Seascape Projects

The forecast again has shifted to the left, and we're now essentially predicting a normal season.  The best analog is 2008.  

I'm intrigued by how the season is progressing in the water.  My sense is that the ocean is warming up faster than on shore, especially below the surface.  I think this might be a result of the series of clear, bright days we've had recently.  The most important component to heating in the ocean is the penetration of light into the water.  On a clear day, especially if the winds are low, there are a lot of photons going into the ocean.  The temperature of the air above the water is a lesser factor, and can really just pull heat from the surface of the ocean.  The forecast for the next few days is for sunny and calm conditions.  My bet is that by the middle of next week, our forecast will be for an slightly early start.

The warming at Buoy E has continued and the temperatures there are now pretty much at the long-term average.  This means that our forecast is for a season that is only delayed two days (now projected for July 1):

As before, the start dates from previous years are plotted at the top, with blue indicating cold years and red indicating warm years.  The diamonds represent a forecast.  The center of the diamond is the forecasted value and the width is the 95% confidence interval.  Each diamond represents a forecast made using data centered on the date on the left.  

Note that the width of the diamond is smaller than before.  This is due to both higher skill for this week compared to previous and to the fact that our forecast is close to the longer-term average.

I don't have a good sense for weather the warming will continue or whether we will remain close to the long term average.  NOAA's Climate Prediction Center is forecasting below average air temperatures over the next few weeks, so perhaps we'll hold steady with a slightly delayed season.

Seascape has a long tradition of trying to develop ecosystem forecasting.  It started with copepods and whales, and last year, we made our first stab at forecasting lobsters, specifically, the date when the Maine lobster landings begin their rapid summer increase.  Our  forecasts were motivated by the extremely warm year in 2012 and the havoc that it caused in the Maine lobster fishery.  

Here is our first attempt at a forecast for the 2014 fishing season:

Just a reminder that this is the forecast for the start of the "high catch period" when landings really begin to increase.  They are based on the temperature at 20m at NERACOOS Buoy E operated by the University of Maine Physical Oceanography Group.  Full details are at the end of the post.

There's a lot on this figure (we're still experimenting with the best way to present these forecasts).  The blue diamonds are the forecasts, with the width of the diamond representing the 95% confidence interval.  Each diamond is the forecast generated on a specific date.  The forecast at the top is the most current, the one on the bottom was made using data in early February.  Our current forecast is for the landings to start picking up around the Fourth of July, about 5 days later than usual (usual is indicated by the black line).

The sticks and the text at the top indicate the start date from the past.  Our current forecast is for the timing of the 2014 season to look a lot like 2003, which was a pretty cold year (indicated by its blue color).  Years that were warmer than normal are indicated in red, years that are near the average temperatures are in black.  Gray indicates that buoy temperatures were not available in those years.

You'll notice that we were forecasting an even later start of the season (about 8 days delayed) back in the end of March.  This coincided with the coldest temperature anomalies. The warming over the last few weeks has been greater than normal, and water temperatures are almost to the 2002-2011 average for this time of year.  This has caused the forecasts to move back towards the center.  I would not be surprised if this trend continues and that our forecast at the end of April is close to the middle.  

We will try to update these forecasts over the next few weeks.  We also have a proposal pending at NASA that would allow us to improve and expand these forecasts in terms of what we forecast (timing and volume of landings by zone, hard/soft-shell mix) and how early we issue them.

We downloaded temperature data from the 20m sensor on NERACOOS buoy E.  The best place to look at the current conditions at E is through the NERACOOS Climatology Tool, but if you want the data, UMaine provides netCDF files.  We created a time series of 8-day average temperatures.  The dates on the left side of the figure are the middle of the 8d period used to generate the forecasts.

For each 8d period, we used historical temperature and landings data to build a simple linear model relating temperature on that date to the date when landings begin to increase.  We generated the lobster landings "start date" using monthly landings for Maine. For each year, we divide the monthly landings by the total landings for the year, and then use linear interpolation to find the point where normalized landings increase to 0.08.  These are the dates indicated by the sticks at the top.

I'm at the NCSE "Climate Solutions Conference" in Washington, DC this week.  I was invited to talk about the lessons we can learn from the 2012 ocean heat wave in a session on "Managing Fisheries Under Climate Change."  For my talk, I was curious whether the warming we've seen in the Gulf of Maine is unique.  Short answers: it is very unique: since 2004, the Gulf of Maine has warmed faster than 99.85% of the global ocean.

To get to this number, I grabbed the global sea surface temperature (SST) data set from NOAA's   optimally interpolated product.  I removed the mean annual cycle (1982-2011) to produce daily anomalies.  I then selected 2,000 point at random (shown over the mean SST from June):

At each point, I computed the linear trend over the last 10 years (January 2004-September 2013).  At this relatively short time scale, I would expect to see a lot of variability--a lot of places warming, but also a lot of places cooling, and this is what I found. The mean trend is 0.006°C/yr and there are slightly more points that are warming (1024 points) than cooling (976).  

The distribution of trends is pretty much normal (in the statistical sense):

but the Gulf of Maine is decidedly abnormal.  The Gulf is warming at 0.23°C/yr (see my previous post), and from my 2000 randomly selected points, I found only four (FOUR!) that were warming faster (red stars on the map).    Turns out one of them was from the Gulf of Maine (go figure), and that the other points were from the Kuroshio extension region northeast of Japan.  This makes some sense oceanographically and is probably driven by a northward shift in the Kuroshio (the Pacific's Gulf Stream). 

According to this analysis, the warming in the Gulf is remarkable, and frankly, a little scary.  However, it is important to remember that this is just a statistical analysis.  At some point, the trend will slow, but we don't know when and by how much.  There is clearly more work to be done to understand the mechanisms that are driving the current trend.   And, there is also an opportunity to use the Gulf of Maine to understand how an ecosystem responds to rapid climate change.

This is essentially the same figure we used to show the 2012 event, but extended through early September (the last available AVHRR OI field).  As you can see, 2013 was cooler than 2012, but was still a very warm year. The recent warming trend has continued (it's actually a little stronger), and 2012 still stands out as unusual.  Of course, this doesn't include the impact of a few very cold days recently (I'm talking about you, polar vortex), nor does it include the very mild weather we're experiencing now.  Temperatures in the ocean change much more slowly than those on land, and water temperatures are a better indicator of climate trends than the weather on a given day.  I hope to update the SST animations in a couple of weeks.