Open Access
Add to BibSonomy
Abstract
The adjacency radiance field surrounding a small island (i.e., the Lampedusa Island in the Central Mediterranean Sea) was theoretically analyzed to address implications on a hypothetical nearby system vicarious calibration (SVC) infrastructure for satellite ocean color sensors. Simulations, performed in the visible and near-infrared regions for the Ocean Land Color Instrument (OLCI) operated onboard Sentinel-3 satellites, show different patterns of adjacency effects (AE) around the island. In the direction of the reflected sunbeam (i.e., in the north-western region), AE mainly originate by missing glint contributions from the sea surface masked by the island. These AE are mainly negative, decrease with wavelength, and strongly depend on sea surface anisotropy (i.e., sea state) and illumination conditions; this hinders the capability to provide a general unique description of their features. In the remaining marine regions, AE are positive and do not exceed the radiometric sensitivity of OLCI data beyond approximately 14 km from the coast. At shorter distances, uncertainties in satellite radiance due to AE would hence not allow fulfilling requirements for SVC.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Corrections
Barbara Bulgarelli and Giuseppe Zibordi, "Adjacency radiance around a small island: implications for system vicarious calibrations: publisher’s note," Appl. Opt. 59, 2592-2592 (2020)https://opg.optica.org/ao/abstract.cfm?uri=ao-59-8-2592
21 February 2020: A correction was made to Eq. (1).
1. INTRODUCTION
Ocean color (OC) is one of the Essential Climate Variables (ECV) indicated by the Global Climate Observation System (GCOS) for the characterization of the Earth’s climate. Products derived from OC remote sensing include the water-leaving radiance ${L_w}$, which quantifies the light emerging from the sea, and the near-surface chlorophyll concentration Chl used as a proxy for phytoplankton biomass. Both products are relevant to evaluate ocean ecosystems health and productivity, to assess the role of the oceans in the global carbon cycle, to manage living marine resources, and to quantify the impact of climate variability and change. While ${L_w}$ is the primary OC product, i.e., directly retrieved from the radiometric signal ${L_{\rm tot}}$ at the satellite sensor, Chl is a derived product resulting from the application of bio-optical algorithms to ${L_w}$. The accuracy of ${L_w}$ hence determines the accuracy of Chl.
Time-series of satellite OC data aiming at supporting climate change investigations must exhibit high accuracy to allow for a confident detection of long-term trends embedded in large natural variations [1]. Specific requirements for OC climate data records of satellite-derived ${L_w}$ are [2]
- • a radiometric uncertainty lower than 5% (with reference to oceanic waters in the blue-green spectral range), and
- • a radiometric stability better than 0.5% per decade.
These accuracy requirements are largely limited by uncertainties affecting the calibration of the satellite sensor and the atmospheric correction process. The latter identifies the procedure leading to the determination of ${L_w}$ from ${L_{\rm tot}}$ through the quantification and removal of all other radiance contributions that originate from sunlight interaction with atmosphere, sea surface, sea-bottom, any nearby land, or clouds.
The impact of uncertainties affecting the sensor absolute calibration and the removal of atmospheric effects is minimized through the so-called System Vicarious Calibration(SVC). This provides mission specific adjustment factors (i.e., g-factors) for pre-launch calibration coefficients, determined from the ratio of predicted to measured ${L_{\rm tot}}$. SVC implies the use of the same atmospheric models and algorithms applied in the operational atmospheric correction process.
In the specific case of the Sentinel-3 Ocean and Land Color Instrument (OLCI), whose baseline atmospheric correction algorithm makes use of spectral bands in the near-infrared (NIR) to estimate the aerosol reflectance [3], the SVC procedure develops in two steps. The first allows adjustment of the relative spectral response of the NIR bands for an accurate retrieval of the aerosol radiance. The second step applies these relatively calibrated NIR bands to retrieve the aerosol contribution in the visible bands and hence determine the remaining adjustment factors [4].
The vicarious calibration of the NIR bands is usually performed in remote, clear open ocean waters (e.g., the South Pacific gyre) where the aerosol type is known [4].
The vicarious calibration of the visible bands instead requires a site where highly accurate in situ reference radiometric data are acquired through state-of-the-art measurement technologies, data reduction methods, and quality assurance/control schemes [5].
An evaluation of SVC requirements for OC climate change applications [5] suggests that an ideal SVC site for the indirect calibration of OC visible bands should allow
- • maximizing the number of high-quality matchups, by trading off factors such as best viewing geometry, sun-glint avoidance, low cloudiness, negligible continental, and nearby land contaminations;
- • minimizing the relative uncertainties in computed $ g $-factors, by exhibiting well-known or accurately modeled optical properties, such as those typical of a maritime atmosphere and oligotrophic/mesotrophic waters (which represent the majority of world oceans); and
- • optimizing the precision of computed $ g $-factors, by displaying high spatial homogeneity and small environmental variability of both atmosphere and ocean.
On the basis of the above requirements, potential European sites for the vicarious calibration of visible radiometric data acquired by the OLCI were identified [6]. Among these, the Sicily Channel embracing the Lampedusa Island (35.52°N, 12.57°E) in the Central Mediterranean Sea is here considered.
The small Lampedusa Island (approximately 12 km long and, on average, 2.5 km wide) hosts a station for climate observations managed by the Italian National Agency for New Technologies, Energy and Sustainable Economic Development (ENEA). The station includes (i) a ground-based laboratory (35.52°N, 12.63°E) dedicated to the investigation of changes in atmospheric composition and structure, and their effects on the surface radiation, and (ii) an oceanic buoy (35.49°N, 12.47°E) located at about 3.3 nautical miles southwest of the island that supports investigations on air–sea interactions and from 2019 also provides OC validation data.
The region is characterized by oligotrophic waters, while the local bathymetry exhibits shallow depths south of the island (i.e., 74 m at the buoy location) and deeper depths north of the island (i.e., 300 m at approximately 10 nautical miles from the coast).
Although maritime aerosol conditions are only seldom met in the region due to frequent occurrence of aerosol dust from the African continent and continental aerosol from Central-Eastern Europe [7,8], recent investigations indicated a sufficient number of high-quality matchups applicable for SVC [6]. Specifically, an analysis based on 5-year SeaWiFS data indicates two potential high-quality matchups per year when very stringent selection criteria are applied ($Chl \le {0}.{1}\;\unicode{x00B5} {{\rm gl}^{ - 1}}$, aerosol optical thickness at 865 nm ${\tau _a}({865}) \le {0}.{10}$, and Ångström exponent $ \alpha \le {1}.{0} $). When relaxing the above thresholds (i.e., by accepting $Chl \le {0}.{2}\;\unicode{x00B5} {{\rm gl}^{ - 1}}$ and ${\tau _a}({865}) \le {0}.{15}$) the number of potential quality matchups significantly increases to 27 per year.
By addressing the case of a hypothetical SVC infrastructure operated nearby the Lampedusa Island, the present study investigates potential perturbations in OLCI marine data originating from the radiance reflected by the island. Standard atmospheric correction methods applied to satellite OC data, indeed, assume an infinite and homogeneous water surface and do not account for radiance contributions from any nearby land, which become a source of uncertainties (the so-called adjacency effects, AE).
The adjacency radiance field around the island is here theoretically investigated by adopting the same procedure applied in previous studies to quantify AE at a specific OC validation site [9] and for typical mid-latitude coastal environments [10].
Simulations are performed at representative visible and NIR center-wavelengths and for typical observation geometries, accounting for annual average atmospheric, land, sea, and illumination conditions. Acknowledging the seasonality of AE [11], simulations have been also performed for the months of January and August, assumed as representative of winter and summer observation conditions, respectively.
2. SIMULATION PROCEDURE
A. Simulation of the Adjacency Contributions at the Sensor
AE are here quantified through the adjacency radiance ${L_{\rm adj}}$, defined as the difference between the actual background radiance at the sensor and the background radiance that would arise from a spatially homogeneous surface exhibiting the same reflecting properties of the sea target. As such, ${L_{\rm adj}}$ can range from negative to positive values.
Simulations of ${L_{\rm adj}}$ are performed at $\lambda = {490}$, 555, 670, and 865 nm by applying the methodology outlined in [9,12] (to which the reader is addressed for more details). Specifically,
The functions ${C^{\rho = 1}}$ and $W$ depend on the illumination and observation geometry, on the land/sea spatial extension, as well as on the atmospheric optical properties. Their simulation, requiring a three-dimensional (3D) description of the propagating system, has been performed with the Novel Adjacency Perturbation Simulator for Coastal Areas (NAUSICAA) 3D backward Monte Carlo (MC) code [9].
The NAUSICAA MC code accounts for multiple scattering, off-nadir illumination and observation geometries, sea surface roughness, actual coastal morphology, and its precision is set to meet actual OC sensor radiometric resolutions [9].
The input parameters ${\rho _l}$ and ${R_{\rm rs}}$ can be extrapolated from satellite-derived or in situ measured data. Hence, once ${C^{\rho = 1}}$ and $W$ are computed for given geometric and atmospheric inputs, the proposed modeling [Eq. (1)] allows a fast evaluation of the adjacency effects for a wide variety of land and water spectral signatures.
The modeling of the propagating system, the selected geometries of illumination and observations, as well as the optical features of atmosphere, water, and land are described in Sections 2.B, 2.C, 2.D, 2.E, and 2.F, respectively.
B. System Modeling
The modeling of the propagating system relies on a number of elements detailed hereafter.
The atmosphere is divided into 14 plane-parallel layers resolving the vertical distribution of aerosol, ozone, and other gas molecules. Each atmospheric layer contains a variable mixture of ozone (only absorbing), other gas molecules (only scattering), and aerosol (scattering and absorbing) modeled in agreement with [9], besides for the aerosol scattering phase function, approximated by a spectral one-term Henyey–Greenstein (OTHG) analytical function.
The surface of non-uniform reflecting properties is centered on the island and divided in ${200} \times {200}$ square elements with a dimension of 232 m (see Fig. 1). The land/sea mask is extracted from the UMD Global 250 m Land Water Mask (MOD44W MODIS product) [14]. A wind-roughened sea surface in the absence of whitecaps is accounted for in the simulations. The latter assumption is reasonable for wind speeds lower than about ${7}\;{{\rm ms}^{ - 1}}$ [15].
Fig. 1. Land/sea mask utilized in the NAUSICAA MC simulations: land elements are indicated in dark gray, while sea elements are in light gray. Each of the ${200} \times {200}$ square elements is 232 m wide. The black lines represent the selected transects, while the black dot indicates the ENEA oceanographic buoy (35.49°N, 12.47°E).
Typical adjacency contributions at the sensor are computed over the whole marine region surrounding the island. Detailed simulations are provided along three study transects: T1 extending for about 13 km from the western tip of the island and intercepting the location of the oceanographic buoy; T2 and T3 extending for about 12 km southward and 17 km northward from the middle part of the island, respectively.
C. Geometrical Observation and Illumination Conditions
Actual observation and illumination conditions encountered by the OLCI in the area of interest have been considered (see Table 1). Solar and sensor zenith angles are determined with respect to the local vertical, while solar and sensor azimuth angles are counted clockwise from the north direction (as generally adopted in satellite geolocation). In accordance with the geometry of observation of the OLCI sensor, which is shifted across track by 12.6° westward away from the sun, the space sensor viewing angle ${\theta _v}$ is chosen to vary from 5° to 50° for the eastern observations (i.e., when the sensor azimuth ${\phi _v}$ equals 99°), and from 5° to 20° for the western observations (i.e., when ${\phi _v} = - {{81}^\circ}$). Average solar zenith and azimuth angles ${\theta _0}$ and ${\phi _0}$ at 10:00 Mean Solar Local time (corresponding to the Equatorial Crossing Time of the OLCI) equal 48° and 130°, respectively. For the August and January observations, (${\theta _0},{\phi _0}$) is set to (35°,118°) and (65°,144°), respectively.
Table 1. Geometric Parameters Defining the Illumination and Observation Geometries
D. Atmospheric Features
Continental and anthropogenic particles originating from Europe, desert dust from Africa, as well as marine aerosols from the North Atlantic and the Mediterranean itself, are commonly present in the basin of the Lampedusa Island [7,8]. The analysis of aerosol observations at the Lampedusa Aerosol Robotic Network (AERONET) site during the period 2004–2007 [16] indicates 36% of cases dominated by desert dust originating from Africa (characterized by ${\tau _a}({500}) \le {0.15}$ and $\alpha \ge {0.5}$), 6% of cases dominated by urban-industrial/biomass-burning aerosol originating from Central-Eastern Europe (characterized by ${\tau _a}({500}) \ge {0.1}$ and $\alpha \ge {1.5}$), and 68% of cases presenting mixed conditions, including mixture of various aerosol types and pure marine aerosol. Notably, pure marine conditions only occur when the influence from the European and African continents is very limited, and thus are rarely observed.
Mixed conditions, representative of average atmospheric conditions, are characterized by a single scattering albedo ${\omega _0}\sim{0.8}$ and an asymmetry parameter of the OTHG aerosol phase function ${g_a}\sim{0.7}$ at all wavelengths [16].
Climatology Level 2.0 quality assured data collected at the Lampedusa AERONET site over the period 2000–2017 indicate a yearly average value of ${\tau _a}({870}) = {0.13}\;{\rm \pm }\;{0.04}$ (corresponding to ${\tau _a}({500}) = {0.18}\;{\rm \pm }\;{0.06}$), and a yearly average value of $\alpha = {0}.{88}\;{\rm \pm }\;{0}.{52}$ (as determined in the 440–870 nm spectral range). Intra-annual values of ${\tau _a}({870})$ and $\alpha $ are illustrated in Fig. 2: they both increase in summer and decrease in winter.
Fig. 2. Intra-annual values of climatological ${\tau _a}({870})$ and $\alpha $ at the Lampedusa AERONET site. Error bars indicate standard deviations.
It is remarked that, on average, the values of $\alpha $ are lower than the threshold of 1.0 applied in an already mentioned study for the selection of high-quality matchups for ocean color SVC [6], but they approach or even exceed such a threshold in summer. For what concerns the aerosol optical thickness, the average value of ${\tau _a}({870})$ slightly exceeds the stringent threshold of 0.10, but it is still lower than the more relaxed threshold of 0.15 [6]. Winter ${\tau _a}({870})$ values are generally lower than 0.10, while they tend to slightly exceed 0.15 in summer. This might suggest that the local summer atmospheric conditions are less favorable for ocean color SVC.
Atmospheric data applied in the simulations are summarized in Table 2 where $\beta $ indicates the Ångström coefficient. Simulations are performed for yearly average conditions, and additionally for the mean atmospheric conditions encountered in August and January. The mid-season conditions are expected to be represented by the yearly mean ones. A direction-independent wind with a mean speed of ${3.3}\;{{\rm ms}^{ - 1}}$, very close to the conditions encountered during AERONET measurements, has been selected.
Table 2. Atmospheric Parameters Used for the Simulations
E. Water Radiometric Features
Water radiometric features have been extracted from a SeaWiFS multi-year climatological analysis of radiometric products [17]. Annual and intra-annual values of the remote sensing reflectance ${R_{\rm rs}}$ illustrated in Fig. 3, indicate a seasonal variation at the blue wavelengths, with lower values in winter and higher values in summer.
Fig. 3. Average annual and intra-annual spectral values of ${R_{\rm rs}}$ applied in the simulations. Error bars indicate standard deviations.
F. Selected Land Reflectance Properties
The directional-hemispherical reflectance (DHR, i.e., the reflectance for incoming light from a single direction [13]) and the isotropic bihemispherical reflectance (${{\rm BHR}_{\rm iso}}$, i.e., the reflectance for an incoming isotropic light field [18]), reported as black-sky and white-sky albedos in the MODIS product suite, have been extracted from the MODIS multi-year climatological snow-free aggregate database [19] at the MODIS land center-wavelengths $\lambda = {470}$, 555, 659, 858 nm and for each MODIS land pixel. These data are the sole providing quality climatological albedo products at suitable spectral and spatial resolution.
It is specified that DHR is the reflectance that would be measured if the atmospheric scattering effects were removed, thus leading to a purely monodirectional incident radiance field. Conversely, ${{\rm BHR}_{\rm iso}}$ is the reflectance that would be measured if the incident radiance field could be assumed isotropic [18]. Both products are independent of the environmental conditions.
In accordance with the applied methodology [9], DHR and ${{\rm BHR}_{\rm iso}}$ data have been spatially averaged over the considered land area, and annual and intra-annual values have been computed (Fig. 4). Differences between the spectral values of time and spatially averaged DHR and ${{\rm BHR}_{\rm iso}}$ (indicated as $\langle{\rm BHR}\rangle$ and $\langle{{\rm BHR}_{\rm iso}}\rangle$) do not appear significant. Nonetheless, the “actual” average land ${\rm BHR}$, hereafter simply termed land albedo ${\rho _l}$, has been computed for each annual and intra-annual period and at each wavelength by weighting the two distinct surface reflectance products through the ratio ${{\cal S}_E}$ between diffuse and direct irradiance at the surface [18]:
Land albedos at the OC wavelengths $\lambda = {490}$, 555, 670, and 865 nm have been taken equal to those of the closest MODIS land center-wavelengths.
Fig. 4. Time and spatially averaged spectral values of climatological DHR and ${{\rm BHR}_{\rm iso}}$ applied in the simulations. Error bars, generally not exceeding the size of the symbols, indicate standard deviations.
Lampedusa is a semi-arid island, whose vegetation typically consists of evergreen spaced shrubs. This explains the low intra-annual variation in land reflectance, and its slight increase (decrease) in winter (summer) in the NIR.
G. Simulation of the Atmospheric Optical Quantities
For the same set of test cases, the plane-parallel finite element method (FEM) numerical algorithm [20,21] has been used to simulate atmospheric optical quantities such as (i) the diffuse transmittance $t$ [22]; (ii) the path radiance ${L_{\rm path}}$ (often indicated as ${L_{\rm atm}}$ [23] in OC remote sensing) that quantifies the radiance due to atmospheric scattering and sea surface reflection of the atmospherically scattered light [24]; and (iii) the diffuse and direct downward irradiances.
It is recalled that the FEM code has been extensively benchmarked with other popular radiative transfer codes [21,25,26] and already applied to perform radiative transfer simulations in realistic conditions [9,26–29].
3. RESULTS AND DISCUSSION
AE are presented and discussed in terms of the percentage contribution ${\xi _{{L_{\rm tot}}}} = 100 \cdot {L_{\rm adj}}/{L_{\rm tot}}$ of the adjacency radiance ${L_{\rm adj}}$ to the total radiance at the sensor ${L_{\rm tot}}$, modeled as ${L_{\rm tot}} = {L_{\rm path}} + t{L_w} + {L_{\rm adj}}$. Adjacency radiance contributions lower than the sensor noise level (NL) [equal to the inverse of the signal-to-noise ratio (SNR)] are regarded as non-discriminable from noise, i.e., non-detectable.
Typical values of ${\xi _{{L_{\rm tot}}}}$ in OLCI data over the whole marine region surrounding the Lampedusa Island are illustrated in Fig. 5 at $\lambda = {865}\;{\rm nm}$ where the adjacency contributions are the largest. Results are presented for average land and water optical conditions with ${\theta _v} = {{20}^\circ}$, ${\phi _v} = {{99}^\circ}$, ${\theta _0} = {{48}^\circ}$, and ${\phi _0} = {{130}^\circ}$.
Fig. 5. ${\xi _{{L_{\rm tot}}}}$ at 865 nm over ${150} \times {150}$ surface elements, for average illumination, land and water optical conditions with ${\theta _v} = {{20}^\circ}$, ${\phi _v} = {{99}^\circ}$, ${\theta _0} = {{48}^\circ}$, and ${\phi _0} = {{130}^\circ}$. The white straight lines indicate the study transects (see Fig. 1). The green and red contour lines identify the noise level (NL) for OLCI-FR and OLCI-RR, respectively. The dashed contour lines are for −NL, while the full contour lines indicate +NL.
Fig. 6. As in Fig. 5 but for $\lambda = {490}\;{\rm nm}$ and for sole ${100} \times {100}$ surface elements.
Contour lines delimiting the NL for the OLCI in full (OLCI-FR) and reduced (OLCI-RR) resolution are also indicated. It is mentioned that the SNR values are scaled to an input radiance typical of cloud-free ocean scenes [10,30].
Remarkably, the adjacency field shows a significantly different pattern around the island.
North of the island in the direction of the reflected sunbeam, the masked glint contributions [term $W$ of Eq. (1)] can be so large to exceed the land contributions. This may lead to slightly negative AE. It is noted that the masked sea surface contributions highly depend on the surface reflectance anisotropy (i.e., on wind speed and direction, and on the sun position), which hinders the possibility to provide a unique description of AE in this area. Yet, standard products do not include quantitative estimates of ${L_w}$ in correspondence of the reflected sunbeam where the overall signal is heavily affected by sun-glint contaminations.
In the remaining marine regions, the masked sea surface contributions are negligible and ${L_{\rm adj}}$ mainly depends on the difference between the land and water reflectance. Adjacency contributions appear always positive approaching 40% at the coast, and exceeding NL up to approximately 5–7 km offshore for OLCI-FR and 10–12 km for OLCI-RR.
Figure 6 shows analogous results at $\lambda = {490}\;{\rm nm}$. Notably, the region affected by significant negative adjacency contributions becomes larger. Indeed, for lower values of the land albedo (see Fig. 4) the contribution of $W$ to ${L_{\rm adj}}$ becomes more important [see Eq. (1)]. South and east of the island, where AE are dominated by the relative difference between land and water reflectance, the region affected by AE is significantly smaller with respect to Fig. 5 because land and water albedos are much closer in the blue spectral region. It is noted that the values of AE are larger than the NL up to about 1.5 and 3 km offshore for OLCI-FR and OLCI-RR, respectively.
AE along the study transects, assumed as representative of the marine region surrounding the island, are further analyzed accounting for the whole set of observation conditions listed in Tables 1 and 2. Mean adjacency contributions at representative center-wavelengths together with the standard deviations $\sigma$, are illustrated in Figs. 7–9 for the transects T1, T2, and T3, respectively. It is recalled that adjacency perturbations larger than the average are expected in summer, while perturbations lower than the average typically occur in winter [9,11].
Fig. 7. Mean values of ${\xi _{{L_{\rm tot}}}}$ at representative center-wavelengths along the transect T1 (see in Fig. 1). The horizontal dotted and dashed lines indicate NL for OLCI-FR and OLCI-RR, respectively. Error bars indicate ${\rm \pm 1}\sigma $.
Data in Figs. 7 and 8 show that south of the island AE becomes lower than the NL for all observation conditions and center-wavelengths with 99.7% (i.e., ${3}\sigma $) level of confidence at approximately 8 and 14 km from the coast for OLCI-FR and OLCI-RR, respectively.
As expected, the average values of ${\xi _{{L_{\rm tot}}}}$ are smaller along T1 than along T2 (e.g., values of ${\xi _{{L_{\rm tot}}}}$ at 865 nm close to the coast are about 7% for T1 and 23% for T2): T2 is indeed closer to a larger portion of land. Conversely, values of σ are larger along T1 due to a stronger dependence of AE on the satellite azimuth. Indeed, for T1 the portion of land below the line-of-sight of the sensor significantly differs between ${\phi _v} = - {{81}^\circ}$ and ${\phi _v} = {{99}^\circ}$.
Notably, the 5% science requirement on ${L_w}$ uncertainties in the blue-green spectral regions implies that ${\xi _{{L_{\rm tot}}}}$ does not exceed (i) the NL at the NIR wavelengths, to ensure an unaffected retrieval of the aerosol optical properties from the NIR bands; and (ii) a tentative maximum uncertainty of 0.3% at the blue-green wavelengths, as determined by an analysis of the impact on SVC of uncertainties in in situ measurements [5]. Overall results (see Figs. 5, 7, and 8) indicate that both conditions could be met for data acquired at distances from the southern and eastern coast of the island larger than approximately 14 km for both OLCI-RR and OLCI-FR.
Values along the northern transect T3 in Fig. 9 exhibit a different trend, showing negative values where the land contributions are smaller than the sky- and sun-glint contributions masked by the island itself. Results nonetheless indicate that, when restricting observations to wind speed lower than ${3.3}\;{{\rm ms}^{ - 1}}$, the basic conditions for SVC could still be met at distances tentatively exceeding 14 km for both OLCI-RR and OLCI-FR.
4. CONCLUSIONS
The adjacency radiance field surrounding the Lampedusa Island has been simulated for typical OLCI observation conditions, except the anisotropy of the land reflectance.
The analysis presented at representative visible and NIR center-wavelengths shows remarkably different adjacency radiance patterns in the marine regions nearby the island.
North of the island in the direction of the reflected sunbeam, AE are mainly dominated by sky- and sun-glint contributions masked by the island itself. As such, AE are slightly negative and decrease with wavelength. Considering that the masked glint contributions strongly depend on the sea state and on the sun position, it is difficult to provide a unique description of the impact of AE on OLCI measurements for this region.
Conversely, south and east of the island AE are mainly dominated by land radiance perturbations. Adjacency contributions are the highest at 865 nm, where they approach 40% in the vicinity of the coast. AE remain larger than the noise level up to approximately 8 and 14 km for OLCI-FR and OLCI-RR, respectively. Beyond such distances from the coast, AE are expected to not affect the 5% science requirement for ${L_w}$ uncertainties in the blue-green spectral region essential for the generation of climate data records.
Conclusively, to avoid the impact of AE, a hypothetical site for the SVC of OLCI observations (in both full and reduced resolution) should be located at distances larger than approximately 14 km from the coast of the considered island, while the region in the direction of the reflected sunbeam should be avoided.
It is finally remarked that the results from this study indicate that an accurate quantification of AE cannot be universally applied on the sole basis of atmospheric and marine optical properties. Additionally, estimates of AE from a specific region cannot be confidently assumed representative of AE for regions exhibiting different albedo and coastline, as well as resulting from diverse viewing and illumination geometries.
Funding
Joint Research Centre; European Association of National Metrology Institutes (16ENV03).
Acknowledgment
The Land Water Mask was retrieved from the online EarthExplorer tool (https://earthexplorer.usgs.gov), courtesy of the NASA EOSDIS Land Processes Distributed Active Archive Center (LP DAAC), USGS/Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota. Dr. D. Meloni, principal investigator of the Lampedusa AERONET site, is duly acknowledged.
Disclosures
The authors declare no conflicts of interest.
REFERENCES
1. S. S. Leroy, J. G. Anderson, and G. Ohring, “Climate signal detection times and constraints on climate benchmark accuracy requirements,” J. Clim. 21, 841–846 (2008). [CrossRef]
2. World Meteorological Organization (WMO), “The global observing system for climate: implementation needs,” GCOS-200 (2016).
3. D. Antoine, “OLCI Level 2—atmospheric corrections over case 1 waters,” Tech. Rep. S3-L2-SD-03-C07-LOV-ATBD (2010).
4. B. A. Franz, S. W. Bailey, P. J. Werdell, and C. R. McClain, “Sensor-independent approach to the vicarious calibration of satellite ocean color radiometry,” Appl. Opt. 46, 5068–5082 (2007). [CrossRef]
5. G. Zibordi, F. Mélin, K. J. Voss, B. C. Johnson, B. A. Franz, E. Kwiatkowska, J.-P. Huot, M. Wang, and D. Antoine, “System vicarious calibration for ocean color climate change applications: requirements for in situ data,” Remote Sens. Environ. 159, 361–369 (2015). [CrossRef]
6. G. Zibordi, F. Melin, and M. Talone, “System vicarious calibration for Copernicus ocean colour missions,” EUR 28851 EN (Publications Office of the European Union, 2017).
7. G. Pace, A. di Sarra, D. Meloni, S. Piacentino, and P. Chamard, “Aerosol optical properties at Lampedusa (Central Mediterranean). 1. Influence of transport and identification of different aerosol types,” Atmos. Chem. Phys. 6, 697–713 (2006). [CrossRef]
8. D. Meloni, A. di Sarra, G. Pace, and F. Monteleone, “Aerosol optical properties at Lampedusa (Central Mediterranean). 2. Determination of single scattering Albedo at two wavelengths for different aerosol types,” Atmos. Chem. Phys. 6, 715–727 (2006). [CrossRef]
9. B. Bulgarelli, V. Kiselev, and G. Zibordi, “Simulation and analysis of adjacency effects in coastal waters: a case study,” Appl. Opt. 53, 1523–1545 (2014). [CrossRef]
10. B. Bulgarelli and G. Zibordi, “On the detectability of adjacency effects in ocean color remote sensing of mid-latitude coastal environments by SeaWiFS, MODIS-A, MERIS, OLCI, OLI and MSI,” Remote Sens. Environ. 209, 423–438 (2018). [CrossRef]
11. B. Bulgarelli and G. Zibordi, “Seasonal impact of adjacency effects on ocean color radiometry at the AAOT validation site,” IEEE Geosci. Remote Sens. Lett. 15, 488–492 (2018). [CrossRef]
12. B. Bulgarelli, G. Zibordi, and F. Melin, “On the minimization of adjacency effects in SeaWiFS primary data products from coastal areas,” Opt. Express 26, A709–A728 (2018). [CrossRef]
13. J. V. Martonchik, C. J. Bruegge, and A. H. Strahler, “A review of reflectance nomenclature used in remote sensing,” Remote Sens. Rev. 19, 9–20 (2000). [CrossRef]
14. M. L. Carroll, J. R. Townshend, C. M. DiMiceli, P. Noojipady, and R. A. Sohlberg, “A new global raster water mask at 250 m resolution,” Int. J. Digit. Earth 2, 291–308 (2009). [CrossRef]
15. E. C. Monahan and I. G. O’Muircheartaigh, “Whitecaps and the passive remote sensing of the ocean surface,” Int. J. Remote Sens. 7, 627–642 (1986). [CrossRef]
16. C. Di Biagio, A. di Sarra, D. Meloni, F. Monteleone, S. Piacentino, and D. Sferlazzo, “Measurements of Mediterranean aerosol radiative forcing and influence of the single scattering albedo,” J. Geophys. Res. 114, 1227 (2009). [CrossRef]
17. G. Zibordi and F. Mélin, “An evaluation of marine regions relevant for ocean color system vicarious calibration,” Remote Sens. Environ. 190, 122–136 (2017). [CrossRef]
18. B. Pinty, A. Lattanzio, J. V. Martonchik, M. M. Verstraete, N. Gobron, M. Taberner, J.-L. Widlowski, R. E. Dickinson, and Y. Govaerts, “Coupling diffuse sky radiation and surface Albedo,” J. Atmos. Sci. 62, 2580–2591 (2005). [CrossRef]
19. E. G. Moody, M. D. King, C. B. Schaaf, and S. Platnick, “MODIS-derived spatially complete surface Albedo products: spatial and temporal pixel distribution and zonal averages,” J. Appl. Meteor. Climatol. 47, 2879–2894 (2008). [CrossRef]
20. V. B. Kisselev, L. Roberti, and G. Perona, “Finite-element algorithm for radiative transfer in vertically inhomogeneous media: numerical scheme and applications,” Appl. Opt. 34, 8460–8471 (1995). [CrossRef]
21. B. Bulgarelli, V. Kisselev, and L. Roberti, “Radiative transfer in the atmosphere-ocean system: the finite-element method,” Appl. Opt. 38, 1530–1542 (1999). [CrossRef]
22. H. R. Gordon and B. A. Franz, “Remote sensing of ocean color: assessment of the water-leaving radiance bidirectional effects on the atmospheric diffuse transmittance for SeaWiFS and MODIS intercomparisons,” Remote Sens. Environ. 112, 2677–2685 (2008). [CrossRef]
23. H. W. Gordon and M. Wang, “Retrieval of water-leaving radiance and aerosol optical thickness over the oceans with SeaWiFS: a preliminary algorithm,” Appl. Opt. 33, 443–452 (1994). [CrossRef]
24. D. Antoine and A. Morel, “A multiple scattering algorithm for atmospheric correction of remotely sensed ocean colour (MERIS instrument): principle and implementation for atmospheres carrying various aerosols including absorbing ones,” Int. J. Remote Sens. 20, 1875–1916 (1999). [CrossRef]
25. B. Bulgarelli and J. Doyle, “Comparison between numerical models for radiative transfer simulation in the atmosphere-ocean system,” J. Quant. Spectrosc. Radiat. Transfer 86, 315–334 (2004). [CrossRef]
26. B. Bulgarelli, G. Zibordi, and J.-F. Berthon, “Measured and modeled radiometric quantities in coastal waters: toward a closure,” Appl. Opt. 42, 5365–5381 (2003). [CrossRef]
27. B. Bulgarelli and G. Zibordi, “Remote sensing of ocean colour: accuracy assessment of an approximate atmospheric correction method,” Int. J. Remote Sens. 24, 491–509 (2003). [CrossRef]
28. B. Bulgarelli and F. Mélin, “SeaWiFS-derived products in the Baltic Sea: performance analysis of a simple atmospheric correction algorithm,” Oceanologia 45, 655–677 (2003), http://agro.icm.edu.pl/agro/element/bwmeta1.element.agro-article-4c93f680-6fcb-4467-aa6c-fe1eb7c6d3ca.
29. G. Zibordi and B. Bulgarelli, “Effects of cosine error in irradiance measurements from field ocean color radiometers,” Appl. Opt. 46, 5529–5538 (2007). [CrossRef]
30. C. Hu, L. Feng, Z. Lee, C. O. Davis, A. Mannino, C. R. McClain, and B. A. Franz, “Dynamic range and sensitivity requirements of satellite ocean color sensors: learning from the past,” Appl. Opt. 51, 6045–6062 (2012). [CrossRef]
