bug with interp2 ?

Question

Timbro 2023-5-2

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1956904-bug-with-interp2

评论： Timbro 2023-5-3

Hello,

I want to interpolate a 2D data "v" from a 2km latitude and longitude grid (xx and yy) to a 10 km grid (Xq and Yq). but with interp2 I get the following error :

>> [Xq,Yq] = ndgrid(lon_extraction,lat_extraction); % 
>> whos xx yy v X Y
  Name        Size               Bytes  Class     Attributes
  X          23x83               15272  double              
  Y          23x83               15272  double              
  v         407x793            2582008  double              
  xx        407x793            2582008  double              
  yy        407x793            2582008  double              
>> h_extr_interp = interp2(xx,yy, v, X,Y);
Error using griddedInterpolant
Grid arrays must have NDGRID structure.
Error in interp2>makegriddedinterp (line 226)
    F = griddedInterpolant(varargin{:});
Error in interp2 (line 134)
        F = makegriddedinterp(X, Y, V, method,extrap);
 
>> 

(I get the same result using meshgrid instead of ndgrid)

any ideas why I get this message??

3 个评论
显示 1更早的评论隐藏 1更早的评论

Stephen23 2023-5-3

编辑：Stephen23 2023-5-3

"(I get the same result using meshgrid instead of ndgrid)"

Your code shows that you do not use the outputs of NDGRID, so changing the function to MESHGRID will not change that.

"Problem solved using griddata instead of interp2"

If you have data in ND-grid format, then use INTERPN:

https://www.mathworks.com/help/matlab/ref/interpn.html

The MATLAB documentation explains how to select the operator based on the data arrangement:

https://www.mathworks.com/help/matlab/interpolation.html

Matt J 2023-5-3

Problem solved using griddata instead of interp2

griddata is quite a bit slower than interp2. You should not use it unless you are certain your xx,yy locations don't define a regular lattice.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2023-5-2

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1956904-bug-with-interp2#answer_1227994

移动：Walter Roberson 2023-5-2

在 MATLAB Online 中打开

[Xq,Yq] = ndgrid(lon_extraction,lat_extraction); 
 h_extr_interp = interp2(xx,yy, v, X,Y);

What do xx and yy or X and Y have to do with Xq and Yq ? You build an ndgrid but you do not use it in the interp2

5 个评论
显示 3更早的评论隐藏 3更早的评论

Stephen23 2023-5-3

编辑：Stephen23 2023-5-3

在 MATLAB Online 中打开

"Maybe interp2 can only be used to switch from a grid to a finer grid, but not the opposite."

No.

Lets try it right now, modifying the first example from the documentation:

[X,Y] = meshgrid(-3:3,-2:2);
V = peaks(X,Y)
V = 5×7
    0.0007    0.0468   -0.5921   -4.7596   -2.1024   -0.0616    0.0004
   -0.0088   -0.1301    1.8559   -0.7239   -0.2729    0.4996    0.0130
   -0.0365   -1.3327   -1.6523    0.9810    2.9369    1.4122    0.0331
   -0.0137   -0.4808    0.2289    3.6886    2.4338    0.5805    0.0125
    0.0000    0.0797    2.0967    5.8591    2.2099    0.1328    0.0013

Finer grid:

[Xq,Yq] = meshgrid(-3:0.25:3,-2:0.25:2);
Vq = interp2(X,Y,V,Xq,Yq)
Vq = 17×25
    0.0007    0.0123    0.0238    0.0353    0.0468   -0.1129   -0.2726   -0.4324   -0.5921   -1.6340   -2.6759   -3.7177   -4.7596   -4.0953   -3.4310   -2.7667   -2.1024   -1.5922   -1.0820   -0.5718   -0.0616   -0.0461   -0.0306   -0.0151    0.0004
   -0.0016   -0.0006    0.0005    0.0016    0.0026    0.0069    0.0112    0.0156    0.0199   -0.9228   -1.8654   -2.8080   -3.7507   -3.2243   -2.6978   -2.1714   -1.6450   -1.2141   -0.7832   -0.3522    0.0787    0.0599    0.0411    0.0223    0.0036
   -0.0040   -0.0134   -0.0228   -0.0322   -0.0416    0.1268    0.2951    0.4635    0.6319   -0.2115   -1.0549   -1.8983   -2.7418   -2.3532   -1.9647   -1.5762   -1.1876   -0.8360   -0.4843   -0.1327    0.2190    0.1659    0.1129    0.0598    0.0067
   -0.0064   -0.0262   -0.0461   -0.0660   -0.0858    0.2466    0.5790    0.9115    1.2439    0.4997   -0.2445   -0.9887   -1.7328   -1.4822   -1.2316   -0.9809   -0.7303   -0.4579   -0.1855    0.0869    0.3593    0.2720    0.1846    0.0972    0.0099
   -0.0088   -0.0391   -0.0694   -0.0997   -0.1301    0.3664    0.8629    1.3594    1.8559    1.2109    0.5660   -0.0790   -0.7239   -0.6112   -0.4984   -0.3857   -0.2729   -0.0798    0.1134    0.3065    0.4996    0.3780    0.2563    0.1347    0.0130
   -0.0157   -0.1195   -0.2232   -0.3270   -0.4307   -0.0783    0.2741    0.6264    0.9788    0.6597    0.3406    0.0215   -0.2977   -0.0909    0.1159    0.3227    0.5295    0.5791    0.6287    0.6782    0.7278    0.5503    0.3729    0.1955    0.0180
   -0.0226   -0.1998   -0.3770   -0.5542   -0.7314   -0.5231   -0.3148   -0.1065    0.1018    0.1085    0.1152    0.1219    0.1286    0.4294    0.7303    1.0311    1.3320    1.2380    1.1440    1.0499    0.9559    0.7227    0.4895    0.2563    0.0231
   -0.0296   -0.2802   -0.5308   -0.7814   -1.0320   -0.9678   -0.9037   -0.8395   -0.7753   -0.4428   -0.1103    0.2223    0.5548    0.9497    1.3446    1.7395    2.1345    1.8969    1.6592    1.4216    1.1840    0.8950    0.6061    0.3171    0.0281
   -0.0365   -0.3606   -0.6846   -1.0086   -1.3327   -1.4126   -1.4925   -1.5724   -1.6523   -0.9940   -0.3357    0.3227    0.9810    1.4700    1.9590    2.4480    2.9369    2.5557    2.1745    1.7934    1.4122    1.0674    0.7226    0.3779    0.0331
   -0.0308   -0.3030   -0.5753   -0.8475   -1.1197   -1.1353   -1.1509   -1.1665   -1.1820   -0.4720    0.2379    0.9479    1.6579    1.9462    2.2345    2.5228    2.8111    2.4094    2.0077    1.6060    1.2042    0.9102    0.6161    0.3220    0.0280

Courser grid:

[Xq,Yq] = meshgrid(-3:2:3,-2:2:2);
Vq = interp2(X,Y,V,Xq,Yq)
Vq = 3×4
    0.0007   -0.5921   -2.1024    0.0004
   -0.0365   -1.6523    2.9369    0.0331
    0.0000    2.0967    2.2099    0.0013

No errors, no problems, everything works exactly as documented.

"bug with interp2 ?"

I very much doubt that.

Most likely the problem is the format of your data, which so far you have not provided us with. For example, if the data are actually in ND-grid format:

[X,Y] = ndgrid(-3:3,-2:2);
V = peaks(X,Y) % not meshgrid data
V = 7×5
    0.0007   -0.0088   -0.0365   -0.0137    0.0000
    0.0468   -0.1301   -1.3327   -0.4808    0.0797
   -0.5921    1.8559   -1.6523    0.2289    2.0967
   -4.7596   -0.7239    0.9810    3.6886    5.8591
   -2.1024   -0.2729    2.9369    2.4338    2.2099
   -0.0616    0.4996    1.4122    0.5805    0.1328
    0.0004    0.0130    0.0331    0.0125    0.0013
[Xq,Yq] = meshgrid(-3:2:3,-2:2:2); % what you use here is irrelevant
Vq = interp2(X,Y,V,Xq,Yq) % ... because X & Y still have the wrong format
Error using interp2>makegriddedinterp
Input grid is not a valid MESHGRID.

Error in interp2 (line 134)
        F = makegriddedinterp(X, Y, V, method,extrap);

If you either uploaded or showed your actual data, then we could check its format. Until you do that, we can safely assume that its format is incorrect for the interpolation function that you have chosen.

Stephen23 2023-5-3

编辑：Stephen23 2023-5-3

在 MATLAB Online 中打开

test_bug.mat

Your data are scattered, they are not on a grid:

S = load('test_bug.mat')
S = struct with fields:
    Xq: [23×83 double]
    Yq: [23×83 double]
     v: [407×793 double]
    xx: [407×793 double]
    yy: [407×793 double]
S.xx
ans = 407×793
  -19.9868  -19.9849  -19.9830  -19.9811  -19.9792  -19.9774  -19.9755  -19.9736  -19.9717  -19.9698  -19.9679  -19.9660  -19.9641  -19.9622  -19.9603  -19.9585  -19.9566  -19.9547  -19.9528  -19.9509  -19.9490  -19.9471  -19.9453  -19.9434  -19.9415  -19.9396  -19.9377  -19.9358  -19.9340  -19.9321
  -19.9627  -19.9608  -19.9589  -19.9570  -19.9551  -19.9532  -19.9513  -19.9494  -19.9475  -19.9456  -19.9437  -19.9419  -19.9400  -19.9381  -19.9362  -19.9343  -19.9324  -19.9305  -19.9286  -19.9267  -19.9249  -19.9230  -19.9211  -19.9192  -19.9173  -19.9154  -19.9135  -19.9117  -19.9098  -19.9079
  -19.9385  -19.9366  -19.9347  -19.9328  -19.9310  -19.9291  -19.9272  -19.9253  -19.9234  -19.9215  -19.9196  -19.9177  -19.9158  -19.9139  -19.9120  -19.9101  -19.9082  -19.9064  -19.9045  -19.9026  -19.9007  -19.8988  -19.8969  -19.8950  -19.8931  -19.8912  -19.8894  -19.8875  -19.8856  -19.8837
  -19.9144  -19.9125  -19.9106  -19.9087  -19.9068  -19.9049  -19.9030  -19.9011  -19.8992  -19.8973  -19.8954  -19.8935  -19.8916  -19.8898  -19.8879  -19.8860  -19.8841  -19.8822  -19.8803  -19.8784  -19.8765  -19.8746  -19.8727  -19.8708  -19.8690  -19.8671  -19.8652  -19.8633  -19.8614  -19.8595
  -19.8902  -19.8883  -19.8864  -19.8846  -19.8827  -19.8808  -19.8789  -19.8770  -19.8751  -19.8732  -19.8713  -19.8694  -19.8675  -19.8656  -19.8637  -19.8618  -19.8599  -19.8580  -19.8561  -19.8542  -19.8523  -19.8505  -19.8486  -19.8467  -19.8448  -19.8429  -19.8410  -19.8391  -19.8372  -19.8353
  -19.8661  -19.8642  -19.8623  -19.8604  -19.8585  -19.8566  -19.8547  -19.8528  -19.8509  -19.8490  -19.8471  -19.8452  -19.8433  -19.8414  -19.8395  -19.8376  -19.8358  -19.8339  -19.8320  -19.8301  -19.8282  -19.8263  -19.8244  -19.8225  -19.8206  -19.8187  -19.8168  -19.8149  -19.8130  -19.8112
  -19.8420  -19.8401  -19.8382  -19.8363  -19.8344  -19.8325  -19.8306  -19.8287  -19.8268  -19.8249  -19.8230  -19.8211  -19.8192  -19.8173  -19.8154  -19.8135  -19.8116  -19.8097  -19.8078  -19.8059  -19.8040  -19.8021  -19.8002  -19.7983  -19.7964  -19.7945  -19.7926  -19.7908  -19.7889  -19.7870
  -19.8178  -19.8159  -19.8140  -19.8121  -19.8102  -19.8083  -19.8064  -19.8045  -19.8026  -19.8007  -19.7988  -19.7969  -19.7950  -19.7931  -19.7912  -19.7893  -19.7874  -19.7855  -19.7836  -19.7817  -19.7798  -19.7779  -19.7760  -19.7741  -19.7723  -19.7704  -19.7685  -19.7666  -19.7647  -19.7628
  -19.7937  -19.7918  -19.7899  -19.7880  -19.7861  -19.7842  -19.7823  -19.7804  -19.7785  -19.7766  -19.7747  -19.7728  -19.7709  -19.7690  -19.7671  -19.7652  -19.7633  -19.7614  -19.7595  -19.7576  -19.7557  -19.7538  -19.7519  -19.7500  -19.7481  -19.7462  -19.7443  -19.7424  -19.7405  -19.7386
  -19.7695  -19.7676  -19.7657  -19.7638  -19.7619  -19.7600  -19.7581  -19.7562  -19.7543  -19.7524  -19.7505  -19.7486  -19.7467  -19.7448  -19.7429  -19.7410  -19.7391  -19.7372  -19.7353  -19.7334  -19.7315  -19.7296  -19.7277  -19.7258  -19.7239  -19.7220  -19.7201  -19.7182  -19.7163  -19.7144
S.yy
ans = 407×793
    7.0762    7.1011    7.1261    7.1510    7.1759    7.2009    7.2258    7.2507    7.2756    7.3006    7.3255    7.3504    7.3754    7.4003    7.4252    7.4501    7.4751    7.5000    7.5249    7.5499    7.5748    7.5997    7.6246    7.6496    7.6745    7.6994    7.7244    7.7493    7.7742    7.7991
    7.0744    7.0993    7.1243    7.1492    7.1741    7.1991    7.2240    7.2489    7.2738    7.2988    7.3237    7.3486    7.3736    7.3985    7.4234    7.4483    7.4733    7.4982    7.5231    7.5481    7.5730    7.5979    7.6228    7.6478    7.6727    7.6976    7.7226    7.7475    7.7724    7.7973
    7.0726    7.0975    7.1225    7.1474    7.1723    7.1973    7.2222    7.2471    7.2720    7.2970    7.3219    7.3468    7.3718    7.3967    7.4216    7.4465    7.4715    7.4964    7.5213    7.5463    7.5712    7.5961    7.6210    7.6460    7.6709    7.6958    7.7208    7.7457    7.7706    7.7956
    7.0708    7.0957    7.1207    7.1456    7.1705    7.1954    7.2204    7.2453    7.2702    7.2952    7.3201    7.3450    7.3700    7.3949    7.4198    7.4447    7.4697    7.4946    7.5195    7.5445    7.5694    7.5943    7.6192    7.6442    7.6691    7.6940    7.7190    7.7439    7.7688    7.7938
    7.0690    7.0939    7.1188    7.1438    7.1687    7.1936    7.2186    7.2435    7.2684    7.2934    7.3183    7.3432    7.3681    7.3931    7.4180    7.4429    7.4679    7.4928    7.5177    7.5427    7.5676    7.5925    7.6174    7.6424    7.6673    7.6922    7.7172    7.7421    7.7670    7.7920
    7.0672    7.0921    7.1170    7.1420    7.1669    7.1918    7.2168    7.2417    7.2666    7.2915    7.3165    7.3414    7.3663    7.3913    7.4162    7.4411    7.4661    7.4910    7.5159    7.5409    7.5658    7.5907    7.6156    7.6406    7.6655    7.6904    7.7154    7.7403    7.7652    7.7902
    7.0654    7.0903    7.1152    7.1402    7.1651    7.1900    7.2149    7.2399    7.2648    7.2897    7.3147    7.3396    7.3645    7.3895    7.4144    7.4393    7.4643    7.4892    7.5141    7.5390    7.5640    7.5889    7.6138    7.6388    7.6637    7.6886    7.7136    7.7385    7.7634    7.7884
    7.0636    7.0885    7.1134    7.1383    7.1633    7.1882    7.2131    7.2381    7.2630    7.2879    7.3129    7.3378    7.3627    7.3877    7.4126    7.4375    7.4624    7.4874    7.5123    7.5372    7.5622    7.5871    7.6120    7.6370    7.6619    7.6868    7.7118    7.7367    7.7616    7.7866
    7.0617    7.0867    7.1116    7.1365    7.1615    7.1864    7.2113    7.2363    7.2612    7.2861    7.3111    7.3360    7.3609    7.3858    7.4108    7.4357    7.4606    7.4856    7.5105    7.5354    7.5604    7.5853    7.6102    7.6352    7.6601    7.6850    7.7100    7.7349    7.7598    7.7848
    7.0599    7.0849    7.1098    7.1347    7.1596    7.1846    7.2095    7.2344    7.2594    7.2843    7.3092    7.3342    7.3591    7.3840    7.4090    7.4339    7.4588    7.4838    7.5087    7.5336    7.5586    7.5835    7.6084    7.6334    7.6583    7.6832    7.7082    7.7331    7.7580    7.7829

The rows and columns clearly do not consist of identical values. Therefore your data are not gridded.

Compare against the perfectly gridded sample points returned by NDGRID or MESHGRID:

[X,Y] = meshgrid(0:3,4:7)
X = 4×4
   1     2     3
   1     2     3
   1     2     3
   1     2     3
Y = 4×4
   4     4     4
   5     5     5
   6     6     6
   7     7     7