Trends are statistically significant or not

I have the p-values for each pixel in idanom_trend_per_winter(:,:), and i am checking which areas show statistically significant values in my dataset. So put the NaN's and values less than 0.05(the default value of alpha, i am using R2012b) as NaN, and plot the results

for ii = 1:361,
      for jj = 1:361
          if ((idanom_trend_per_winter(:,:) > 0.05) | (isnan(idanom_trend_per_winter(:,:)))) 
              sgt(ii,jj) = NaN;
          else
              sgt(ii,jj) = idanom_trend_per_winter;
          end
      end
  end

_ Error Subscripted assignment dimension mismatch._

Answer 1

[David Barry]

Presumably instead of:

if ((idanom_trend_per_winter(:,:) > 0.05) | (isnan(idanom_trend_per_winter(:,:))))
    sgt(ii,jj) = NaN;
else
    sgt(ii,jj) = idanom_trend_per_winter;
end

You meant:

if ((idanom_trend_per_winter(ii,jj) > 0.05) | (isnan(idanom_trend_per_winter(ii,jj))))
  sgt(ii,jj) = NaN;
else
  sgt(ii,jj) = idanom_trend_per_winter(ii,jj);
end

Anyway, you can do all of this without the need for a loop. I am assuming sgt is meant to be the same size as idanom_trend_per_winter but with the substituted values, so:

sgt = idanom_trend_per_winter;
sgt(idanom_trend_per_winter < 0.05 | isnan(idanom_trend_per_winter)) = NaN;

Oh and also, you said less than 0.05 in your description but your code is checking for greater than.

[Sophia]

Sorry my bad.. the result is the values less than 0.05, so all values above 0.05 as NaN

Thanks!

Answer 2

[Image Analyst]

This: "idanom_trend_per_winter(:,:) > 0.05" is going to give you a 2-D array, not a single number like you might be expecting. Use all() or any() depending on how you want to handle that case.