Carbon and REDD+ operators

`carbon.emissions`#

The carbon emitted and removed by land change over a series of maps (the IPCC's stock-difference method): every cell's stock at every date from its class (a stock per class, tC/ha; or a curve of regrowth, Chapman-Richards or linear, from the date the class appeared in the cell, for secondary forest), a fall an emission, a rise a removal; by period, by transition, and every cell's net emission over the series. Transitions happen at the later date of a period.

Input	Type	What it is
`maps`required	raster	the maps of classes, a layer per date

Output	Type	What it is
`periods`	table	a row per period: year_from, year_to, emissions, removals, net (t of CO2, or of C)
`transitions`	table	a row per period and pair of classes with cells: year_from, year_to, from, to, area_ha, emissions, removals
`stocks`	stack	the stock of every cell at every date, tC/ha
`net`	map	every cell's net emission over the series, t per hectare

Setting	Type	What it is
`stock`required	any	the stock of each class, tC/ha: {class: stock}
`regrowth`	any	the classes that grow from the date they appear: {class: {max, rate, shape, form}} (tC/ha; Chapman-Richards's max (1 - exp(-rate age))^shape, or linear: rate tC/ha a year up to max)
`co2`	bool	in tonnes of CO2 (C × 44/12; default true), else of carbon
`times`	any	the dates of the maps (default: the layers' names)
`cell_area`	number	a cell's area in m2 (default: the cells' sides)

`redd.allocate`#

VT0007's allocation (UDef-ARP's prediction): the fitted relative frequency of every modelling region (a region the fit lacks takes its class's average over the subdivisions) gives each cell a density; the densities are scaled to the activity data, the expected deforestation (or a period's observed one, to confirm the model), never above a cell's area, again until the adjustment ratio is 1.00001 or less.

Input	Type	What it is
`classes`required	map	the vulnerability classes of the period
`table`required	table	the relative frequency table (redd.fit)
`subdivisions`	map	the subdivisions, as in the fit
`expected`	number	the expected deforestation, hectares
`deforestation`	map	or the period's observed deforestation (1)

Output	Type	What it is
`density`	map	the expected deforestation of each cell, hectares
`info`	table	adjustment_ratio, iterations, expected_ha, allocated_ha, filled (regions the fit lacked)

Setting	Type	What it is
`max_iterations`	integer	the adjustments at most (default 50)

`redd.assess`#

VT0007's assessment of an allocation (UDef-ARP's model evaluation): the predicted and the actual deforestation in the square cells of an assessment grid wholly inside the jurisdiction (centred on the grid's nodes from the map's corner, as UDef-ARP's Thiessen polygons), and their agreement: the mean and median absolute errors (also as shares of a cell), the intersection over union, R², the ordinary least squares and Theil-Sen lines.

Input	Type	What it is
`density`required	map	the predicted deforestation of each cell, hectares
`deforestation`required	map	the actual deforestation (1)
`mask`	map	the jurisdiction (non-zero; default all the map)

Output	Type	What it is
`table`	table	a row per assessment cell: cell, x, y (its centre), actual_ha, predicted_ha, residual_ha
`metrics`	table	samples, mae, mae_percent, medae, medae_percent, iou_percent, agreement, difference, r2, ols_slope, ols_intercept, theil_sen_slope, theil_sen_intercept
`residuals`	map	each assessment cell's residual (predicted - actual), ha

Setting	Type	What it is
`grid_area`required	number	the area of an assessment cell, hectares

`redd.fit`#

VT0007's fit of a vulnerability map to a period's deforestation (UDef-ARP's fitting): the relative frequency of deforestation in every modelling region (a vulnerability class in a subdivision of the jurisdiction) and the density it gives every cell, in hectares.

Input	Type	What it is
`classes`required	map	the vulnerability classes (redd.vulnerability)
`deforestation`required	map	the deforestation of the period (1)
`subdivisions`	map	the subdivisions of the jurisdiction (ids 0-999)

Output	Type	What it is
`table`	table	a row per modelling region: id (class × 1000 + subdivision), class, subdivision, deforested and all its cells, frequency
`density`	map	the fitted density of deforestation, hectares per cell

`redd.nrt`#

VT0007's negligible risk threshold (UDef-ARP's nrt_calculation): the distance to the forest's edge below which 99.5 % of the deforestation of a period happened, the middle of the first bin (a cell wide) where the cumulative share of the deforested cells reaches 0.995, in whole map units.

Input	Type	What it is
`distance`required	map	the distance of the forest's cells to the edge, in map units
`deforestation`required	map	the deforestation of the period (1)
`mask`	map	the jurisdiction (non-zero; default all the map)

Output	Type	What it is
`nrt`	number	the negligible risk threshold, in map units

`redd.vulnerability`#

VT0007's vulnerability map in classes of geometric widths (UDef-ARP's geometric classification): the benchmark from the distance to the forest's edge (classes 2 to n + 1 from the edge to the negligible risk threshold, 1 beyond it), or an alternative from a map of vulnerability (a potential, a probability; classes 1 to n from the lowest), inside the jurisdiction; 0 elsewhere.

Input	Type	What it is
`distance`	map	the benchmark: the distance to the forest's edge
`nrt`	number	with the distance: the negligible risk threshold
`vulnerability`	map	or an alternative: a map of vulnerability
`forest`	map	with a vulnerability: the forest (non-zero)
`mask`	map	the jurisdiction (non-zero; default all the map)

Output	Type	What it is
`classes`	map	the vulnerability class of each cell, 0 outside

Setting	Type	What it is
`classes`	integer	the classes of geometric widths (default 29)

carbon.emissions#

redd.allocate#

redd.assess#

redd.fit#

redd.nrt#