Skip to content

Ratio Estimates

The oft overlooked ratio estimator is a powerful component of FIA analyses, enabling the estimation of ratios that can normalize estimates to forested areas, yield averages of tree-level quantities, estimate mortality rates, and more. An understanding of the more basic status estimates vignette is also required.

As the name suggests, the ratio estimator is formed by dividing a numerator estimate by a denominator estimate, making an analysis dependent on pairs of variables. In fiaplyr, this is accomplished by using a function within the estimate context called ratio, where users specify numerator and denominator attributes. Both attributes are evaluated from the same handler. The handler’s domains apply to the numerator and, by default, the denominator. Use den_partitions when the denominator requires different domain variables.

To illustrate the power of the ratio estimator, we will provide three examples:

  1. Growing stock volume divided by forested area.
  2. Diameter height ratio by species.
  3. Proportional representation of species.

Growing Stock Volume divided by Forested Area

Section titled “Growing Stock Volume divided by Forested Area”

Normalizing estimates by forested area is a common practice in FIA analyses. For example, in states with sparse forest cover, per-acre densities that can be made with the standard post-stratified estimator may seem small to the lay user because the density is expressed over the entire state, rather than only for forested lands within it. The ratio estimator resolves this by dividing the per-acre density by the proportion of forested land, inflating the estimate to represent the density exclusively within forests. This is straightforward using the ratio helper. First, establish a handler

library(fiaplyr)
library(DBI)
library(duckdb)
library(dplyr)
# Connect to the Vermont mini database
con <- dbConnect(duckdb(), fiadb_vt_mini_path())
# Create a handler for the 2003 to 2006 evaluation for Vermont status variables
handler <- eval_handler(con, 500601)

Next, we subset to growing stock trees on forested land, and then specify the ratio arguments. The first argument forms the output numerators. In this case, we have not defined a partition, so this represents just one estimate, representing the growing stock volume across Vermont. Then, we specify the denominator as cond(), which estimates the forested proportion noting that we subset to COND_STATUS_CD == 1 to only include forested conditions.

handler |>
subset(tree(TREECLCD == 2), cond(COND_STATUS_CD == 1)) |>
estimate(
ratio(tree(VOLCFNET), cond())
)
# A query: ?? x 4
# Database: DuckDB 1.5.5
var_n var_d estimate se
<chr> <chr> <dbl> <dbl>
1 VOLCFNET prop 1808. 41.9

Here, we see the density of growing stock per acre on forested land, which is 1,808 cubic feet per acre. The table defines the numerator variable var_n, and the denominator variable var_d. The point estimate and standard error are provided.

A representation of the slenderness of trees is the diameter-height ratio, which has implications for taper, merchantability, and other characteristics of trees. We will estimate the diameter height ratio for growing stock trees within the 10 to 15 inch diameter class for each species. By default, the ratio estimator generates estimates for all possible domain pairings, but for this question it is sufficient to only generate ratios when the species match between the numerator and denominator, this can be done by using domain_pairing = 'matched' argument.

tree_handler <- handler |>
subset(tree(TREECLCD == 2, DIA >= 10, DIA <= 15, ACTUALHT == HT)) |>
partition(tree(SPCD))
dhr_ests <- tree_handler |>
estimate(
ratio(
tree(HT),
tree(DIA)
)
) |> filter(SPCD_n == SPCD_d)
# Prepare readable species labels
species_labels <- handler@tables$ref_species |>
select(SPCD, COMMON_NAME)
dhr <- dhr_ests |>
arrange(desc(estimate)) |>
filter(se != 0) |>
left_join(species_labels, by = c("SPCD_n" = "SPCD"))
head(dhr)
# A query: ?? x 7
# Database: DuckDB 1.5.5
# Ordered by: desc(estimate)
SPCD_n SPCD_d var_n var_d estimate se COMMON_NAME
<dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 823 823 HT DIA 6 0.000000101 bur oak
2 832 832 HT DIA 4.59 0.135 chestnut oak
3 833 833 HT DIA 5.24 0.164 northern red oak
4 901 901 HT DIA 6.73 0.000000147 black locust
5 951 951 HT DIA 5.65 0.185 American basswood
6 972 972 HT DIA 5.91 0.245 American elm

Notice that we filtered the ratio estimates to only include those where the species matched between the numerator and denominator. This is because the ratio estimator will produce estimates for all possible domain pairings, e.g., the diameter of SPCD=71 divided by the height of SPCD=125. This is useful in some contexts, but not here, so they are discarded outright.

It may be of interest to estimate the proportion a species represents within a state. This requires the denominator to omit the species domain inherited from the handler, which can be expressed with a denominator override.

# First specify a handler for growing stock trees, partitioned by species
prop_handler <- handler |>
subset(tree(TREECLCD == 2)) |>
partition(tree(SPCD))
prop_ests <- prop_handler |>
estimate(
ratio(
tree(),
tree(),
den_partitions = list(tree())
)
)
prop_ests |>
arrange(desc(estimate)) |>
left_join(species_labels, by = c("SPCD_n" = "SPCD")) |>
select(SPCD_n, COMMON_NAME, estimate, se) |>
head()
# A query: ?? x 4
# Database: DuckDB 1.5.5
# Ordered by: desc(estimate)
SPCD_n COMMON_NAME estimate se
<dbl> <chr> <dbl> <dbl>
1 823 bur oak 0.0000304 0.0000333
2 832 chestnut oak 0.000160 0.000113
3 833 northern red oak 0.00879 0.00157
4 837 black oak 0.0000180 0.0000176
5 901 black locust 0.000124 0.000131
6 951 American basswood 0.00223 0.000982

Here, we can see the proportion of each species within the state of Vermont for the specific case of growing stock trees. The same approach supports other custom ratios, such as the proportion of trees with defects or static mortality proportions.