I’m very pleased to report our new paper is now in open discussion in The Cryosphere. The paper presents a new model for predicting the spectral bioalbedo of snow and ice, which confirms that ice algae on ice surfaces can change its colour and by doing so enhance its melt rate (“bioalbedo”). We also used the model to critique the techniques used to measure bioalbedo in the field. The model is based on the SNow ICe and Atmosphere Radiative model (SNICAR), but adapted to interface with a mixing model for pigments in algal cells. We refer to the coupled models as BioSNICAR.
The model uses Mie theory to work out the optical properties of individual algal cells with refractive indices calculated using a pigment mixing model. The user can decide how much of each pigment the cell contains, the cell size, the biomass concentration in each of n vertical layers, the snow/ice optical properties, angle and spectral distribution of incoming sunlight and the mass concentration, optical properties and distribution of inorganic impurities including mineral dusts and black carbon (soot). From this information, the model predicts the albedo of the surface for each wavelength in the solar spectrum. This can then be used to inform an energy balance model to see how much melt results from changes to any of the input values, including growth or pigmentation of algae.
The model shows that smaller cells with photoprotective pigments have the greatest albedo-reducing effect. The model experiments suggest that in most cases algal cells have a greater albedo-reducing effect than mineral dusts (depending upon optical properties) but less than soot.
As well as making predictions about albedo change, the modelling is useful for designing field experiments, as it can quantify the error resulting from certain practises, such as using devices with limited wavelength ranges, or neglecting to characterise the vertical distribution of cells. I’ll cover this in some further posts. The most important thing is metadata collection, since standardising this enables the measurement conditions to be as transparent as possible and encourages complementarity between different projects. Importantly, following a protocol for albedo measurements and collecting sufficient metadata will make it easier to couple ground measurements to satellite data. We outline two key procedures: hemispheric albedo measurement, and hemispherical-conical reflectance factor measurement. To accompany the discussion in our paper, we’ve produced some metadata collection sheets that might be useful to other researchers making albedo measurements in the field (download here: metadata sheets) and made our code and data available in an open repository.
On Wednesday last week I traveled to the University of Bristol to give a seminar at the Centre for Glaciology. I presented a new physical model for the spectral albedo of ice with algal growth, along with some field data from 2016. Preparing for the talk, discussions with fellow researchers and insightful questions in the Q&A all reinforced some key issues that remain unresolved in bioalbedo studies – fundamental questions that have proven difficult to answer. First, do algae darken ice? Second, are they widespread enough to have ice sheet scale impact?
The answer to the first question is a clear yes. That dark materials contaminating an ice surface lower its albedo is not surprising. However, the crucial follow-up question is “by how much?” and this is much more challenging to answer; however, physical modelling provides a clear framework for determining the impact of an algal bloom on ice albedo. With sufficient information from empirical lab and field studies, we can quantify the bioalbedo effect and characterize its variability over space and time.
Standing in the so called ‘dark zone’ on the Greenland ice sheet, the answer to the second question also seems to be a clear ‘yes’. The ice surface is dark for as far as the eye can see in all directions, and wherever ice is sampled and examined under the microscope, it is found to be teeming with algal cells. However, what is visible from standing in the dark zone and what is important at the ice-sheet scale are two different things. To quantify algal coverage over the ice sheet we need to be able to detect blooms remotely, ideally from space using spectral data from satellites. This method of mapping is routine for terrestrial vegetation and algal blooms in the ocean; however, there are specific challenges to doing the same for algal blooms on ice.
The most common way to identify photosynthetic life in satellite reflectance data is to apply the ‘red-edge’ biomarker. This refers to a sharp rise in the reflectance spectrum of a surface due to vegetation because of efficient absorption by chlorophyll and very little absorption at near-infrared wavelengths (which has been suggested to be the result of evolutionary pressure to avoid overheating, or alternatively a side-effect of the evolution of cell-spacing in early aqueous plants). This has also been proposed as a spectral feature that could be used to map photosynthetic life on other planets. Amazingly, the red-edge has been detected in Earth-shine (light that has reflected multiple times between the Earth and moon and faintly illuminates the dark part of crescent moons), which provides a hemisphere-integrated reflectance signal for our planet. Since ice algae is photosynthetic, it follows that it could be mapped using the red-edge biomarker.
However, there are several issues that may complicate matters and increase the risk of a ‘false-positive’ result from applying the red-edge biomarker to Earth’s ice. These are
1. Carotenoids obscuring chlorophyll
Ice algae produce photoprotective carotenoid pigments that absorb over a wide range of visible wavelengths. They have a strong but broad absorption spectrum (which is why they protect the algae from ‘sunburn’). This could obscure the chlorophyll ‘bump’ near 500 nm and make interpretation of the red-edge more difficult. While the carotenoids themselves might provide a diagnostic reflectance spectrum, they too are hard to distinguish from other reflectance-reducers on ice.
Dust also absorbs strongly in visible wavelengths and also reflects effectively at red wavelengths, leading to a pseudo-red-edge feature in the reflectance spectrum. The precise shape of the reflectance spectrum varies for each mineral, and actually no mineral exactly replicates the vegetation red-edge signal. However, dust on ice is not composed of a single mineral, and both the dust and any biological impurities are mixed together and set in a complex ice matrix with its own reflectance spectra. It is feasible that the slope of the red-edge might be diagnostic of biological impurities, but this requires truly hyperspectral (i.e. spectral resolution of 1-2 nm) and will not be achievable using current satellite data. These issues combined lead to a high chance of a false positive result from the application of the red-edge biomarker to ice surfaces. This is especially important for explaining the ‘dark ice’ on the Greenland ice sheet since the two leading hypotheses are biological growth and outcropping dust.
3. Spatial integration reducing signal
An additional important issue is that any biomarker signal will be diluted by spatial integration over the viewing footprint of a satellite sensor. The presence of clean ice, ponded water, cryoconite, abiotic impurities or roughness elements will decrease the signal to noise ratio, probably further obscuring the red-edge signal.
These issues do not necessarily prohibit the use of the red-edge biomarker, but they do necessitate robust correction for abiotic impurities (particularly dusts) and rigorous ground truthing to validate the application of the biomarker to satellite data. There was a fascinating discussion in the planetary sciences in the early-mid twentieth century surrounding a reflectance signal detected on Mars which spread to cover wider areas each spring. This was proposed to be evidence of Martian plant life (e.g. Lowell, 1911); however, this hypothesis was discredited by further spectral analysis (Millman, 1939) and was then shown to be due to blowing dusts (Sagan and Pollack, 1969).
While physical modelling paired with ground reflectance measurements and sample analysis can answer the first fundamental question (do algae darken ice?), the second question (are they widespread enough to have an albedo-lowering effect at the ice sheet scale?) may prove challenging to answer robustly.