How to replace no data value -3.4028231e+38 to numpy.nan

Question

I have an 2d array constructed from raster image. The raster image has no data value assigned to -3.4028231e+38, I am trying to replace this value with 'nan' but I am unable to find this value when I apply conditional operator on it.

my data is as following:

>>> slice22 = inndvi[0:2,0:2]
>>> slice22
array([[ -3.40282306e+38,  -3.40282306e+38],
       [ -3.40282306e+38,  -3.40282306e+38]], dtype=float32)

when I try to check these value in if statement:

>>> if slice22[0][0] ==-3.40282306e+38:
...     print "yes"
... else:
...     print "no"
... 
no

the output is 'no'

Due to this I am not able to assign 3.40282306e+38 to numpy.nan as following:

slice22[slice22 == 3.40282306e+38] = numpy.nan

One more thing I would like to mention is that my dataset ranges from +2 to -2 in the raster. I have tried to use the range to eliminate the 3.40282306e+38 value, but still i get errors.

>>> slice22 [slice22 < 2 and slice22 >2 ]= np.nan 
Runtime error 
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

You can not use equality operations on floating point numbers. — Konstantin, Jan 27 '16 at 06:16
you can use `slice22[slice22 < a_very_small_value] = numpy.nan` — realli, Jan 27 '16 at 06:17
@Alik ist here any to type cast this floating point into integer — Gurminder Bharani, Jan 27 '16 at 06:17
More precisely - http://floating-point-gui.de/errors/comparison/ — Konstantin, Jan 27 '16 at 06:18
@chickenjoe even when you cast this to integer you will have to compare floats to `-3.4028231e+38` — Konstantin, Jan 27 '16 at 06:28

score 2 · Accepted Answer · answered Jan 27 '16 at 07:24

2

Considering you know your real values is between -2 and 2, you can easily filter out anything outside of this range.

a[(a < -2) | (a > 2)] = np.nan   #option 1
a[np.abs(a) > 2] = np.nan   #option 2
a[np.logical_or(a < -2, a > 2)] = np.nan   #option 3

answered Jan 27 '16 at 07:24

M4rtini

13,186
4
35
42

thank you for taking out your time and solving my problem option 1 works wonders, once again thank you. – Gurminder Bharani Jan 27 '16 at 07:43

score 1 · Answer 2 · answered Jan 27 '16 at 06:17

Is it crucial for it to have to equal -3.40282306e+38 for your algorithm to work? If nor try an inequality instead of equals. Equlaity works on floating point numbers if they are perfect integers such as 0,for numbers like youre using an inequality maybe the answer. Wont something like

(-3.40282306e+38-smallnumber) fix your problem?-You have to set small number to somthing in the context of your problem which I dont know what it is.

How to replace no data value -3.4028231e+38 to numpy.nan

2 Answers2