NAME
t.rast.mapcalc - Performs spatio-temporal mapcalc expressions on temporally sampled maps of space time raster datasets.
KEYWORDS
temporal,
algebra,
raster,
time
SYNOPSIS
t.rast.mapcalc
t.rast.mapcalc --help
t.rast.mapcalc [-ns] inputs=name[,name,...] expression=string [method=name[,name,...]] output=name basename=basename [nprocs=integer] [--overwrite] [--help] [--verbose] [--quiet] [--ui]
Flags:
- -n
- Register Null maps
- -s
- Check the spatial topology of temporally related maps and process only spatially related maps
- --overwrite
- Allow output files to overwrite existing files
- --help
- Print usage summary
- --verbose
- Verbose module output
- --quiet
- Quiet module output
- --ui
- Force launching GUI dialog
Parameters:
- inputs=name[,name,...] [required]
- Name of the input space time raster datasets
- expression=string [required]
- Spatio-temporal mapcalc expression
- method=name[,name,...]
- The method to be used for sampling the input dataset
- Options: start, during, overlap, contain, equal, follows, precedes
- Default: equal
- output=name [required]
- Name of the output space time raster dataset
- basename=basename [required]
- Basename for output raster maps
- A numerical suffix separated by an underscore will be attached to create a unique identifier
- nprocs=integer
- Number of r.mapcalc processes to run in parallel
- Default: 1
t.rast.mapcalc performs spatio-temporal
mapcalc expressions on maps of temporally sampled space time
raster datasets (STRDS). Spatial and temporal operators and internal
variables are available in the expression string. The description of
the spatial operators, functions and internal variables is available in
the
r.mapcalc manual page. The temporal
functions are described in detail below.
This module expects several parameters. All space time raster datasets
that are referenced in the mapcalc expression must be listed
in the inputs option. The first space time raster
dataset that is listed as input will be used to temporally sample all
other space time raster datasets. The temporal sampling method can be
chosen using the method option. The order of the STRDS's in
the mapcalc expression can be different to the order of the STRDS's in
the input option. The resulting space time raster dataset must be
specified in the output option together with the basename
of generated raster maps that are registered in the resulting
STRDS. Empty maps resulting from map-calculation are not registered by
default. This behavior can be changed with the -n flag. The
flag -s can be used to assure that only spatially related maps
in the STRDS's are processed. Spatially related means that temporally
related maps overlap in their spatial extent.
The module t.rast.mapcalc supports parallel processing. The option
nprocs specifies the number of processes that can be started in
parallel.
A mapcalc expression must be provided to process the temporal
sampled maps. Temporal internal variables are available in addition to
the r.mapcalc spatial operators and functions:
The supported internal variables for relative and absolute time are:
- td() - This internal variable represents the size of the
current sample time interval in days and fraction of days for absolute
time, and in relative units in case of relative time.
- start_time() - This internal variable represents
the time difference between the start time of the sample space time
raster dataset and the start time of the current sample interval or instance.
The time is measured in days and fraction of days for absolute time,
and in relative units in case of relative time.
- end_time() - This internal variable represents
the time difference between the start time of the sample space time
raster dataset and the end time of the current sample interval. The
time is measured in days and fraction of days for absolute time,
and in relative units in case of relative time.
The end_time() will be represented by null() in case of a time instance.
The supported internal variables for the current sample interval or instance
for absolute time are:
- start_doy() - Day of year (doy) from the start time [1 - 366]
- start_dow() - Day of week (dow) from the start time [1 - 7],
the start of the week is Monday == 1
- start_year() - The year of the start time [0 - 9999]
- start_month() - The month of the start time [1 - 12]
- start_week() - Week of year of the start time [1 - 54]
- start_day() - Day of month from the start time [1 - 31]
- start_hour() - The hour of the start time [0 - 23]
- start_minute() - The minute of the start time [0 - 59]
- start_second() - The second of the start time [0 - 59]
- end_doy() - Day of year (doy) from the end time [1 - 366]
- end_dow() - Day of week (dow) from the end time [1 - 7],
the start of the week is Monday == 1
- end_year() - The year of the end time [0 - 9999]
- end_month() - The month of the end time [1 - 12]
- end_woy() - Week of year (woy) of the end time [1 - 54]
- end_day() - Day of month from the start time [1 - 31]
- end_hour() - The hour of the end time [0 - 23]
- end_minute() - The minute of the end time [0 - 59]
- end_second() - The second of the end time [0 - 59].
The
end_* functions are represented by the null() internal variable
in case of time instances.
We will discuss the internal work of
t.rast.mapcalc with an
example. Imagine we have two STRDS as input, each one of monthly
granularity. The first one
A has 6 raster maps (a3 ... a8)
with a temporal range from March to August. The second STRDS
B
has 12 raster maps (b1 ... b12) ranging from January to December. The
value of the raster maps is the number of the month from their interval
start time. Dataset
A will be used to sample dataset
B to create a dataset
C. We want to add all maps with
equal time stamps if the month of the start time is May or June,
otherwise we multiply the maps. The command will look as follows:
t.rast.mapcalc input=A,B output=C basename=c method=equal \
expression="if(start_month() == 5 || start_month() == 6, (A + B), (A * B))"
The resulting raster maps in dataset C can be listed with t.rast.list:
name start_time min max
c_1 2001-03-01 00:00:00 9.0 9.0
c_2 2001-04-01 00:00:00 16.0 16.0
c_3 2001-05-01 00:00:00 10.0 10.0
c_4 2001-06-01 00:00:00 12.0 12.0
c_5 2001-07-01 00:00:00 49.0 49.0
c_6 2001-08-01 00:00:00 64.0 64.0
Internally the spatio-temporal expression will be analyzed for each
time interval of the sample dataset A, the temporal functions will be
replaced by numerical values, the names of the space time raster
datasets will be replaced by the corresponding raster maps. The final
expression will be passed to r.mapcalc, resulting in 6 runs:
r.mapcalc expression="c_1 = if(3 == 5 || 3 == 6, (a3 + b3), (a3 * b3))"
r.mapcalc expression="c_2 = if(4 == 5 || 4 == 6, (a4 + b4), (a4 * b4))"
r.mapcalc expression="c_3 = if(5 == 5 || 5 == 6, (a5 + b5), (a5 * b5))"
r.mapcalc expression="c_4 = if(6 == 5 || 6 == 6, (a6 + b6), (a6 * b6))"
r.mapcalc expression="c_5 = if(7 == 5 || 7 == 6, (a7 + b7), (a7 * b7))"
r.mapcalc expression="c_6 = if(8 == 5 || 8 == 6, (a8 + b8), (a8 * b8))"
Semantic labels present in the sample dataset A will be transferred to
the output dataset.
The following command creates a new space time raster dataset
january_under_0 that will set to null all cells with
temperature above zero in the January maps while keeping all the rest
as in the original time series. This will change the maximum values
of all January maps in the new STRDS as compared to the original one,
tempmean_monthly.
t.rast.mapcalc input=tempmean_monthly output=january_under_0 basename=january_under_0 \
expression="if(start_month() == 1 && tempmean_monthly > 0, null(), tempmean_monthly)"
# print minimum and maximum only for January in the new strds
t.rast.list january_under_0 columns=name,start_time,min,max | grep 01-01
name|start_time|min|max
january_under_0_01|2009-01-01 00:00:00|-3.380823|-7e-06
january_under_0_13|2010-01-01 00:00:00|-5.266929|-0.000154
january_under_0_25|2011-01-01 00:00:00|-4.968747|-6.1e-05
january_under_0_37|2012-01-01 00:00:00|-0.534994|-0.014581
# print minimum and maximum only for January in the original strds,
# note that the maximum is different
t.rast.list tempmean_monthly columns=name,start_time,min,max | grep 01-01
2009_01_tempmean|2009-01-01 00:00:00|-3.380823|7.426054
2010_01_tempmean|2010-01-01 00:00:00|-5.266929|5.71131
2011_01_tempmean|2011-01-01 00:00:00|-4.968747|4.967295
2012_01_tempmean|2012-01-01 00:00:00|-0.534994|9.69511
t.rast.mapcalc supports semantic label filtering similarly
to
t.rast.list. In
example below a new STRDS will be created and filled by NDVI products.
t.rast.mapcalc inputs=test.S2_8,test.S2_4 output=ndvi basename=ndvi \
expression="float(test.S2_8 - test.S2_4) / (test.S2_8 + test.S2_4)"
For more information about semantic label concept
see
i.band.library module.
r.mapcalc,
t.register,
t.rast.list,
t.info
Temporal data processing Wiki
Sören Gebbert, Thünen Institute of Climate-Smart Agriculture
SOURCE CODE
Available at:
t.rast.mapcalc source code
(history)
Latest change: Thursday Jan 26 14:10:26 2023 in commit: cdd84c130cea04b204479e2efdc75c742efc4843
Main index |
Temporal index |
Topics index |
Keywords index |
Graphical index |
Full index
© 2003-2024
GRASS Development Team,
GRASS GIS 8.4.1dev Reference Manual