Nonsensical results from r.resamp.interp with certain region settings

[sorokine]

I am resampling a raster to higher resolution using bilinear or bicubic methods but I am getting erroneous result.

Steps to reproduce and more information:

r.info rast1
...
 |   Type of Map:  raster               Number of Categories: 255             |
 |   Data Type:    FCELL                                                      |
 |   Rows:         128                                                        |
 |   Columns:      256                                                        |
 |   Total Cells:  32768                                                      |
 |        Projection: Latitude-Longitude                                      |
 |            N: 89:37:40.633467N    S:        90S   Res: 1:24:12.036199      |
 |            E: 179:17:48.75E    W: 179:17:48.75E   Res: 1:24:22.5           |
 |   Range of data:    min = 0.000000  max = 0.000000                         |
...

r.region rast=rast1
g.region nsres=0.2 ewres=0.2
r.resamp.interp --o input=rast1 output=rast1resampled

rast1resampled does not look like anything close to expected.

Workaround

Shove 3 degrees from the region before resampling:

g.region n=87 s=-87 w=-177 e=177

Resampled raster will be correct.