Trends in spectrally resolved outgoing longwave radiation from 10 years of satellite measurements

Earth's climate is intrinsically linked to energy flow in and out of the Earth-atmosphere system. Equilibrium is maintained when outgoing longwave radiation (OLR) and reflected shortwave radiation compensate for incoming solar radiation. Any disruption to this balance, a radiative forcing, triggers climate feedbacks. Understanding long-term changes necessitates accurate evaluation of climate drivers' radiative effects on OLR. Broadband instruments have yielded improvements, but techniques like radiative kernel and partial radiative perturbation rely on accurate parameter knowledge, potentially introducing regional biases. Hyperspectral sounders offer independent constraints via spectrally resolved OLR (SOLR), revealing individual climate processes' spectral signatures. While the potential of SOLR for climate change studies was recognized earlier, its widespread use has only recently emerged with the availability of stable multi-year observations. Studies using Atmospheric Infrared Sounder (AIRS) data have produced significant results, highlighting biases in climate models not apparent in broadband flux comparisons and revealing cooling trends in the CO₂ v₂ band for stratospherically sensitive channels. The Infrared Atmospheric Sounding Interferometer (IASI) onboard Metop satellites offers advantages, covering the 645–2760 cm⁻¹ thermal infrared region without gaps, making it ideal for detecting SOLR changes linked to trace gas variations. IASI's long-term stability and use in sensor inter-calibration make it a valuable tool. Previous studies using IASI data have shown small variations allowing the identification of robust changes, and identified clear spectral signatures related to greenhouse gas increases. This study analyzes 10 years of IASI clear-sky SOLR data to assess the sensitivity of IASI-derived SOLR in detecting and quantifying small changes and linking them to surface and atmospheric conditions. The primary research questions are: 1. Can statistically significant SOLR trends be detected from IASI channels sensitive to different altitudes? 2. Can these trends be linked to known climate processes? 3. Can the effect of increasing greenhouse gases be detected?

The literature review highlights the existing research on satellite-derived hyperspectral radiance measurements for assessing climate drivers and their feedback on OLR. It discusses previous work using broadband instruments and the limitations of those methods. It also reviews prior studies utilizing AIRS data, emphasizing their success in identifying model biases and detecting significant cooling trends in CO2 absorption bands. The review covers previous research using IASI measurements to investigate interannual radiance variability and detect changes in the Earth's OLR spectrum, highlighting the instrument's suitability for this type of analysis. The review sets the stage for the current study, justifying the use of IASI data and outlining the specific research questions that the study aims to address.

Clear-sky SOLR data were derived from a reprocessed dataset of IASI/Metop-A radiance measurements (645–2300 cm⁻¹ at 0.25 cm⁻¹ spectral sampling). The algorithm for converting spectra to fluxes is detailed in Whitburn et al. (2020). The data achieve good precision (±0.005 W m⁻² (cm⁻¹)⁻¹). Cloud-free scenes were selected using Advanced Very High Resolution Radiometer (AVHRR) cloud information, resulting in ~14% of observations. The dataset comprises 10 years (2008–2017) of daily global SOLR (2° × 2° grid), focusing on daytime measurements over sea to minimize local variations. Linear trends (LT) were computed from daily, zonally averaged (2° latitude) SOLR using Gardiner et al.'s (2008) method, fitting a low-order Fourier series (n=3) for intra-annual variability plus a linear term for the annual trend. Bootstrap resampling determined confidence limits, with significant trends defined as those where the 95% confidence interval doesn't include zero. The impact of land measurements and cloud distribution heterogeneity on zonal trends is acknowledged. Additional analysis included calculating global LT on a 2° × 2° grid for selected channels sensitive to different altitudes, and global temperature LT from ERA5 reanalysis data. The altitude of peak radiance changes was determined by calculating relative changes in Earth's radiation after passage through 1-km-thick atmospheric layers. This involved using a line-by-line radiative transfer code (Atmosphit) with tropical and subarctic standard atmospheres.

Analysis of 10 years of IASI data revealed spatial inhomogeneities in SOLR trends across spectral regions. In atmospheric window regions (795–970 cm⁻¹, 1070–1230 cm⁻¹, 2090–2170 cm⁻¹), changes primarily reflected surface temperature changes, influenced by El Niño/Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO) events. Positive trends (+0.03 to +0.05% per year) were observed in the tropics and mid-latitudes, aligning with warmer mean sea surface temperature (SST) towards the end of the period. Cooling trends were identified in the North Atlantic (related to the North Atlantic Warming Hole), and the Western Pacific (part of the PDO pattern). At high latitudes (above 60°N and S), negative trends were observed, attributed to cooling in the North Atlantic. In the Southern Hemisphere, warming and cooling areas reflected Antarctic Dipole (ADP) activity. Spectral structures in the window regions indicated changes in atmospheric absorption lines, reflecting changes in CO2 and H2O concentrations. The increase in CO2 and CH4 resulted in negative LT (-0.05 to -0.3% per year) in their absorption bands, due to increased absorption of OLR. For mid- and upper-tropospheric channels, trends reflected greenhouse gas concentrations and dynamical processes driven by SST changes. The strong H2O absorption channels showed negative LT in the tropics and mid-latitudes, likely due to stratospheric cooling and increased upper-tropospheric H2O. The ν3 O3 band exhibited positive zonal LT (+0.05 to +0.12% per year), with spatial heterogeneity reflecting ENSO cycles and potential variations in O3 concentrations. At high latitudes, trends were largely explained by changes in temperature at different altitudes, with positive trends in the center of CO2 and CH4 absorption bands reflecting stratospheric warming. Negative trends in the wings of CO2 bands were due to increased transparency to surface emissions and CO2 increase. Trends in the ν3 O3 band reflected combined effects of stratospheric warming and O3 increase.

The findings demonstrate the potential of IASI-derived SOLR for understanding Earth's climate and its changes. The high spectral sampling allowed unambiguous identification of greenhouse gas concentration changes (CO2, CH4, CFC-11, CFC-12), especially in mid- and upper-tropospheric channels. Trends reflecting temperature changes were mostly linked to atmospheric circulation shifts driven by ENSO and PDO. Regional patterns revealed by H2O channels reflected changes in convergence and subsidence zones. The high spectral resolution is beneficial in identifying compensating errors in climate model outputs. Limitations include uncertainties at high latitudes due to lower observations and potential cloud filter issues. The study highlights the value of long-term, spectrally resolved OLR datasets.

This study, using 10 years of IASI data, successfully identified clear spectral signatures of long-term greenhouse gas concentration changes and their impact on OLR. The high spectral resolution of IASI is crucial for this. The results underscore the importance of high spectral resolution measurements for understanding climate change. Future studies with longer IASI time series and the upcoming IASI-NG will provide further insights.

The study's limitations include potential inaccuracies in cloud filtering at high latitudes, leading to uncertainties in the zonal trends. The relatively short 10-year time period may not fully capture all climate variability. The focus on oceanic regions minimizes the impact of land surface heterogeneity but still influences zonal trends near coasts. The analysis relies on linear trend fitting, which may not capture non-linear changes in the system.