬度带和天气系统活跃区域3个方面,将变分反演结果（VAR SST）与利用统计回归方法反演相同卫星得到的海表温度产品（GBL SST）、其他海温融合产品（OISST）及实际浮标观测数据等进行一系列评估。从全球评估指标看出,以OISST为参照,VAR SST要优于GBL SST;以浮标观测为参照,VAR SST略逊于GBL SST,而且VAR SST还改进了GBL SST随时间波动大的缺点;从分纬度带对比看出,在与OISST对比时,VAR SST在低纬度地区和北半球中纬度地区的质量要优于GBL SST,海温反演精度较高。研究还表明,由于变分方法考虑了大气状态的变化,能够更加有效订正卫星遥感过程中大气的削弱作用,从而反演出精度更高的海表温度,尤其在天气系统较为复杂的区域效果明显。     

In-situ diurnal sea surface temperature variations and near-surface thermal structure in the tropical hot event of the Indo-Pacific warm pool

Diurnal Sea Surface Temperature (SST) variations and the near-surface thermal structure of the tropical hot event (HE) have been investigated using advanced in-situ equatorial observations with hourly temporal resolution. The information on the HE area defined by the satellite cloud-free SSTs is used to sample the in-situ observations. The in-situ SSTs sampled for the HE conditions show that a maximum (minimum) SST has a histogram mode at 30.8°C (29.0°C), and frequently appears at 15:00 (07:00) local time. The amplitude of the diurnal SST variation (DSST) is defined by the difference between the maximum and minimum SSTs. The mean DSST during HEs is greater than 0.5°C, and has a maximum of about 0.75°C at the HE peak. The time series of mean DSST gradually increases (rapidly decreases) before (after) the peak. The satellite SST has a systematic positive bias against the corresponding daytime SST measured by the Triangle Trans-Ocean buoy Network. This bias is enhanced under conditions of large in-situ DSST. One-dimensional numerical model simulation suggests that the systematic bias is caused by the sharp vertical temperature gradient in the surface layer of HE. The near-surface thermal structure is generated by conditions of high insolation and low wind speed, which is the typical HE condition.     

Validity of sea surface temperature observed with the TRITON buoy under diurnal heating conditions

In order to investigate the validity of buoy-observed sea surface temperature (SST), we installed special instruments to measure near-surface ocean temperature on the TRITON buoy moored at 2.07°N, 138.06°E from 2 to 13 March 2004, in addition to a standard buoy sensor for the regular SST measurement at 1.5-m depth. Large diurnal SST variations were observed during this period, and the variations of the temperatures at about 0.3-m depth could be approximately simulated by a one-dimensional numerical model. However, there was a notable discrepancy between the buoy-observed 1.5-m-depth SST (SST_1.5m) and the corresponding model-simulated temperature only during the daytime when the diurnal rise was large. The evaluation of the heat balance in the sea surface layer showed that the diurnal rise of the SST_1.5m in these cases could not be accounted for by solar heating alone. We examined the depth of the SST_1.5m sensor and the near-surface temperature observed from a ship near the buoy, and came to the conclusion that the solar heating of the buoy hull and/or a disturbance in the temperature field around the buoy hull would contribute to the excessive diurnal rise of the SST_1.5m observed with the TRITON buoy. However, the temperature around the hull was not sufficiently homogenized, as suggested in a previous paper. For the diurnal rise of the SST_1.5m exceeding 0.5 K, the daytime buoy data became doubtful, through dynamics that remain to be clarified. A simple formula is proposed to correct the unexpected diurnal amplitude of the buoy SST_1.5m.     

环渤海沿岸海表温度资料的均一性检验与订正

本文对环渤海沿岸具有代表性且资料完整的6个海洋观测站的月均海表温度(SST)序列作均一性检验和订正。我国海洋观测站密集度低,难以选择参考序列,本文首先采用不依赖参考序列的惩罚最大F检验(PMFT)方法对SST序列检验,利用详尽的元数据对检验结果进行确认,再对不连续点订正,该方法适用于元数据详尽的海洋观测站。对于元数据不详尽的观测站,使用惩罚最大T检验(PMT)方法,选取与海洋台站距离近且相关显著的气象观测站的均一化地面气温序列来制作参考序列,对SST序列进行检验和订正。结果表明,环渤海地区SST序列都存在一定非均一性,观测站较大距离迁移和观测系统变更(从人工观测到自动化观测)是造成非均一性的重要原因。订正后的环渤海地区年平均SST增温趋势更加明显。本文使用不同方法来检验SST序列的均一性,该思路对沿海其他海区观测站SST均一性检验和订正有一定参考价值和应用前景,可为沿海气候变化研究提供科学准确的第一手资料。