Mean of 3D matrix based on range of first dimension

I have a 3d matrix of size (365,360,720) with daily half degree global temperature data. I'd like to take the average over a range of days, say 100-200. So effectively I'd like a mean (100:200,:,:) into a matrix of size (360,720). I'm stuck on how to quickly do this. Thanks.

Answer 1

[Star Strider]

Perhaps something like this --

A = randn(360,360,720);

At = A(100:200,:,:);

Atm = mean(At, 1);

AtmSize = size(Atm)

AtmSize = 1×3

1 360 720

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Atm = squeeze(Atm)

Atm = 360×720

-0.0563 -0.0387 -0.0523 0.0762 -0.1329 -0.0509 0.0523 0.0332 -0.2267 -0.1836 -0.0127 -0.0402 -0.0735 0.1392 0.0235 0.0728 -0.0159 0.1024 0.0071 0.0243 -0.0520 -0.0282 0.0572 -0.0392 -0.0686 0.0673 0.0416 0.0407 0.0097 0.0737 0.0469 -0.0999 0.0627 0.0330 -0.0786 -0.0207 -0.0271 -0.1168 0.0365 0.1092 0.0897 0.0962 0.1419 0.0330 -0.0173 0.1815 -0.0110 0.0431 0.1018 -0.0074 -0.0927 -0.1060 0.0711 -0.0474 0.1863 0.1030 -0.0059 0.0058 0.0447 -0.0826 0.1906 0.0866 -0.2160 -0.2412 0.1257 0.1013 -0.0033 -0.1745 -0.0701 0.0379 -0.0344 0.0819 0.0688 -0.0467 0.1903 -0.0803 0.2209 0.2378 -0.0555 0.0585 -0.0532 -0.0156 -0.0321 0.1816 0.0592 0.0265 -0.2052 -0.0298 -0.0451 0.0030 -0.0708 0.0826 0.1403 -0.0555 -0.0678 0.0524 -0.0569 0.1118 -0.0676 -0.0215 0.1798 -0.1957 -0.0228 -0.1010 -0.1025 -0.1472 0.0396 -0.0681 -0.0649 -0.0266 0.1162 0.1447 0.1257 -0.0216 -0.0912 0.0638 0.0584 -0.0496 -0.1483 0.0754 -0.0800 -0.0840 0.1532 -0.0463 0.0745 -0.0121 -0.1472 -0.0364 -0.3824 -0.0633 -0.0319 -0.0034 0.0276 -0.0376 0.0569 -0.0032 0.1157 -0.0350 -0.1528 0.0798 -0.0070 0.0010 0.0776 0.0248 0.2047 0.0146 -0.0516 -0.1305 0.1231 -0.1489 0.0909 0.0122 0.0778 0.1473 -0.0347 -0.0859 0.1104 -0.1017 -0.0618 -0.1294 0.1392 0.1307 0.0435 -0.0578 0.0485 0.0177 0.0465 0.0444 0.1278 -0.0675 -0.0554 0.0631 0.0257 -0.0960 0.1843 -0.2281 -0.0226 -0.0779 -0.0059 0.0229 0.1359 0.0326 0.1371 -0.0836 0.0047 0.0734 0.0038 0.0420 -0.0356 -0.0251 0.1240 0.0632 -0.0085 -0.1897 0.0406 0.1528 -0.1264 -0.0851 -0.0461 -0.1030 -0.0925 0.2176 0.0395 0.2468 0.0988 0.0961 -0.0090 0.1564 0.0624 -0.0553 0.1334 -0.2958 -0.0802 -0.0781 0.0857 -0.0047 -0.1466 0.0133 -0.0773 -0.1473 -0.0936 0.1276 -0.0578 -0.0330 -0.0075 0.0237 0.0484 -0.0033 -0.1414 0.0139 0.1428 0.0325 0.1843 -0.0210 -0.0465 0.0411 -0.0249 0.1968 -0.1111 0.0148 0.0386 -0.1297 -0.0755 0.1352 0.0123 -0.1832 -0.0537 -0.1020 -0.0122 0.0468 -0.1295 -0.1598 0.1508 -0.0296 -0.1561 -0.0962 -0.1094 0.0313 0.0405 -0.0062 -0.0712 0.0468 0.1133 -0.1454 0.0240 0.0102 0.1901 0.1144 -0.0110 -0.0791 -0.0127 0.0958 -0.0837 -0.1954 -0.0482 0.0117 0.1084 -0.1303 -0.0329 0.0282 0.2331 0.0007 -0.0029 -0.0874 0.0730 -0.0068 -0.0071 -0.1582 -0.1228 -0.0003 0.1739 -0.1212 -0.0658 0.0427 -0.0349 0.1073 -0.1038 0.2022 -0.0739 -0.1452 0.1411 -0.0730 0.0093 0.0812 -0.0779 -0.0931 0.1126 0.0753 -0.1362 0.0440 -0.1879 -0.0543 -0.1332 0.1606 0.0242 -0.1093 0.1185 -0.1820 -0.0239 -0.2374 0.1241 0.0591 0.1297 0.1133 0.1033 -0.0727 -0.1683 0.0495 0.0288 0.1134 0.2239 -0.1604 -0.0833 -0.0084 0.0000 0.2061 0.0766 -0.0148 0.1127 -0.1086 -0.0613 0.0319 0.0190 -0.2960 -0.0999 -0.0719 0.0750 -0.0828 -0.1119 0.0608 -0.0867 0.1495 -0.1421 -0.0346 -0.0366 -0.1248 -0.0670 -0.1580 0.1969 0.1194 0.0338 0.0934 -0.0507 0.0570 0.1196 0.0105 -0.0177 -0.0464 0.0478 -0.1821 0.0919 -0.1310 -0.1995 0.0044 0.0293 0.0679 0.1780 0.0508 -0.0188 0.1694 0.1436 -0.0651 0.0600 0.0862 -0.0105 -0.0182 0.0309 -0.1479 -0.1018 -0.1670 0.1421 -0.0875 0.0314 0.0523 0.1086 -0.0037 -0.0581 -0.0768 -0.1244 0.0780 0.0105 -0.0382 0.0635 0.0418 0.0433 -0.2012 0.0198 -0.1365 0.0818 -0.0022 0.0028 0.1736 -0.0620 0.1086 -0.1206 -0.0350 -0.0608 0.1143 -0.0987 -0.0882 -0.0075 0.0770 0.1313 -0.0606 -0.0470 -0.0827 -0.0535 -0.0111 0.1745 -0.0278 0.0906 0.0134 0.0343 -0.0381 -0.1171 -0.0575 0.2197 0.0188 0.1222 -0.1152 0.0107 0.0537 -0.0222 0.0070 -0.0486 -0.0998 0.0843 -0.1137 -0.0296 0.0089

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

AtmSize = size(Atm)

AtmSize = 1×2

360 720

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

.

Answer 2

[Matt J]

This might seem a little counter-intuitive, but the key is that it avoids extracting a submatrix (an expensive op) from the input matrix A,

s=zeros(1,365); s(100:200)=1; s=s/sum(s);

result = reshape(s*A(:,:),360,720);

[Matt J]

A short execution time comparison:

A = randn(365,360,720);

timeit(@()methodStandard(A))

ans = 0.1238

timeit(@()methodProposed(A))

ans = 0.0075

function methodStandard(A)

 At = A(100:200,:,:);
 result = reshape(mean(At, 1),360,720);

end

function methodProposed(A)

 s=zeros(1,365); s(100:200)=1; s=s/sum(s);

 result = reshape(s*A(:,:),360,720);

end