Function type: Reporting/derived function – summarises or
plots the output of
another function; it does not compute any new statistic. Not
exported. Unlike the other functions internalised alongside it, this
one has no direct single-object S3 wrapper – it takes a direction
raster (from direction_map(), also not exported) rather
than a "trend_test"/"fdr" object, so there is nothing for a method to
dispatch on. Reachable from outside the package via
plot(tst_result, which = "direction")/plot(rta_result, which = "direction") for workflow_tst()/workflow_rta() results
specifically; both this
function and direction_map() are reachable with ::: for any
other case.
Arguments
- direction
Output of
direction_map().- path
Character or
NULL. If supplied, a PNG is written there.- main
Plot title. Defaults to the generic
"Binarised trend map";workflow_tst()/workflow_rta()results pass their own"TST direction map"/"RTA direction map"instead.
References
This combination of FDR-corrected significance with trend direction is this package's own contribution, not from an external method paper; cited here as the source of the overall TST workflow it belongs to:
Gutiérrez-Hernández, O. and García, L.V. (2025) Uncovering true significant trends in global greening. Remote Sensing Applications: Society and Environment, 37, 101377. doi:10.1016/j.rsase.2024.101377
Underlying theoretical justification for the FDR-BH assumption this significance is based on:
Benjamini, Y., & Yekutieli, D. (2001) The control of the false discovery rate in multiple testing under dependency. Annals of Statistics, 29(4), 1165-1188. doi:10.1214/aos/1013699998
See also
Other FDR correction functions:
fdr_bh(),
fdr_bky(),
fdr_by(),
fdr_comparison_barplot(),
fdr_correction(),
fdr_direction_summary(),
fdr_pvalue_histogram(),
fdr_significance_maps(),
fdr_summary(),
fdr_threshold_plot()
Examples
r <- read_ordered_stack(example_data("vhp_ndvi"))
#> Temporal order auto-detected with pattern '(19[0-9]{2}|20[0-9]{2})'.
#> Automatic mode: order detected from file names. For higher reliability -- especially if the series is not annual -- supplying 'files' explicitly (with 'time' or 'cycle_type') is recommended. See ?read_ordered_stack.
#> Temporal order verification (mandatory, cannot be skipped):
#> stack_position detected_number file
#> 1 1982 VHP_SMN_annual_ndvi_1982.tif
#> 2 1983 VHP_SMN_annual_ndvi_1983.tif
#> 3 1984 VHP_SMN_annual_ndvi_1984.tif
#> 4 1985 VHP_SMN_annual_ndvi_1985.tif
#> 5 1986 VHP_SMN_annual_ndvi_1986.tif
#> 6 1987 VHP_SMN_annual_ndvi_1987.tif
#> 7 1988 VHP_SMN_annual_ndvi_1988.tif
#> 8 1989 VHP_SMN_annual_ndvi_1989.tif
#> 9 1990 VHP_SMN_annual_ndvi_1990.tif
#> 10 1991 VHP_SMN_annual_ndvi_1991.tif
#> 11 1992 VHP_SMN_annual_ndvi_1992.tif
#> 12 1993 VHP_SMN_annual_ndvi_1993.tif
#> 13 1994 VHP_SMN_annual_ndvi_1994.tif
#> 14 1995 VHP_SMN_annual_ndvi_1995.tif
#> 15 1996 VHP_SMN_annual_ndvi_1996.tif
#> 16 1997 VHP_SMN_annual_ndvi_1997.tif
#> 17 1998 VHP_SMN_annual_ndvi_1998.tif
#> 18 1999 VHP_SMN_annual_ndvi_1999.tif
#> 19 2000 VHP_SMN_annual_ndvi_2000.tif
#> 20 2001 VHP_SMN_annual_ndvi_2001.tif
#> 21 2002 VHP_SMN_annual_ndvi_2002.tif
#> 22 2003 VHP_SMN_annual_ndvi_2003.tif
#> 23 2004 VHP_SMN_annual_ndvi_2004.tif
#> 24 2005 VHP_SMN_annual_ndvi_2005.tif
#> 25 2006 VHP_SMN_annual_ndvi_2006.tif
#> 26 2007 VHP_SMN_annual_ndvi_2007.tif
#> 27 2008 VHP_SMN_annual_ndvi_2008.tif
#> 28 2009 VHP_SMN_annual_ndvi_2009.tif
#> 29 2010 VHP_SMN_annual_ndvi_2010.tif
#> 30 2011 VHP_SMN_annual_ndvi_2011.tif
#> 31 2012 VHP_SMN_annual_ndvi_2012.tif
#> 32 2013 VHP_SMN_annual_ndvi_2013.tif
#> 33 2014 VHP_SMN_annual_ndvi_2014.tif
#> 34 2015 VHP_SMN_annual_ndvi_2015.tif
#> 35 2016 VHP_SMN_annual_ndvi_2016.tif
#> 36 2017 VHP_SMN_annual_ndvi_2017.tif
#> 37 2018 VHP_SMN_annual_ndvi_2018.tif
#> 38 2019 VHP_SMN_annual_ndvi_2019.tif
#> 39 2020 VHP_SMN_annual_ndvi_2020.tif
#> 40 2021 VHP_SMN_annual_ndvi_2021.tif
#> 41 2022 VHP_SMN_annual_ndvi_2022.tif
#> 42 2023 VHP_SMN_annual_ndvi_2023.tif
#> Stack built: 42 layers, 146 x 338 cells.
#> >> [read_ordered_stack()] elapsed: 0.15 s
trend <- trend_test(r, report = FALSE, verbose = FALSE)
fdr_result <- fdr_correction(trend$stats$p, report = FALSE, verbose = FALSE)
# direction_map() and fdr_direction_plot() are used together
# internally by workflow_tst()/workflow_trends() (see their own
# source) to draw the "TST/RTA direction map" panel automatically
# when report = TRUE (the default) -- there is no standalone public
# wrapper for this specific combination outside those workflows.