Yep, it's possible.
Check out these examples using patch(). Your x and y values will be defined by the curve and the limits of your axes. The c value will be defined by GDP.
Demo
y = rand(1,100); % this would be GDP
x = linspace(1,200,100);
% x,y define the 'top' of the shape. Now we define and sides and bottom.
yy = [y, zeros(size(x))];
xx = [x, fliplr(x)];
figure()
patch(xx,yy, [y,zeros(size(y))])
colorbar
Note the redundancy between the y axis values and the colorbar values which indicate the same thing. The only added benefit is visual aesthetic.
