Academia.eduAcademia.edu

Figure 3 – uploaded by Steffen Dransfeld

See full PDFdownloadDownload figure
Figure 3. Difference OLCI of observations against SuomiNet (a, b) and IGRA (ce, d) for OLCI-A (a, c) and OLCI-B (b, d). The solid line shows the median of each 2 kg m7? wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the 5th and 95th percentiles. Bins with fewer than 20 matchups were omitted from the figure.

Figure 3 Difference OLCI of observations against SuomiNet (a, b) and IGRA (ce, d) for OLCI-A (a, c) and OLCI-B (b, d). The solid line shows the median of each 2 kg m7? wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the 5th and 95th percentiles. Bins with fewer than 20 matchups were omitted from the figure.

Related Figures (7)

Figure 1. Locations of OLCI-A matchups with IGRA (a) and SuomiNet (b) networks.
Figure 1. Locations of OLCI-A matchups with IGRA (a) and SuomiNet (b) networks.
Figure 2. Density scatter plot of LAND comparisons of OLCI-A (a, c) and OLCI-B (b, d) against SuomiNet (a, b) and IGRA (ce, d) observations, with CLOUD_MARGIN and CLOUD_AMBIGUOUS matchups removed. The colour field shows the percentage of matchups in each category, with the darkest colours showing the highest density of matchups. The blue dashed line shows the x = y line and the black lines show the median (dashed) and 16th and 84th percentile (dotted) OLCI-A observations for each 2 kg m7? bin of reference observations. Linear fits of the matchups (not shown) for SuomiNet are y = 1.12x — 0.31 (OLCI-A) and y = 1.11x — 0.29 (OLCI-B) and for IGRA y = 1.07x + 0.69 (OLCI-A) and y = 1.07x +0.64 (OLCI-B). Correlation coefficients are 0.98 (SuomiNet) and 0.90 (GRA) for both instruments.
Figure 2. Density scatter plot of LAND comparisons of OLCI-A (a, c) and OLCI-B (b, d) against SuomiNet (a, b) and IGRA (ce, d) observations, with CLOUD_MARGIN and CLOUD_AMBIGUOUS matchups removed. The colour field shows the percentage of matchups in each category, with the darkest colours showing the highest density of matchups. The blue dashed line shows the x = y line and the black lines show the median (dashed) and 16th and 84th percentile (dotted) OLCI-A observations for each 2 kg m7? bin of reference observations. Linear fits of the matchups (not shown) for SuomiNet are y = 1.12x — 0.31 (OLCI-A) and y = 1.11x — 0.29 (OLCI-B) and for IGRA y = 1.07x + 0.69 (OLCI-A) and y = 1.07x +0.64 (OLCI-B). Correlation coefficients are 0.98 (SuomiNet) and 0.90 (GRA) for both instruments.
Figure 4. Difference OLCI-A (a, c) and OLCI-B (b, d) of observations against SuomiNet as a function of latitude (a, b) and solar zenith angle (c, d). The solid line shows the median of each 2° wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the 5th and 95th percentiles. Bins with fewer than 20 matchups were omitted from the figure.
Figure 4. Difference OLCI-A (a, c) and OLCI-B (b, d) of observations against SuomiNet as a function of latitude (a, b) and solar zenith angle (c, d). The solid line shows the median of each 2° wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the 5th and 95th percentiles. Bins with fewer than 20 matchups were omitted from the figure.
Table 1. Statistics of general comparisons shown in Fig. 2. Figure 5. Difference OLCI-A (a, c) and OLCI-B (b, d) of observations against IGRA as a function of latitude (a, b) and solar zenith angle (c, d). The solid line shows the median of each 2° wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the 5th and 95th percentiles. Bins with fewer than 20 matchups were omitted from the figure.
Table 1. Statistics of general comparisons shown in Fig. 2. Figure 5. Difference OLCI-A (a, c) and OLCI-B (b, d) of observations against IGRA as a function of latitude (a, b) and solar zenith angle (c, d). The solid line shows the median of each 2° wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the 5th and 95th percentiles. Bins with fewer than 20 matchups were omitted from the figure.
Figure 6. Dependency of OLCI-SuomiNet relative difference on camera index. (a, b) Distributions of relative differences for OLCI-A (a) and OLCI-B (b) matchups with SuomiNet observations for OLCI cameras 1—5. Red lines show the medians of the distributions and blue boxes the 16th and 84th percentiles of the distribution with black whiskers showing percentiles 2.5 and 97.5. Red crosses show the means of the distributions. (c, d) Relative difference as a function of central wavelength of the detector for OLCI-A (c) and OLCI-B (d). Colours represent cameras 1—5 (see legend), and the solid and dashed lines represent the median and the interpercentile range (16th—84th) percentiles, respectively.
Figure 6. Dependency of OLCI-SuomiNet relative difference on camera index. (a, b) Distributions of relative differences for OLCI-A (a) and OLCI-B (b) matchups with SuomiNet observations for OLCI cameras 1—5. Red lines show the medians of the distributions and blue boxes the 16th and 84th percentiles of the distribution with black whiskers showing percentiles 2.5 and 97.5. Red crosses show the means of the distributions. (c, d) Relative difference as a function of central wavelength of the detector for OLCI-A (c) and OLCI-B (d). Colours represent cameras 1—5 (see legend), and the solid and dashed lines represent the median and the interpercentile range (16th—84th) percentiles, respectively.
Figure 7. (a—d) Structure function (./D(p) in kg m7) within macropixels, compared to the centre of the macropixel (a, b) and the square root of mean error estimate of corresponding OLCI pixels (c, d). (e, f) Distribution of experimental uncertainty estimates (8 pixels around the centre of the macropixel) using the structure function method ./D(p) and the IWV uncertainty estimates from the inversion algorithm. Red lines show the medians of the distributions and blue boxes the 16th and 84th percentiles of the distribution, with black whiskers showing the Sth and 95th percentiles. Red crosses show the means of the distributions.
Figure 7. (a—d) Structure function (./D(p) in kg m7) within macropixels, compared to the centre of the macropixel (a, b) and the square root of mean error estimate of corresponding OLCI pixels (c, d). (e, f) Distribution of experimental uncertainty estimates (8 pixels around the centre of the macropixel) using the structure function method ./D(p) and the IWV uncertainty estimates from the inversion algorithm. Red lines show the medians of the distributions and blue boxes the 16th and 84th percentiles of the distribution, with black whiskers showing the Sth and 95th percentiles. Red crosses show the means of the distributions.
Figure A1. Dependency of observed OLCI difference against SuomiNet (a, b) and IGRA (ce, d) observations to distance to closest cloud- flagged pixel for OLCI-A (a, c) and OLCI-B (b, d). Black lines show the difference in kg m2, while the red lines show the relative difference in percent. The solid line shows the median of each 2 kg m2 wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the Sth and 95th percentiles. Appendix A
Figure A1. Dependency of observed OLCI difference against SuomiNet (a, b) and IGRA (ce, d) observations to distance to closest cloud- flagged pixel for OLCI-A (a, c) and OLCI-B (b, d). Black lines show the difference in kg m2, while the red lines show the relative difference in percent. The solid line shows the median of each 2 kg m2 wide bin, while the dashed lines show the 16th and 84th percentiles and the dotted lines the Sth and 95th percentiles. Appendix A
Related topics:
Environmental ScienceRemote SensingAtmospheric sciencesAtmospheric Measurement TechniquesPrecipitable Water
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved