Estimating Underwater Light Regime under Spatially Heterogeneous Sea Ice in the Arctic

# Estimating Underwater Light Regime under Spatially Heterogeneous Sea Ice in the Arctic
## ⚔<br/>GreenEdge legacy meeting
### P. Massicotte, G. Bécu, S. Lambert-Girard, E. Leymarie and M. Babin
### 2019-11-06 (Nice, France)

---

# The vertical distribution of underwater light in open water

An overview

---

# The importance of underwater light

An adequate description of the underwater light regime is mandatory to understand energy fluxes in aquatic ecosystems.

- Primary production

- Photochemical reactions (photo-degradation)

- Energy budget in the water column

---

# Measuring light in open water

In open water, downwelling irradiance, `$E_d$`, decreases exponentially with increasing depth.

---

# Mathematical formulation

**Assuming an optically homogeneous water column**, the decrease of light with increasing depth can be modeled as follow:

$$
E_d = E_d{(0^-)}e^{-K_d(z)z}
$$

- `$K_d(z)~[m^{-1}]$` is the vertical diffuse attenuation coefficient describing the rate at which light decreases with increasing depth.

- An important metric used by biologists and to parameter models (**need for precise estimates**).

---

# Vertical light distribution in open water

In **optically homogeneous water**, downward irradiance follows quite well a monotonically exponential decrease with depth.

---

# Vertical light distribution in ice-covered water

Current challenges

---

# Vertical light distribution in ice-covered water

Due to **spatial horizontal heterogeneity**, measuring vertical light profiles under ice cover presents considerable challenges in comparison to open water.

<center>
<img src="img/IMG_1093_v2.jpg" height="300" />
<figcaption>Photo: Joannie Ferland</figcaption>
</center>

---

# Vertical light distribution in ice-covered water

In ice-covered water, **subsurface light maxima at a depth of around 10 m** are visible between 400 and 560 nm.

---

# Estimating `$K_d$` underice

Estimating `$K_{Ed}$` under ice represents a considerable challenge.

---

# The main objective

<b><i>Propose a new method to estimate average irradiance profile, `$\overline{E_d}(z)$`, over a large spatially heterogeneous area.</i></b>

From a given `$E_d(0^-)$` measured or modelled right under the ice sheet.

---

# Upward radiance: A New Hope

One promising way is to use **upwelling radiance, `$L_u$`,** which is far less influenced by the surface spatial heterogeneity.

---

# Upward radiance: A New Hope

Upward radiance is less influenced by the surface spatial heterogeneity (**no subsurface maximum**).

---

# Why upward radiance?

**Below 10 meters**, the downward and upward attenuation coefficients are correlated.

---

# What we did

- Show that `$K_{Lu}$` can be used as a proxy to estimate the mean `$\overline{K_{d}}$` in the water column.

- **In other words, is it better to use `$K_{Lu}$` rather then `$K_{Ed}$`?**

Based on:
  
- *in-situ* C-OPS measurements
  
- Monte-Carlo simulations (using SimulO, a software developped by Edouard Leymarie)

---

# In-situ underwater light measurements at the ice camp

---

# Sampling at the ice camp

Between 20 April and 27 July 2016 (water depth: 360 m).

---

# C-OPS measurements

- A total of 83 vertical light profiles were acquired during the 2016 field campaigns.

- Equipped with both **downward plane irradiance** `$E_d(z)$` and **upward radiance** `$L_u(z)$` radiometers.

- At 19 wavelengths between 380 and 875 nm.

---

# C-OPS measurements

---

# C-OPS data processing (smoothing)

Data were smoothed locally (loess) on a depth grid with a higher vertical resolution at the surface.

---

# C-OPS: quality of the measurements

The C-OPS has an exceptional signal-to-noise ratio (SNR). Note that this is for upwelling radiance.

---

# Contribution of the red wavelengths to PAR

Even if the signal gets a bit noisier in the red, this spectral region has a reduced contribution to the PAR.

---

# 3D Monte Carlo numerical simulations of radiative transfer

---

# 3D Monte Carlo numerical simulations (SimulO)

Propagate light under a sea-ice surface *(50-meters radius)* containing *a 5-meters radius melt pond*. **The melt pond represents 1% of the surface area.**

---

# Main results

---

# Simulating the under-ice light field

The under-ice radiance light field `$L_u$` **is less influenced by the melt pond** compared to the irradiance light field `$E_u$`.

---

# Simulating the under-ice light field

A greater dispersion around the reference profile occurred when using `$K_d$` compared to those generated with `$K_{Lu}$`.

---

# The mean relative errors

- Decrease as measurements are made away from the melt pond.
- **Lower by approximately a factor of two** when using `$K_{Lu}$` (~7%) compared to `$K_{Ed}$` (~12%).

---

# Take home messages

---

# Take home messages

1. Our results show that under spatially heterogeneous sea ice at the surface (and for a homogeneous water column), the average irradiance profile, `$\overline{E_d}(z)$`, is well reproduced by a single exponential function.

2. `$K_{Lu}$`, which is up to two times less influenced by a heterogeneous incident light field than `$K_{Ed}$` in the vicinity of a melt pond, can be used as a proxy to estimate `$E_d(z)$` in the water column.

---

<br><SPAN STYLE="color: #B2CCE5; font-size: 50pt";><b>Thank you!</b></SPAN><br><br>

Massicotte, P., Bécu, G., Lambert-Girard, S., Leymarie, E., & Babin, M. (2018). Estimating Underwater Light Regime under Spatially Heterogeneous Sea Ice in the Arctic. Applied Sciences, 8(12), 2693. https://doi.org/10.3390/app8122693