Extending regression lines beyond data span

Hi everyone,

I'm working on a code that would predict Arctic ice coverage during the rest of the 21st century, like so:

%load ice coverage data
data = load('Ice_North.txt');

%define variables
year = data(:,1);
data_month(:,1:12) = data(:,2:13); %a new matrix with monthly values only

%calculate the yearly average time series
yearData = nanmean(data_month,2); 

%plot yearly averaged time series
figure(1)
plot(year, yearData, 'LineWidth',2,'Color','b')
xlabel('Year')
ylabel('Sea ice coverage (milions of km^2)')
grid on
title('Annual average Arctic sea ice coverage')


%regression analysis

%linear
coeff1 = polyfit(year,yearData,1);
f1 = polyval(coeff1,year); 
hold on
plot(year,f1,'LineWidth',1.5,'Color','r')

%quadratic 
coeff2 = polyfit(year,yearData,2);
f2 = polyval(coeff2,year); 
hold on
plot(year,f2,'LineWidth',1.5,'Color','g')
legend('Data','Linear interpolation','Quadratic interpolation')

I would like to extend both linear and quadratic interpolations, so that they reach up to, for example 2100. Any advice on how to do that?

All the best,

Maja

[Jonas]

just add the numbers you asked for to the year variable in the polyval function, it evaluates the regression model at the point you enter there

[Maja Zdulska]

Thanks a lot!

[Scott MacKenzie]

I made the modification Jona suggested. Seems the model predicts 2066 as the year all the sea ice is gone.

Maja, may I ask: Where did you get this data?

[Maja Zdulska]

Hi Scott,

Data is from the National Snow and Ice Data Center:

https://nsidc.org/data/g02135

[Scott MacKenzie]

Great. Thanks very much. Very nice work.