Sea Surface Temperatures and Ice Cover in the Bering Sea

Background

We decided to analyze the sea surface temperatures and ice cover data for the Bering Sea. We used the following coordinates:

We selected this region due to its recent drastic decline in sea ice, as shown by the satellite images below. In addition, climate change has a more severe impact on the sea surface temperatures in the northern hemisphere, as explained in the “Environmental Context” section.

Environmental Context

According to a 2020 study published in Global and Planetary Change, anthropogenic climate change has a long-run heating effect on sea surface temperatures. However, projections of SST are different depending on the region. These thermal changes are more extreme in the Northern Hemisphere  (Ruela et al., 2020) . This influenced our choice to use the Bering Sea as our location for our analysis. 

Motivation

Goal:

Methods

Results

The figure to the right is the graph of sea surface temperature (in Celsius) deviation from the mean. For example, if the value is 1 degree C, this means that the temperature was 1 degree C above the average temperature for that time of year. The data used for this graph is from NCEP Olv2 ¼ SST anom, which can be found on Climate Explorer. The data was formatted into two columns in excel: date and sea surface temperature (SST). The excel document was then uploaded to R as a csv file. Summary data for SST anomalies was mean of 0 C, a minimum of -2.75 C, and a max of 3.67 C. It can be observed that the minimum value was recorded around 1985 and the maximum value was recorded around 2018. This was an early indicator that temperature anomalies are becoming warmer. We then produced this graph, which contains all of the data that was available. As one can see in the graph, the temperature varies, going both above and below the baseline, which is the red line set at 0 degrees Celsius. Additionally, in recent years (2012-2021) the temperature deviation from the mean has been consistently positive, even reaching levels of +3 degrees C from the mean. 

The figure to the right is the graph of sea ice cover, as a percentage change from baseline level (median). For example, if the percent ice change is at -.3, this means that the ice is 30% lower than the median ice level. In order to maintain the same ice level, there would need to be equal deviations below and above median levels. The data used for this graph is from Reynolds v2 ice cover, which can be found on Climate Explorer. The data was formatted into two columns in excel: date and sea ice cover percent. The excel document was then uploaded to R as a csv file. The summary data for the ice coverage deviation from baseline was a mean and median of zero (because this was the baseline), a minimum of -53%, and a maximum of 42%. Because the maximum value was recorded in 1985 and the minimum value was recorded in 2018, this suggests that sea ice coverage fluctuation may have had greater highs in the past and greater lows currently. We then produced this graph, which contains all of the data that was available (1980-2021). In the above graph, we can see that the percent ice change varies throughout time and depending on the time of year. From 2013 to 2021 it can be observed that the sea ice cover has been at all-time highs. It is interesting that this increase in sea ice cover is happening during the same period that the SST deviation from the mean is at all-time highs. By comparing Figures 3 and 4, it can be observed that temperature and sea ice cover levels appear to have an inverse relationship. When temperature deviation is negative - meaning cooler than normal - (for example in 2005-2012) ice levels increase. Through layering the graphs on top of each other, the suspected inverse relationship can be seen visually.

The above graph overlays the two variables of interest: SST and sea ice cover. SST is the gray line and sea ice cover is the red line. It is important to note that these two variables do not have the same units. SST is measured in temperature (in Celsius) deviation from the mean, while sea ice cover is measured as a percentage change from baseline. To compare the data visually, we multiplied the sea ice cover by 6.8 so that it would have the same maximum value as SST, and therefore have the same scale as the SST. As a result, we could observe the data together on one graph. This graph should only be used to observe the visual inverse relationship between the two variables, considering the scale of the sea ice cover has been altered for this graph. 

While this graph is visually effective, the most accurate way to establish an inverse relationship between the two variables is to run a regression analysis of the impact of SST on sea ice cover. 

The table on the right contains the results of a regression of SST on sea ice cover. The data used in the regression is the same data depicted in Figures 3 and 4. The most important values to pay attention to in the table are the R Squared value, the SST Coefficient, t Stat, and P-value. The R Squared value explains how much of the variation in y (sea ice cover) is explained by the variation in x (SST). According to our regression, 11.33 percent of the variation in sea ice cover is explained by the SST. This value shows that although SST explains part of the variation in sea ice cover, many additional variables contribute to declining sea ice cover. In a further study of this lab, it would be interesting to add additional variables to the regression to see what other factors contribute to a reduction in the sea ice cover. The SST Coefficient can be interpreted to mean that a positive deviation of 1 degree Celsius from the mean temperature results in a 4% decrease in sea ice. The t Stat and the P-value show that this variable is highly statistically significant. For a regression with 478 observations, a t Stat with an absolute value above 1.96 would be required to reject the null hypothesis. Our regression has a t statistic with an absolute value of 7.87. Therefore, we can reject the null hypothesis that SST has no effect on sea ice cover. Additionally, a P-value lower than .05 shows significance at the 95% confidence level. Considering that the P-value in the regression is significantly lower than .05, we can be confident that the variable is statistically significant. 

We also ran regression analysis for different periods of time within our data; however, because the values stayed similar and highly significant, we have only included the regression that spans the entire period from 1981 to 2021. 

In summary, this regression analysis showed that the relationship between sea surface temperature and sea ice cover is highly significant. By providing a coefficient value, we can start to discuss the precise impact that the changes of SST have on sea ice cover. 

Discussion

The results above show that sea ice cover and sea surface temperature have an inverse relationship, meaning that as temperature increases, sea ice cover decreases. In the context of the climate, this relationship is considered healthy and normal. As seasons change and temperatures fluctuate, the ice cover on the Bering Sea should be expected to fluctuate. However, this becomes an issue when the temperature fluctuates unevenly. If there is not the same amount of temperature fluctuation above and below the mean, ice cover will either decrease or increase. As seen from 2010 to 2021 in the graph to the right, temperature levels are consistently above the mean, which means the ice cover will decrease without getting the chance to replenish.  

By applying the SST coefficient we found in our regression, we can better understand the future implications. For each 1 degree Celsius, the temperature goes above the monthly mean, ice cover decreases by 4% from the monthly baseline. This means that if the temperature is 1-3 degrees above the mean (which it has been in recent years), sea ice coverage in the Bering sea can decrease as much as 12% from the monthly baseline.

Effects of Melting Sea Ice:

Although the melting sea ice in the Bering Sea will not directly affect rising sea levels, it will reduce the overall albedo of the Earth. Albedo is the proportion of light reflected by a surface  (What Is Albedo? | MyNASAData, n.d.) . Since ice has greater albedo than the ocean, less energy will be reflected into space when the sea ice melts. This will further increase the warming effect of climate change, speeding up the melting of sea ice and resulting in a positive feedback loop. The reduction of sea ice also disrupts marine ecosystems since many forms of wildlife use the ice as their habitat.  (Hancock, n.d.) 

Conclusion

Our research question examined the relationship between sea surface temperature and sea ice coverage. Through graphing the two data sets, we observed possible relationships between SST and sea ice coverage, and by running a regression, we established the inverse relationship between the two variables. 

References

Bering Sea Location. Find Latitude and Longitude. (2022). Retrieved May 9, 2022, from https://www.findlatitudeandlongitude.com/?msclkid=385950f7cfc911ec85191f2494219438

Climate explorer: Select a monthly field. (2022). Retrieved May 9, 2022, from https://climexp.knmi.nl/selectfield_obs2.cgi?id=someone%40somewhere

Hancock, Lorin (n.d.). World Wildlife Fund. Retrieved May 9, 2022, from https://www.worldwildlife.org/pages/why-are-glaciers-and-sea-ice-melting

Hansen, K. (2018). Historic low sea ice in the Bering Sea. Climate Change: Vital Signs of the Planet. Retrieved May 9, 2022, from https://climate.nasa.gov/news/2726/historic-low-sea-ice-in-the-bering-sea?msclkid=5e9ceee0cf6d11ecb0192bf4d5253da1

Stabeno, P. J., & Bell, S. W. (2019). Extreme Conditions in the Bering Sea (2017–2018): Record-Breaking Low Sea-Ice Extent. Geophysical Research Letters, 46(15), 8952–8959. https://doi.org/10.1029/2019GL083816

What is Albedo? | MyNASAData. (n.d.). Retrieved May 9, 2022, from https://mynasadata.larc.nasa.gov/mini-lessonactivity/what-albedo?msclkid=9cb6ef70cfe011ec8f94851f5efda7f2