Satellite observations of cloud regime development: the role of aerosol processes

Edward Gryspeerdt, Philip Stier, Daniel G. Partridge

TL;DR

This study uses a novel regime-based approach with temporally spaced satellite data to disentangle aerosol effects on cloud development from meteorological influences. By accounting for the aerosol-cloud fraction relationship, they find that simple "snapshot" correlations can overestimate aerosol impacts on cloud fraction by at least a factor of two, suggesting aerosols can increase stratocumulus persistence and invigorate deep convection.

What Problem Does This Paper Address?

Understanding how atmospheric aerosols influence clouds and, consequently, Earth's climate is a major challenge in climate modeling. While correlations between aerosol properties (like aerosol optical depth, AOD) and cloud properties (like cloud fraction, CF) have been observed, it's been difficult to determine if these are direct aerosol effects or if they are driven by other factors. Specifically, meteorological conditions and retrieval biases (like cloud contamination of aerosol measurements) can create spurious correlations. Previous studies often relied on single satellite overpasses ("snapshot" studies), making it hard to capture the dynamic evolution of clouds and disentangle aerosol impacts from meteorological covariations. This paper aims to address this by observing cloud development over time in different aerosol environments.

Methods

The authors employ a "regime-based approach" to classify cloud types. They use k-means clustering on Cloud-Top-Pressure (CTP) and Cloud-Optical-Depth (COD) histograms from MODIS data to define distinct cloud regimes. The core innovation is using temporally spaced satellite retrievals (Terra and Aqua MODIS, with ~3-hour separation) to observe transitions between these cloud regimes.

To isolate aerosol effects, they meticulously account for confounding factors:

  • Cloud Fraction (CF) Influence: They ensure that for a given cloud regime, the distribution of cloud fraction is the same for both high and low aerosol populations at the start of the observation period. This is crucial because aerosol retrieval can be biased by CF.
  • Meteorological Covariation: They extend this CF-matching approach to other meteorological variables (like relative humidity, 10m windspeed, static stability, and vertical velocity) that are known to influence cloud and aerosol properties. This helps remove spurious correlations arising from these factors.
  • Regime Transitions: By studying how often clouds transition between regimes over the ~3-hour period, they can infer aerosol influences on cloud development, which are less susceptible to the initial AI-CF relationship.

Datasets

  • MODIS (Moderate Resolution Imaging Spectroradiometer) Data: Used for cloud properties (CF, CTP, COD) and aerosol index (AI) from both Terra and Aqua satellites (Collection 5.1).
  • ECMWF ERA-Interim Reanalysis Data: Used for meteorological variables (850 hPa relative humidity, 10m wind speed, 500 hPa vertical velocity, and low troposphere static stability - LTSS).
  • Global Precipitation Climatology Project (GPCP) 1° product: Used for annual mean rain rates in the tropics.
  • ECMWF Monitoring Atmospheric Composition and Climate (MACC) AOD product: Used as an alternative aerosol dataset to assess potential sampling errors.
  • ISCCP (International Satellite Cloud Climatology Project) Data: Used for comparison with MODIS results, employing geostationary satellite data.
  • GlobAerosol AOD retrieval from AATSR instrument: Also used for comparison.

Key Insights and Ideas

  • "Snapshot" correlations are misleading: The study strongly suggests that simple correlations between aerosol and cloud properties from single satellite passes significantly overestimate aerosol effects on cloud fraction.
  • Aerosol impacts on cloud development are key: By observing how clouds evolve over a few hours, rather than just looking at a single point in time, the authors can better isolate aerosol influences.
  • Aerosols might stabilize stratocumulus: An increase in aerosol index (AI) is linked to more transitions into the stratocumulus regime over the ocean, suggesting aerosols help maintain these clouds.
  • Aerosols invigorate convection over land: Higher AI is associated with more transitions into the deep convective regime over land, supporting the "aerosol invigoration" hypothesis.
  • Meteorology matters, especially stability and wind: The observed aerosol effects on cloud development are significantly modulated by meteorological factors like low troposphere static stability and 10m windspeed.

Key Results

  • Simple AI-CF correlations overestimate aerosol effects on cloud fraction by at least a factor of two.
  • Increased AI is associated with:
  • An increase in stratocumulus regime persistence over oceans.
  • An increase in transitions into the deep convective regime over land.
  • Changes in transitions from shallow cumulus, with the strength of these changes dependent on LTSS and 10m windspeed.
  • The study successfully reduces the influence of meteorological covariations and CF-related biases on aerosol retrievals.

Limitations and Future Work

  • Upper Bound: The authors acknowledge that while they've reduced errors from meteorological factors and CF, complete removal of meteorological covariation is difficult, meaning their results likely represent an upper bound on aerosol effects.
  • Advection: While they investigate advection and find its impact on their results to be small, it's a factor that could still introduce some uncertainty.
  • Wet Scavenging: The study notes that wet scavenging processes, particularly in high CF/high precipitation scenes, can influence aerosol properties and complicate comparisons between different aerosol retrieval methods (e.g., MODIS vs. MACC AOD).
  • Future Work: The paper suggests further work is needed to better understand the relationship between developmental studies (like this one) and "snapshot" studies to leverage the strengths of both. It also highlights the need for improved accounting for meteorological covariations.

Summary updated

Abstract

Many different interactions between aerosols and clouds have been postulated, based on correlations between satellite retrieved aerosol and cloud properties. Previous studies highlighted the importance of meteorological covariations to the observed correlations. In this work, we make use of multiple temporally-spaced satellite retrievals to observe the development of cloud regimes. The observation of cloud regime development allows us to account for the influences of cloud fraction (CF) and meteorological factors on the aerosol retrieval. By accounting for the aerosol index (AI)-CF relationship, we reduce the influence of meteorological correlations compared to "snapshot" studies, finding that simple correlations overestimate any aerosol effect on CF by at least a factor of two. We find an increased occurrence of transitions into the stratocumulus regime over ocean with increases in MODIS AI, consistent with the hypothesis that aerosols increase stratocumulus persistence. We also observe an increase in transitions into the deep convective regime over land, consistent with the aerosol invigoration hypothesis. We find changes in the transitions from the shallow cumulus regime in different aerosol environments. The strength of these changes is strongly dependent on Low Troposphere Static Stability and 10 m windspeed, but less so on other meteorological factors. Whilst we have reduced the error due to meteorological and CF effects on the aerosol retrieval, meteorological covariation with the cloud and aerosol properties is harder to remove, so these results likely represent an upper bound on the effect of aerosols on cloud development and CF.
BibTeX
@inproceedings{gryspeerdt2014satellite,
  author = {Edward Gryspeerdt and Philip Stier and Daniel G. Partridge},
  title = {{Satellite observations of cloud regime development: the role of aerosol processes}},
  booktitle = {Atmospheric},
  year = {2014},
  doi = {10.5194/acp-14-1141-2014},
}
Figure