Status of LUE

File format

TODO

Modelling operations

Currently, PCRaster contains implementations of more modelling operations than LUE. Over time more operations will be added to LUE. Here we compare the list of PCRaster operations with the current set of operations implemented in LUE. When planning to port a PCRaster model to LUE, it is a good idea to compare the information in the next table with the operations used by the model.

Here we only mention LUE operations that implement an operation that is also in PCRaster. LUE also contains other operations, not listed here.

If an operation used in a PCRaster model is not available in LUE yet, it is a good idea to report it to the project team. Various factors determine the order in which operations are added to the LUE framework. The number of model developers requesting one is one of them.

Note

A LUE operation that implements an operation that is also in PCRaster may be different from the PCRaster operation. It may be named different, it may accept different arguments, or accept them in a different order. We provide a special module that can be used to quickly port PCRaster model by hiding these differences as much as possible.

Table 1 Legend

✅	LUE supports this PCRaster operation, or an operation like it. It may be that the LUE operation is named differently, that it supports different parameters, or that it accepts parameters in a different order.
❌	This PCRaster operations is not considered yet. It may be in the future.
🗑	This PCRaster operation is considered, but skipped. See notes for reason.
1, 2, 3	Guesstimate of the amount of effort required to add it: 1 little, 2 somewhat, 3 serious. This is only relevant for operations that have not been added yet.

Table 2 Status of LUE operations vs PCRaster operations

PCRaster	LUE	Notes
+	✅
-	✅	subtract and negate
/ or div	✅	div is only part of lue.pcraster
*	✅
**	✅
abs	✅	abs with non-spatial expression is not available
accucapacityflux accucapacitystate	❌	1
accuflux	✅	not yet in Python package accuflux with non-spatial material is not available
accufractionflux accufractionstate	✅	accu_fraction accu_fraction with non-spatial material and non-spatial transport fraction is not available
accuthresholdflux accuthresholdstate	✅	accu_threshold3 accu_threshold with non-spatial material and non-spatial threshold is not available
accutriggerflux accutriggerstate	❌	1
accutraveltimestate accutraveltimeflux accutraveltimeremoved	❌	2
accutraveltimefractionstate accutraveltimefractionflux accutraveltimefractionremoved	❌	2
acos	✅
and, &, pcrand	✅
argorder argorderwithid	❌	?
argorderaddarealimited argorderwithidaddarealimited	❌	?
argorderarealimited argorderwithidarealimited	❌	?
areaarea	✅	zonal_area Returns the number of cells per class
areaaverage	✅	zonal_mean
areadiversity	✅	zonal_diversity
areamajority	✅	zonal_majority
areamaximum	✅	zonal_maximum
areaminimum	✅	zonal_minimum
areanormal	✅	zonal_normal
areaorder	❌	1
areatotal	✅	zonal_sum
areauniform	✅	zonal_uniform Returned values are in range [0, 1)
asin	✅
aspect	✅
atan	✅
boolean	✅	boolean with non-spatial argument is not available
catchment	❌	2
catchmenttotal	❌	1
cellarea	❌	1
celllength	❌	1
clump	✅	LUE has support for diagonal and nondiagonal connectivity. Use diagonal connectivity to get the same behaviour as PCRaster’s clump.
cos	✅
cover	❌	1
defined	✅
directional	🗑	LUE currently does not support value scales.
downstream	✅
downstreamdist	✅
dynwavestate dynwaveflux lookuppotential lookupstate \	❌	3
dynamicwave	❌	3
eq or ==	✅
exp	✅
extentofview	❌	2
fac	❌	1
ge or >=	✅
gt or >	✅
horizontan	❌	1
idiv, //	❌	1
ifthen	✅	where(condition_expression, true_expression)
ifthenelse	✅	where(condition_expression, true_expression, false_expression)
influencesimplegauss	❌	?
inversedistance	❌	2
kinematic	✅	kinematic_wave The nrTimeSlices argument is removed. Negative values for the the discharge, lateral inflow and channel length arguments result in no-data values in the result (Github issue 524)
kinwavestate kinwaveflux	❌	?
ldd	🗑	LUE currently does not support value scales.
lddcreate	✅	The operation assumes the DEM passed in is sound: depressions are filled and flats are handled.
lddcreatedem	❌	3
ldddist	❌	2
lddmask	❌	2
lddrepair	❌	2
le or <=	✅
ln	✅	log
log10	✅
lookup	❌	?
lookuplinear	❌	?
lookupmapstack	❌	?
lt or <	✅
maparea	❌	1
mapmaximum	✅	maximum
mapminimum	✅	minimum
mapnormal	✅	normal(array, dtype, 0, 1), normal(array_shape, partition_shape, dtype, 0, 1)
maptotal	✅	sum
mapuniform	✅	uniform(array, dtype, 0, 1), uniform(array_shape, partition_shape, dtype, 0, 1)
markwhilesumle markwhilesumge	❌	2
max	❌	1
min	❌	1
mod, %	❌	1
ne or !=	✅
nodirection	❌	1
nominal	✅
normal	✅
not ~ pcrnot	✅
or | pcror	✅
order	❌	? Idea is easy, parallel implementation seems not.
ordinal	✅
path	❌	2
pit	❌	1
plancurv	❌	2
pred	🗑	Not sure why this is useful…
profcurv	❌	2
rounddown	✅	floor
roundoff	✅	round
roundup	✅	ceil
scalar	✅
shift shift0	🗑	Not sure why this is useful…
sin	✅
slope	✅
slopelength	❌	3
spatial	✅
spread	❌	3
spreadldd	❌	3
spreadlddzone	❌	3
spreadmax	❌	3
spreadmaxzone	❌	3
spreadzone	❌	3
sqr	✅	Not as an operation. Just use value * value.
sqrt	✅
streamorder	❌	2
subcatchment	❌	3
succ	🗑	Not sure why this is useful…
tan	✅
time	🗑	Things work differently in a LUE C++/Python model.
timeinput…	❌	?
timeinput	❌	?
timeinputmodulo	❌	?
timeinputsparse	❌	?
timeoutput	❌	?
timeslice	❌	?
transient	❌	2
uniform	✅
uniqueid	✅	unique_id
upstream	✅
view	❌	3
window4total	🗑	Not needed, LUE’s focal_mean accepts a custom kernel
windowaverage	✅	focal_mean
windowdiversity	✅	focal_diversity
windowhighpass	✅	focal_high_pass
windowmajority	✅	focal_majority Note that, unlike in PCRaster, the kernel window is not enlarged when no unique most frequent value is encountered within the focal window. In that case, no-data is output.
windowmaximum	✅	focal_maximum
windowminimum	✅	focal_minimum
windowtotal	✅	focal_sum
xcoordinate	🗑	Not sure why this is useful…
xor ^ pcrxor	✅
ycoordinate	🗑	Not sure why this is useful…

Modelling framework

Like PCRaster, LUE contains helper classes for implementing forward numerical simulation models as well. The lue.pcraster.framework sub package contains classes that behave like the PCRaster Python Framework classes, but are implemented in terms of the LUE classes. Not all PCRaster classes have been ported to LUE yet. The next table lists the current status.

Table 3 Legend

✅	LUE supports this class. It may be that the LUE class and its member functions are named differently, that the member functions support different parameters, or that they accept parameters in a different order.
❌	This PCRaster class is not considered yet. It may be in the future.
🗑	This PCRaster class is considered, but skipped. See notes for reason.
1, 2, 3	Guesstimate of the amount of effort required to add it: 1 little, 2 somewhat, 3 serious. This is only relevant for classes that have not been added yet.

Table 4 Status of LUE model development support classes vs PCRaster Python Framework classes

PCRaster	LUE	Notes
StaticModel	❌	Work in progress
StaticFramework	❌	Work in progress
DynamicModel	❌	Work in progress
DynamicFramework	❌	Work in progress
MonteCarloModel	❌
MonteCarloFramework	❌
ParticleFilterModel	❌
ParticleFilterFramework	❌
ResidualResamplingFramework	❌
SequentialImportanceResamplingFramework	❌
EnKfModel	❌
EnsKalmanFilterFramework	❌