rgee and the Earth Engine Python API share the same classes, functions, and methods. Therefore, is expected that works identical to all user requests. However, differences in the language design of R and Python might cause some problems. We identify three bug-potential scenarios. Each of them are explained in-depth below.
rgee may show an error message when valid requests to the Earth Engine server are made. This issue happens when the map method is used under the next two cases: (1) users employing a reticulate version lower than < 1.14 (please update it!); and (2) if you are leading with ee$List objects. For instance:
library(rgee) ee$Initialize() mylist = ee$List$sequence(10) mylist$map(function(x) ee$Number(x)$add(1)) #> Error in py_call_impl(callable, dots$args, dots$keywords): RuntimeError: Evaluation error: argument "x" is missing, with no default. #> #> Detailed traceback: #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/apifunction.py", line 205, in <lambda> #> return lambda *args, **kwargs: func.call(*args, **kwargs) # pylint: disable=unnecessary-lambda #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/function.py", line 67, in call #> return self.apply(self.nameArgs(args, kwargs)) #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/function.py", line 80, in apply #> result = computedobject.ComputedObject(self, self.promoteArgs(named_args)) #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/function.py", line 107, in promoteArgs #> promoted_args[name] = Function._promoter(args[name], spec['type']) #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/__init__.py", line 242, in _Promote #> return CustomFunction.create(arg, 'Object', ['Object'] * args_count) #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/customfunction.py", line 121, in create #> return CustomFunction(signature, func) #> File "/home/aybarpc01/.virtualenvs/r-reticulate/lib/python3.7/site-packages/ee/customfunction.py", line 47, in __init__ #> self._body = body(*variables) #> File "/home/aybarpc01/R/x86_64-pc-linux-gnu-library/3.6/reticulate/python/rpytools/call.py", line 21, in python_function #> raise RuntimeError(res[kErrorKey])
The code before is perfectly valid but rgee will produce an error. This problem should be easily solved by adding the function ee_utils_pyfunc. It will permit to wrap R functions before to send it to reticulate. Let’s see:
library(rgee) ee$Initialize() mylist = ee$List$sequence(0,10) mynewlist = mylist$map( ee_utils_pyfunc( function(x) ee$Number(x)$add(1) ) ) mynewlist$getInfo() #> [1] 1 2 3 4 5 6 7 8 9 10 11
By default, when you define a number in R it will produce a double precision value. This does not happen in Python because, by default it will create a int value.
Python
R
class(1) #> [1] "numeric"
But why does this matter? Let’s explain it with an example:
Python
R
and_bitwise = ee$Number(32)$bitwiseAnd(100) #caution: silent error
and_bitwise$getInfo()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/aybarpc01/.local/lib/python3.7/site-packages/ee/computedobject.py", line 95, in getInfo
return data.computeValue(self)
File "/home/aybarpc01/.local/lib/python3.7/site-packages/ee/data.py", line 490, in computeValue
return send_('/value', ({'json': obj.serialize(), 'json_format': 'v2'}))
File "/home/aybarpc01/.local/lib/python3.7/site-packages/ee/data.py", line 1186, in send_
raise ee_exception.EEException(json_content['error']['message'])
ee.ee_exception.EEException: Number.bitwiseAnd: Bitwise operands must be integer only.Users need to take into consideration that most of the arguments of the Earth Engine methods are strict to admit only integer values. The creation of integers in R is quite simple, you just need to add the letter L to the end of the specific number or employ the function as.integer. The correct code in R would be:
and_bitwise = ee$Number(32L)$bitwiseAnd(100L) and_bitwise$getInfo() #> [1] 32
This problem also appears due to differences between the design of R and Python as programming languages. Currently, R’s only integer data type is a 32 bit signed integer. Such integers can only count up to about 2 billion. This range is extremely insufficient to deal with Google Earth Engine timestamp which is saved in milliseconds since the UNIX epoch.
Python
R
my_date <- ee$Date('1990-01-01') my_date$getInfo() #> $type #> [1] "Date" #> #> $value #> [1] -208192512
The problems with ee$Date just appear in the last mile (Python to R or vice-versa, reticulate), and it should not be a trouble if treated with care. rgee implement two functions to deal with Earth Engine dates: eedate_to_rdate and rdate_to_eedate.
# Era5 dataset era_img <- ee$ImageCollection("ECMWF/ERA5/DAILY")$ filterDate("2019-01-01", "2019-12-31")$ first() # Extracting init date ee_date <- era_img$get('system:time_start') ee_date$getInfo() # Silent error #> [1] 112573440 eedate_to_rdate(ee_date = ee_date, timestamp = TRUE) #> [1] 1.546301e+12