Presentation Information
[AAS08-P06]Validation of Pandora NO2 remote sensing observation
*Sumire Ishii1, Kazuyuki Kita1, Tamaki Fujinawa2, Satoshi Inomata2, Takafumi Sugita2, Astrid Müller2, Kei Shiomi3, Hiroshi Tanimoto2 (1.College of Science, Ibaraki university, 2.National Institute for Environmental Studies, 3.Earth Observation Research Center, Japan Aerospace Exploration Agency)
Keywords:
Pandora,Remote sensing,NO2,Validation,GOSAT-GW
1.Introduction
The Global Observing SATellite for Greenhouse gases and Water cycle (GOSAT-GW), launched on June, 2025, can observe NO2 in addition to the CO2 and CH4 observed by GOSAT-1/2. NO2 is emitted alongside CO2 in the combustion of fossil/biomass fuels. Because it has a shorter atmospheric lifetime than CO2, its amount exhibits more evident increase within air masses affected by the combustion. Therefore, simultaneous observation of NO2 and CO2 is expected to improve the accuracy of CO2 emission estimates.
Validation of GOSAT-GW NO2 utilizes ground-based Pandora spectrometer operated within the Pandonia Global Network (PGN). This study aims to clarify the magnitude of errors and contribution of each error source in Pandora NO2 data. This is to verify the accuracy of Pandora NO2 data, in order to assess their accuracy prior to validating GOSAT-GW.
2. Observation data
The Pandora spectrometer measure UV-VIS spectra of direct sunlight and sky-scattered radiation, from which NO2 vertical column density (VCD) are derived. In Direct-sun (DS) mode, it measures the direct sunlight spectrum and determines the total NO2 amount using the DOAS method. By subtracting the climatological stratospheric VCD of NO2 derived from satellite observations from the total amount, the tropospheric VCD (t-VCD) of NO2 is also estimated. In Sky-scan (SS) mode, the scattered solar light spectrum is measured from multiple elevation angles, and the t-VCD and surface NO2 concentration are evaluated using the MAX-DOAS method.
This study uses data from Pandora instruments 163 and 176 installed at the National Institute for Environmental Studies (NIES) in Tsukuba, and Pandora 193 installed at the JAXA Tsukuba Space Center, to (1)compare Pandora surface NO2 concentration data with in situ surface NO2 observation data (the data from the CAPS instrument installed at NIES and median of four surrounding continuous monitoring stations), (2)compare t-VCD observed in DS and SS modes for each instrument, and (3)perform cross-comparisons of t-VCDs among the three Pandora instruments. The target period was February 1, 2025, to August 31, 2025, when data from all Pandora instruments were available.
3.Results and Discussion
Fig.1 (left) compares the hourly average t-VCD measured by Pandora176 in DS mode and SS mode. During the GOSAT-GW observation period from 13:00 to 14:00, the values were in good agreement, with SS/DS = 1.02 ± 0.015 [10-3 mol/m2]. However, as shown in Fig.1 (right), comparing the tropospheric VCD measurements from both modes with the surface NO2 concentrations measured with CAPS revealed that the t-VCD values, particularly in DS mode, exhibits an offset of approximately 0.04×10-3 mol/m2 when the surface NO2 concentration is close to zero. Considering that most of the tropospheric VCD of NO2 is present near the surface, this offset is likely significant. It may arise because the DS mode estimates t-VCD by measuring the total amount of NO2 and subtracting the stratospheric VCD climatological value whereas the SS mode estimates t-VCD from the NO2 altitude distribution derived from observational data.
Fig.2 shows histograms of the t-VCD differences between individual three simultaneous measurements and their average (SS mode on the left, DS mode on the right). While no clear instrumental difference was found in SS mode, the differences from the average in DS mode were, on average, -0.0004 [10-3 mol/m2] for P163, 0.0067 for P176, and -0.0062 for P193, indicating an instrumental difference of magnitude comparable to the standard deviation. Differences between SS and DS modes stem partly from Pandora's SS mode exhibiting larger VCD standard deviation and DS mode's wider elevation angle viewing a narrower horizontal field than SS mode, amplifying location-dependent variations. However, in both modes, Pandora176 indicates systematically larger values, suggesting instrument error originating from the spectrometer device function inaccuracies.
The primary causes of random error were also investigated. In DS mode, the atmospheric variability value, indicating the influence of clouds on the direct solar path, significantly affected random error. In contrast, in SS mode, spectral fitting error, considered as the main error factor, did not significantly affect random error.
The Global Observing SATellite for Greenhouse gases and Water cycle (GOSAT-GW), launched on June, 2025, can observe NO2 in addition to the CO2 and CH4 observed by GOSAT-1/2. NO2 is emitted alongside CO2 in the combustion of fossil/biomass fuels. Because it has a shorter atmospheric lifetime than CO2, its amount exhibits more evident increase within air masses affected by the combustion. Therefore, simultaneous observation of NO2 and CO2 is expected to improve the accuracy of CO2 emission estimates.
Validation of GOSAT-GW NO2 utilizes ground-based Pandora spectrometer operated within the Pandonia Global Network (PGN). This study aims to clarify the magnitude of errors and contribution of each error source in Pandora NO2 data. This is to verify the accuracy of Pandora NO2 data, in order to assess their accuracy prior to validating GOSAT-GW.
2. Observation data
The Pandora spectrometer measure UV-VIS spectra of direct sunlight and sky-scattered radiation, from which NO2 vertical column density (VCD) are derived. In Direct-sun (DS) mode, it measures the direct sunlight spectrum and determines the total NO2 amount using the DOAS method. By subtracting the climatological stratospheric VCD of NO2 derived from satellite observations from the total amount, the tropospheric VCD (t-VCD) of NO2 is also estimated. In Sky-scan (SS) mode, the scattered solar light spectrum is measured from multiple elevation angles, and the t-VCD and surface NO2 concentration are evaluated using the MAX-DOAS method.
This study uses data from Pandora instruments 163 and 176 installed at the National Institute for Environmental Studies (NIES) in Tsukuba, and Pandora 193 installed at the JAXA Tsukuba Space Center, to (1)compare Pandora surface NO2 concentration data with in situ surface NO2 observation data (the data from the CAPS instrument installed at NIES and median of four surrounding continuous monitoring stations), (2)compare t-VCD observed in DS and SS modes for each instrument, and (3)perform cross-comparisons of t-VCDs among the three Pandora instruments. The target period was February 1, 2025, to August 31, 2025, when data from all Pandora instruments were available.
3.Results and Discussion
Fig.1 (left) compares the hourly average t-VCD measured by Pandora176 in DS mode and SS mode. During the GOSAT-GW observation period from 13:00 to 14:00, the values were in good agreement, with SS/DS = 1.02 ± 0.015 [10-3 mol/m2]. However, as shown in Fig.1 (right), comparing the tropospheric VCD measurements from both modes with the surface NO2 concentrations measured with CAPS revealed that the t-VCD values, particularly in DS mode, exhibits an offset of approximately 0.04×10-3 mol/m2 when the surface NO2 concentration is close to zero. Considering that most of the tropospheric VCD of NO2 is present near the surface, this offset is likely significant. It may arise because the DS mode estimates t-VCD by measuring the total amount of NO2 and subtracting the stratospheric VCD climatological value whereas the SS mode estimates t-VCD from the NO2 altitude distribution derived from observational data.
Fig.2 shows histograms of the t-VCD differences between individual three simultaneous measurements and their average (SS mode on the left, DS mode on the right). While no clear instrumental difference was found in SS mode, the differences from the average in DS mode were, on average, -0.0004 [10-3 mol/m2] for P163, 0.0067 for P176, and -0.0062 for P193, indicating an instrumental difference of magnitude comparable to the standard deviation. Differences between SS and DS modes stem partly from Pandora's SS mode exhibiting larger VCD standard deviation and DS mode's wider elevation angle viewing a narrower horizontal field than SS mode, amplifying location-dependent variations. However, in both modes, Pandora176 indicates systematically larger values, suggesting instrument error originating from the spectrometer device function inaccuracies.
The primary causes of random error were also investigated. In DS mode, the atmospheric variability value, indicating the influence of clouds on the direct solar path, significantly affected random error. In contrast, in SS mode, spectral fitting error, considered as the main error factor, did not significantly affect random error.