Volcanic Ash Cloud Height via CO2 Slicing

The study addresses the significant aviation hazard posed by geographically widespread volcanic ash clouds. Accurate monitoring of ash cloud emission and atmospheric dispersion, especially plume altitude, is crucial for minimizing risks and aviation disruption. The altitude is a key input for various ash cloud propagation models. The research leverages the CO2 slicing technique, a method proven effective for determining aqueous cloud top height, and explores its adaptation and application to volcanic ash. This technique offers a potential solution for improving the accuracy and efficiency of volcanic ash cloud height estimations, thereby enhancing aviation safety and reducing flight disruptions. Prior research suggested the potential of this approach, notably using the Moderate Resolution Imaging Spectroradiometer (MODIS). This study extends this work by adapting the technique for use with the Infrared Atmospheric Sounding Interferometer (IASI), a more advanced instrument for satellite remote sensing.

The core methodology utilizes the CO2 slicing technique, an established method for determining cloud top height/pressure in aqueous clouds. This technique analyzes the CO2 absorption feature in the thermal infrared spectrum (665-750 cm⁻¹ or 13.3-15 µm). The increasing radiance observed within this region, coupled with the decreasing temperature with height in the troposphere, allows for the identification of cloud or ash top heights by observing spectral divergence between clear and cloudy/ashy conditions. The research utilizes data from the Infrared Atmospheric Sounding Interferometer (IASI), a high-spectral-resolution instrument onboard meteorological satellites. IASI scans provide data across a wide swath, offering extensive coverage for monitoring volcanic ash plumes. Simulated ash spectra are used to select the optimal IASI channels for the application of the CO2 slicing technique to ensure computational efficiency and accuracy of the volcanic ash cloud height estimation.

The study begins by applying the CO2 slicing technique to simulated ash spectra, allowing for the optimization of channel selection for different atmospheric conditions. This simulation phase provided an important benchmark for evaluating the technique's effectiveness. The Root Mean Square Error (RMSE) obtained from simulated data was less than 800m, indicating a strong correlation between the simulated and retrieved heights. Next, the validated method is applied to actual IASI spectra obtained during the Eyjafjallajökull (2010) and Grímsvötn (2011) volcanic eruptions. Both eruptions produced well-documented tropospheric ash plumes, making them ideal case studies. The CO2 slicing results are meticulously compared with an existing optimal estimation (OE) scheme for IASI and validated using data from CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization), a lidar instrument renowned for its accuracy in cloud and ash height measurements. The comparison shows that the CO2 slicing results demonstrate a lower RMSE value (2.2 km) compared to the OE scheme (2.8 km) when validated against CALIOP data.

A critical aspect of the study involves validation and comparison of the CO2 slicing technique against established methods. Direct comparison with an independent optimal estimation (OE) scheme, also designed for IASI data, provides insight into the relative strengths and weaknesses of the two retrieval approaches. Furthermore, validation against CALIOP lidar data provides an independent ground truth assessment of the volcanic ash cloud height estimates. The comparison demonstrates the effectiveness of the proposed CO2 slicing approach in capturing the height of volcanic ash plumes, particularly in comparison to the OE scheme. While discrepancies exist due to varying spatial and temporal resolutions, and inherent assumptions within each technique, CO2 slicing frequently exhibited a closer agreement with the CALIOP lidar data, indicating its potential for improved accuracy in volcanic ash cloud height retrieval using satellite remote sensing technology.

The study leverages data from the Infrared Atmospheric Sounding Interferometer (IASI), a key instrument onboard the Metop A, B, and C meteorological satellites. Launched in 2006, 2012, and 2018 respectively, each IASI instrument orbits the Earth twice daily. Its wide swath width (2200 km) and high spectral resolution (0.5 cm⁻¹) in the infrared range (645-2760 cm⁻¹ or 3.62-15.5 µm) are critical for the research. The instrument scans consist of groups of four circular pixels (12 km diameter at nadir), providing detailed spatial information. This high spectral and spatial resolution of IASI makes it a valuable tool for remote sensing, capable of accurately measuring atmospheric properties. The use of IASI is vital for the high accuracy and resolution of the volcanic ash cloud height measurements. The spectral resolution of IASI allows the fine differentiation of the CO2 slicing technique to accurately measure the height of the plume by analyzing the CO2 absorption feature within the infrared data, crucial for precise satellite remote sensing.

The core of the research employs the established CO2 slicing technique. This method, initially developed for determining aqueous cloud top height, is adapted here for estimating volcanic ash cloud height. The technique utilizes the CO2 absorption feature in the thermal infrared spectrum. In the Earth's troposphere, where temperature decreases with altitude, measured radiances are directly proportional to atmospheric transparency for each channel. As wavenumber and measured radiance increase within the CO2 absorption band, channels become progressively more transparent. This allows for identifying the altitude at which an ash cloud begins to deviate from a clear atmospheric spectrum; this point of divergence reveals the volcanic ash cloud height. The study emphasizes the importance of channel selection, utilizing simulated ash spectra to identify the most suitable IASI channels for optimal CO2 slicing performance. This optimized channel selection improves the accuracy and reliability of the height estimations using satellite remote sensing data. The process involves careful consideration of factors such as signal-to-noise ratios and emissivity ratios to ensure robust height retrieval.

Application of the CO2 slicing technique to IASI data requires careful consideration of several factors. Since the method has historically been used for tropospheric aqueous clouds, this study focuses solely on tropospheric solutions for volcanic ash cloud height. Prior information, such as ground observations or pilot reports, may be necessary to confirm that the plume is indeed within the troposphere. The study also incorporates quality control measures to ensure the reliability of the results. These criteria involve checking if observed-minus-clear radiance values fall within the instrument's noise level for selected CO2 and reference channels. The algorithm also filters out results with effective emissivity values outside the 0-1.05 range. Multiple height solutions, if present, are addressed using established techniques like a top-down approach or a method based on the radiative transfer equation, resulting in a final volcanic ash cloud height estimation. The algorithm's performance is also impacted by atmospheric conditions, especially near the tropopause and the surface where the distinction between clear and ashy spectra may be less pronounced, potentially causing issues for satellite remote sensing of low-level or thin ash clouds.

The study begins with a crucial step: applying the CO2 slicing technique to simulated ash spectra. This allows for a controlled environment to test and refine the method. A total of 384 simulated ash spectra were generated, representing a variety of atmospheric profiles (high latitude, mid-latitude day/night, tropical daytime, polar summer/winter) and ash properties. These properties included variations in cloud height (200-900 hPa), ash effective radius (5-10 µm), and ash optical depth (5-15 at 550nm). The range of ash optical depth explored is significant because it encompasses values from highly dispersed plumes (less than 1.2, as observed in the Eyjafjallajökull eruption) to much higher values expected closer to the volcanic vent following major eruptions. The primary goal of this stage was the selection of optimal IASI channels for the CO2 slicing technique. Channel selection was based on minimizing the difference between true and retrieved pressures, maximizing the percentage of successful retrievals, and considering the emissivity ratio between channel pairs. The results of this simulated data analysis guided the later application to real-world data and are essential for the accuracy of volcanic ash cloud height estimations using satellite remote sensing techniques. The RMSE for the simulated data was under 800m, showcasing the strong potential of the method before moving to real-world datasets.

Following the simulated data analysis, the adapted CO2 slicing technique was applied to IASI spectra obtained during the Eyjafjallajökull (2010) and Grímsvötn (2011) eruptions. These eruptions were chosen for their extensive documentation through satellite imagery and their tropospheric ash plume emissions, making them ideal for validating the technique. The CO2 slicing results were compared to those from an optimal estimation (OE) scheme, also developed for IASI, providing a comparative analysis of two different retrieval methods. Further validation was conducted using data from a satellite-borne lidar (CALIOP), an independent source of ground truth data, to assess the accuracy of both the CO2 slicing and OE methods. This multifaceted approach (CO2 slicing, OE, and CALIOP) ensures a comprehensive validation. The Eyjafjallajökull and Grímsvötn eruptions provide extensive real-world validation data, and comparisons were conducted with a spatial and temporal tolerance of 50km and 2hrs. The results show that the CO2 slicing method demonstrates a lower RMSE (2.2 km) compared to the OE scheme (2.8 km) when compared against CALIOP data, confirming its potential for accurate volcanic ash cloud height determination.

The comparison of CO2 slicing results with the OE scheme and CALIOP lidar data reveals both strengths and limitations. While both the CO2 slicing and OE methods generally capture the overall height of the ash layer, there are notable differences in specific cases, particularly when multiple cloud layers are present. The comparison highlights how the varying spatial resolutions (335m for CALIOP versus 12km for IASI) and the difference in measurements (backscattering for CALIOP versus thermal emission for IASI) impact the comparison. The analysis of the Eyjafjallajökull eruption shows temporal variations in average plume height, which are better captured by the CO2 slicing method compared to the OE scheme, especially at higher altitudes. Differences were also noted at lower altitudes close to the surface, where the accuracy of both techniques is reduced due to less distinction between the clear and cloudy or ashy spectra. The study identifies cases where one technique outperforms the other, illustrating the need for a multi-method approach to improve confidence in retrieved volcanic ash cloud height using satellite remote sensing. The findings underscore the potential of CO2 slicing as a quick and efficient method, but emphasize the need for future research to address its limitations, especially for thin or low-level ash clouds.

To assess the accuracy of the CO2 slicing technique, a comparison was conducted against an existing optimal estimation (OE) scheme also designed for IASI data. Both methods were applied to the same IASI data from the Eyjafjallajökull and Grímsvötn eruptions. The distribution of heights obtained from both methods showed that the peak of the distribution for the CO2 slicing heights was consistently higher than that of the OE scheme. A time series analysis of the Eyjafjallajökull eruption's average retrieved height revealed that the CO2 slicing method captured the temporal variations in plume height more effectively, showing a significant decrease over the study period, correlating with observed changes in volcanic activity. The OE scheme, however, showed less variability in average height throughout the study period. These differences are likely due to the inherent differences in the design, assumptions, and limitations of each technique. The OE scheme uses fewer channels (105 compared to the 57 in the CO2 slicing method), and importantly, these channels were not optimized for ash height retrieval, unlike those selected in the CO2 slicing approach using simulated data. The OE method’s reliance on a priori height is a significant difference between the two techniques and could be a source of discrepancy.

Further validation was performed using data from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument onboard the CALIPSO satellite. CALIOP provides high-resolution backscatter profiles, offering an independent assessment of ash cloud height. To compare the results, CALIOP overpasses intersecting with the ash plumes (identified using SEVIRI imagery) were collocated with IASI pixels within a 50km radius and a 2-hour time window. The spatial resolutions of CALIOP (335m) and IASI (12km) were considered in the collocation process, with the CALIOP data smoothed to match the IASI resolution. Cloud top height from CALIOP backscatter profiles was determined by calculating cumulative backscatter and identifying the threshold where atmospheric extinction exceeded a manually set level. This level was chosen for each scene to best match observed cloud top heights. The comparison of heights from CO2 slicing, OE, and CALIOP showed that both the CO2 slicing and OE methods generally captured the shape of the ash layer, although there were instances where one method performed better than the other. The differences could be related to factors such as thick underlying cloud layers, which might lead to both techniques underestimating the cloud top height. The lower RMSE of 2.2 km for the CO2 slicing technique compared to the 2.8 km for the OE scheme, when compared to CALIOP measurements, further reinforces the accuracy and potential of the new approach in volcanic ash cloud height estimation.

While comparisons with CALIOP lidar data are common for validating cloud and ash height retrievals, the study acknowledges limitations. The differences in measurement types (backscattering vs thermal emission), spatial resolutions, and measurement locations introduce uncertainty in direct comparisons. The temporal differences were minimized (2 hour window), but rapid changes in cloud structure still pose a challenge. Despite these limitations, the comparisons offered valuable insight into the performance of the CO2 slicing and OE techniques. The study suggests that differences between the CO2 slicing and OE results could be partially attributed to the differing channel selections and the influence of a priori height in the OE scheme. The authors propose that using the CO2 slicing results as a priori height in the OE scheme, or modifying the channels used in the OE method, may improve its performance. The results, however, overall demonstrate the accuracy of the CO2 slicing technique and the value of utilizing lidar data for validation in volcanic ash cloud height estimation using satellite remote sensing.

The study concludes that the adapted CO2 slicing technique shows considerable promise for quickly estimating volcanic ash cloud height from IASI data. The method's relatively fast processing speed is a key advantage, making it suitable for near real-time applications in aviation hazard mitigation. The RMSE of 2.2km when compared to CALIOP lidar data demonstrates improved accuracy compared to the existing optimal estimation (OE) scheme for IASI (RMSE of 2.8km). The results obtained from the Eyjafjallajökull and Grímsvötn eruptions confirm the method's capability to capture the dynamic changes in plume height. The successful application to real-world data from these well-studied eruptions, combined with the validation using CALIOP lidar data, provides strong evidence supporting the effectiveness of the adapted CO2 slicing technique. However, the study acknowledges that the technique's performance can be limited in cases of low-level or optically thin ash clouds or where steep temperature gradients exist, highlighting the need for continued improvement and refinement.

The authors suggest several avenues for future research to improve the accuracy and applicability of the CO2 slicing technique. One approach involves creating synthetic channels (by averaging multiple channels) to enhance the algorithm's sensitivity to lower-level clouds, which is a known limitation identified in the study. Further exploration into alternative channel selection methods or different approaches to determine the final cloud height is also warranted. The current channel selection process used simulated data from six atmospheric profiles; future work might focus on tailoring channel selection to specific atmospheric climatologies to further improve accuracy. A deeper investigation into the strengths and limitations of the technique under different atmospheric conditions is critical for understanding its suitability in various scenarios. Investigating the potential use of the CO2 slicing results as a priori data for other retrieval schemes, such as the OE scheme, is also a promising area of research. This could lead to improved volcanic ash cloud height estimations via satellite remote sensing.