You are here

Updated Apostle Islands Scenarios

This page is part of the project: Apostle Islands

Revised Climate Scenarios

The climate scenarios developed for the Apostle Islands Scenario Planning Workshop held in 2015 (see Workshop Report) have been revised to better reflect updated knowledge about lake ice and lake levels.  In addition, the updated projections are based on a set of climate models that were designed specifically for the Great Lakes Region, which is an improvement over the previous models used.  Future projections are for the late 21st century.  There are six models to draw information from and all are based on the assumption that there will be increasing climate forcing into the future (RCP8.5).  Using models based on this "higher" amount of forcing is a conservative approach to adaptation planning because planners can explore an upper range of possible impacts - the greater the amount of climate change and variability that can be incorporated into adaptation plans increases the potential for climate resilience.  The table below is broken down by climate variable and separated into columns representing the historical observed trend (which sets the context for thinking about the future), the range of the model projections for end-of-century, and descriptions of each variable for two chosen scenarios.  The future values represent the spatial average around Apostle Islands, as depicted in the map below.

Map of the region used for the Apostle Islands future projections in the Climate Scenario Table

   

Apostle Islands Climate Scenario Table

Climate Variable

Observed Trend

Summary statement about what's already been observed in each season

Range of Model Projections 

Values reported represent the amount of change for the Apostle Island Region by the end-of-century

Soggy Scenario
 

Projections are best on the wettest model (CNRM)

Values reported represent the amount of change for the Apostle Island Region by the end-of-century

Hot and Bothered Scenario

Projections are based on the warmest model (MIROC5)

Values reported represent the amount of change for the Apostle Island Region by the end-of-century

Temperature (by season)

Annual average temperature in Northwestern WI has increased 3.2 deg F compared to the 1951-1980 historical average (source: GLISA Northwestern WI Climate Division Climatology)

Seasonally, the greatest warming has occurred during winter (5.4 deg. F) and spring (3.3 deg. F) over the last 50 years (source: GLISA Northwestern WI Climate Division Climatology Madeline Island Station data) 

Winter = 6 to 11 degrees F
Spring = 4 to 11 degrees F
Summer = 7 to 13 degrees F
Fall = 5 to 12 degrees F

Winter = + 8 degrees F
Spring = + 7 degrees F 
Summer  = + 10 degrees F
Fall  = + 6 degrees F

Winter = + 11 degrees F
Spring = + 9 degrees F 
Summer  = + 13 degrees F
Fall  = + 11 degrees F

Precipitation (by season)

Annual average precipitation in Northwestern WI has increased 3.41% compared to the 1951-1980 historical average (source: GLISA Northwestern WI Climate Division Climatology Madeline Island Station data)

Regionally, fall precipitation has increased the most (20.4%) over the last 57 years. Summer precipitation has declined most (-8.9%) over the period. (source: GLISA Northwestern WI Climate Division Climatology Madeline Island Station data) 

Winter = 1 to 6 inches
Spring = -0.5 to 2 inches
Summer = -3 to 3 inches
Fall = 0.5 to 3 inches

Winter = +  6 inches
Spring = + 1 inches
Summer = + 0.5 inches
Fall = + 3 inches

Winter = + 3 inches
Spring = + 2 inches
Summer = -2 inches
Fall = +2 inches

Heavy precipitation events

The Midwest has seen large increases in extreme precipitation events. (source: NCA Figure 2.17)

1" events: 9 to 31 more days per decade

3" events: 3 to 5 more days per decade

1" events: 23 more days per decade

3" events: 4 more days per decade

1" events: 20 more days per decade

3" events: 5 more days per decade

Date of First Fall Freeze The average length of the frost free season from 1980-2000 was about 120 days (source: Technical Input to NCA) 20 to 28 days later

20 days later

26 days later

Date of Last Spring Freeze The average length of the frost free season from 1980-2000 was about 120 days (source: Technical Input to NCA) 13 to 23 days earlier

20 days earlier

23 days earlier

Growing season length

The growing season increased by about 2 weeks across the Midwest since 1950 mainly due to earlier last spring freezes. (source: NCA)

33 to 48 days longer

Based on the changes in when the first/last freeze occurs, the growing season is expected to start 2-3 weeks earlier in the spring and extend 3-4 weeks longer into fall.

40 days longer

48 days longer

Evapotranspiration  

  Winter = 0 to 2 inches
Spring = 0 to 1 inch
Summer = -1 to 1 inch
Fall = 0.5 to 1 inch

Winter = +1.8 inches
Spring = +0.4 inches
Summer = +0.3 inches
Fall =  +1 inch

 

Winter = +1.6 inches
Spring =  +0.6 inches
Summer =  -0.1 inches
Fall =   +1 inch

Cloud cover fraction

The frequency of clouds in winter (summer) is increased (decreased) over and downwind of Lake Superior (Scott and Huff 1996).  The recent warming of Lake Superior's summer water temperatures has narrowed the the lake-to-air temperature difference creating a less stable atmosphere (more potential for clouds) but only increased winds have been reported in the literature (Lenters et al 2013). Representing clouds well in climate models is an active area of research.  Clouds play an important role in Earth's radiative feedbacks, but the spatial scales at which cloud processes occur is much smaller than what climate models can simulate, so clouds are parameterized (estimated) in the models. As Lake Superior's summer water temperatures continue to warm, the lake will have less  of a stabilizing effect on the lower atmosphere, and there is greater potential for increased cloudiness and precipitation during summer.  As Lake Superior's summer water temperatures continue to warm, the lake will have less  of a stabilizing effect on the lower atmosphere, and there is greater potential for increased cloudiness and precipitation during summer. 

Snowfall

Northwest WI has on average 55" +/- roughly 20" of snowfall each year. During the 1950- 2010 period, the earlier years were characterized by less snowfall and later years characterized by more snowfall on average. (source: U. Wisconsin climate division data)

 

Roughly 25% of winter precipitation on the Bayfield Peninsula is from lake effects.

-41 to -14 cm 

Snowfall downstream of Lake Superior is projected to *increase* by mid-century during late winter (Feb and Mar) and then decrease below historic levels by late-century (source: Notaro Presentation, slide 21).

-36 cm

-41 cm

Days with snowfall

 

10 to 40 less days

36 less days

37 less days

Heavy Lake-Effect Snow

In the region of Apostle Islands, the model simulated on the order of 1 to 2 heavy lake-effect snowfall days per year historically (late 20th century) using the following definition (found also on slide 12):

LOCATION - Grid cell must be located over land and within 100km of one of the Great Lakes’ shorelines.

WIND) For at least 6 hours, the mean 10m wind direction must be off one of the lakes, allowing for sufficient fetch

ICE) The ice cover fraction on the lake, off which the wind is flowing, must be less than 70%

AMOUNT) Local daily snowfall must be at least 10 cm

ENHANCEMENT) Local, near-lake daily snowfall must exceed the mean non-local/continental snowfall, far from the lake, by at least 4cm

Projections around Lake Superior are less certain - by mid-century, models suggest an *increase* in heavy lake-effect snowfall near the Apostle Islands even in the face of winter warming, because air temperatures are still cold enough for snow to form.  By late century, models suggest a slight increase to a slight decrease in the number of days with heavy lake-effect snowfall (source: Notaro Presentation, slide 24). 

-0.5 days per year 

This figure was adopted from Notaro's presentation, slide 24.  Units are the change in Mean Number of Simulated Heavy Lake-Effect Snowfall Days Per Year

-1 day per year

This figure was adopted from Notaro's presentation, slide 24.  Units are the change in Mean Number of Simulated Heavy Lake-Effect Snowfall Days Per Year

Lake ice duration

Near-shore ice cover on Lake Superior has been declining on average 1-2 days/year over the 1994-2003 observational period (Mason et al 2016).  Off-shore ice cover has declined less rapidly (~1 day/year) over the same period.  Lake Superior ice cover has undergone a dramatic shift from a period of more ice (60 days on average between 1973-1998) to a period of less ice (30 days on average from 1998-present).  However, there is still strong year-to-year variability with two recent years (2013/14  and 2014/15) reaching near 100% coverage at its maximum on Lake Superior. Although the models project a compressed lake ice season and less ice in the region of Apostle Islands, indicators from the past would suggest a less certain future.  High ice years are still possible under the right conditions (particularly high fall evaporation rates off Lake Superior, which is driven by 1) large temperature contrast between water and air, 2) low humidity, and 3) high winds)   (source: Lenters et al 2013). Although the models suggest considerably less ice on average in the future, year-to-year variability will likely remain high and the potential for high-ice years is not eliminated under the right conditions.   Although the models suggest considerably less ice on average in the future, year-to-year variability will likely remain high and the potential for high-ice years is not eliminated under the right conditions. 

Lake ice extent

Lake ice has and will continue to form first near shore and in protected areas and last off-shore.  Ice extent has been decreasing since the 1970s - annual mean ice cover (%) has a significant negative trend for all lakes.  (Wang et al 2012).         Ice becomes primarily restricted to the northern shore of Lake Superior (source: Notaro et al 2015, Figure 9).

Ice becomes primarily restricted to the northern shore of Lake Superior (source: Notaro et al 2015, Figure 9).

When ice forms, it will still form in shallow, protected areas first.

Ice becomes primarily restricted to the northern shore of Lake Superior (source: Notaro et al 2015, Figure 9).

When ice forms, it will still form in shallow, protected areas first.

Lake ice seasonality (does the timing of lake ice formation change?)

Lake Superior typically starts to form ice in late December/early January, peaks in Feb/March, and is close to ice-free by May. The ice season is projected to compress to the months of Jan-March (source: Notaro et al 2015, Figure 9).

The ice season is projected to compress to the months of Jan-March.  There are not strong differences in ice cover between the two models used in these scenarios.  

(source: Notaro et al 2015, Figure 9)

The ice season is projected to compress to the months of Jan-March.  There are not strong differences in ice cover between the two models used in these scenarios.  

(source: Notaro et al 2015, Figure 9)

Lake temperature (by season?)

Lake Superior summer surface water temperatures have risen approximately 6 deg (F) over the last 100 years with most of the warming occurring during the last three decades. Water temperatures have varied up to 18 deg (F) during summer from year-to-year and by up to 10 deg (F) over multiple winters.(source: Great Lakes Statistics)

Winter surface water temperatures are projected to increase only slightly, whereas spring and summer temperatures are projected to increase more.  The monthly timing of when this increased warming occurs is less certain.  The models are biased towards an earlier spring stratification season, so although they suggest Lake Superior's surface water is projected to warm the most during spring, this is earlier than historical observations of rapid mid-to-late summer warming (source: Notaro et al, 2015, Figure 5).

12-18 degrees F warmer surface temperatures during spring/summer

Winter: + 2 degrees F

Summer: +12 degrees F

The models have a bias towards earlier stratification, so although the models suggest this summer warming will occur during May/June, historical observations indicate Lake Superior's greatest warming has occurred mid-to-late summer so later timing may be more likely.  

(source: Notaro et al, 2015, Figure 5)

Winter: + 2 degrees F

Summer: +18 degrees F

The models have a bias towards earlier stratification, so although the models suggest this summer warming will occur during May/June, historical observations indicate Lake Superior's greatest warming has occurred mid-to-late summer so later timing may be more likely.  

(source: Notaro et al, 2015, Figure 5). 

Lake levels

Lake Levels primarily depend on the balance between over-lake precipitation, over-lake evaporation, and the horizontal (landscape) flow of water into/out of the lake (these are the net basin supply components).

Figure: Net Basin Supply Components for Lake Superior (click to enlarge)

Lake Superior lake levels show a slight delay (about a month) in response to changes in the difference between precipitation and evaporation. As there are net gains (precipitation > evaporation) lake levels increase and vice versa. 

Lake Superior historical high: 603.4 ft above sea level (1 feet above present) 

Lake Superior historical low: 599.5 ft above sea level (-3 feet below present) (source:GLERL Lake Level Observations) 
 
Intra-annual variability is about 1-2 feet (Great Lakes Water Level Dashboard) 
 
Lows occur in spring (Mar/Apr) 
 
Highs occur in late summer/early fall (Aug-Oct) 
 
Lake Superior water levels show strong evidence for non-random trends. Levels increased from 1860-1980, then experienced a 30 cm decrease from 1980- 2007. Since May of 2014 monthly mean water levels have been at or above the long term (1918-2015) record, with the exception of March 2017. There is an earlier shift to the spring maximum and slight decrease in net basin supply.

-97 mm to 134 mm

Lake levels are very complex to simulate and projections have many sources of error.  These projections are based on some of the best available models, but they are still subject to considerable bias.  The two models (CNRM and MIROC5) used in these scenarios capture the seasonal cycle of over-lake precipitation well, but over-land precipitation that runs off into the lakes during  winter is underestimated.  Lake evaporation is the most difficult components to simulate, which resulted in an "unreasonable seasonal cycle" but a more reasonable annual measure (Notaro et al, 2015).  The errors in the representation of the annual cycle for net basin supply components make it difficult to project future changes in annual variability, so changes are discussed only in terms of annual means.

Lake Superior levels rise 134 mm on average (source: Notaro et al, 2015)

The greatest increase in precipitation for the "soggy" scenario occurs during winter, which is also when lake evaporation is high, so these two forces will compete against each other in determining lake levels.  If the precipitation comes in the form of snow, runoff-the third main component for determining lake levels- may be delayed until the snow melts.  If precipitation comes in the form of rain, which is more likely by late-century, runoff will play a larger role during winter (as opposed to spring), especially if the ground is frozen or semi-frozen.    

Lake Superior levels decline -97 mm on average (source: Notaro et al, 2015)

The high magnitude of summertime warming in the "hot and bothered" scenario may contribute to large lake level drops during summer, especially when coupled to a drop in summer precipitation.  However, winter lake levels are more complex.  If the precipitation comes in the form of snow, runoff-the third main component for determining lake levels- may be delayed until the snow melts.  If winter precipitation comes in the form of rain, which is more likely by late-century, runoff will play a larger role during winter (as opposed to spring), especially if the ground is frozen or semi-frozen.       

Wind speed 

Lake Superior has seen a 5% increase per decade in surface wind speeds measured by buoys from 1985-2008. Wind is a very difficult variable to simulate in the models with much credibility.  The historical increase in wind speeds over Lake Superior have been partly attributed to the destabilizing nature of the Lake as its summer surface waters warm in relation to the air temperature.  Summer lake temperatures are projected to continue warming in the future, so there is potential for even greater increases in wind events over and near the lake.    

Wind direction 

       

Arctic Oscillation/ cold air outbreaks

It is difficult to predict the mode of the AO and one extreme negative mode can be followed by an extreme positive mode. The modes determine the type of weather that is experienced: warmer and drier air (+) versus cooler and wetter air (- ). The AO is primarily a wintertime variable (DJFM).

The Great Lakes tend to have lower (higher) ice cover during the positive (negative) NAO. The negative phase of the AO is more strongly correlated with positive snowfall  anomalies over North America than correlations of negative anomalies with a positive AO mode. In general, the AO is more strongly correlated with snowfall over Eurasia than North America

  The models have not been explicitly analyzed for their AO mode.  This scenario was initially based on the assumption of more predominantly negative phase AO, leading to cooler winters on average.  However, the wettest model used to represent a "soggy" scenario indicates winter warming of 8 degrees F, only second to summer warming of 10 degrees F.   The models have not been explicitly analyzed for their AO mode.  This scenario was initially based on the assumption of more predominantly positive phase AO, leading to warmer winters on average.  The warmest model used to represent a "hot" scenario indicates winter warming of 11 degrees F.  

 

Scenarios Narrative Revisions

A set of narratives (scenarios), based on future climate change projections, were developed for the Park that more clearly outlined potential climate related impacts.  The update to the state of knowledge for the climate information requires a subsequent update to the narratives of climate impacts.  Below are the two original impact scenarios summarized with revision statements pertaining to specific parts of the scenarios.  The Climate Scenario Table also reflects the revisions of the climate information.

 

Scenario: Hot and Bothered

 A world of higher temperatures and lower precipitation

Original Framework:

  1. Warm winters predominate
  2. Hot and dry summer
  3. Stable lake ice extremely rare1
  4. Lake level decreases1
  5. Bothered: major disturbance complexes (wind and fire) hammer ecosystems

Main Impacts (from Original Narrative, page 10)

  1. More land is converted to agriculture2 - L. Superior water quality declines from increased nutrient-laden runoff
  2. Significant changes in ecosystem dynamics and species ranges
  3. Loss of northern species
  4. Terrestrial and aquatic invasives become more prevalent
  5. Lake dynamics are affected, leading to fish range and depth changes
  6. Longer warm-season for Park visitor activities

Revisions:

1Warmer summer temperatures and lower precipitation amounts would lead to lower summertime lake levels, but wintertime lake levels and lake ice would not necessarily decline.  Lake evaporation during fall is a major factor for lake ice development.  This scenario calls for lower summertime humidity (from drier conditions) and the chance of increased winds late summer/early fall, which increase the potential for higher evaporation rates leading to increased ice cover.  However, ice formation requires below freezing temperatures, so the amount of winter warming is important.  In the observational record, two nearby weather stations indicate fewer days per year below freezing (Duluth, MN, not tested for significance) and no trend (Ironwood, MI).  A possible explanation for this is that Lake Superior is modifying air temperatures at downwind locations preventing them from rising like in the rest of the region.   The timing of warmer (above freezing) winter days also matters - this scenario may want to consider a shortening of the winter season or intermittent (day to week-long) warm periods, or both.  These would all have different effects on ice cover/lake levels. 

2Although the scenario calls for increased land-use conversion from natural landscape to agriculture, this may not be practical under a lower precipitation scenario as agriculture can be water intensive 

 

Scenario: Soggy

A wetter scenario where lake levels rise

Original Framework:

  1. Cool winters predominate (-AO)
  2. Mild and moist summers
  3. Stable lake ice regularly forms1
  4. Lake levels increase

Main Impacts (from Original Narrative, page 9)

  1. Great erosion of cliffscapes and sandscapes
  2. Water quality declines from increased runoff
  3. Access to smaller beaches becomes limited as lake levels rise - increased trampling of sensitive dune vegetation
  4. Species range shifts
  5. Increased visitor numbers to winter ice caves1

Revisions

1Cold winters provide a clear potential for lake ice formation, but there is historically strong year-to-year variability of ice cover which will likely persist into the future under this scenario.  So, years of very high ice cover may be followed by years with low ice cover, and vise versa.  In addition, a high ice year actually has the potential to limit the amount of ice that forms the following year (see Figure 6 of Lenters et al 2013).  Even in a scenario where there are more high ice cover years, this does not necessary translate to stable ice formation.  Here are some of the factors that affect the stability of the ice caves:

  1. Duration of ice cover (park institutes a threshold that the 10-day average ice cover must exceed)
  2. Strong winds causing ice breakup
  3. Geographic origin of the ice (on-site versus floated in from offshore)
  4. Others?