Summarizing FIA estimates by attributes based on current, previous, midpoint, or accounting values

Analysts often require change estimates (e.g., change in number of trees) summarized by condition (e.g., owner) or tree (e.g., diameter) attribute values (EVALIDator page, row and column classification variables). Attributes often change between plot visits so the change estimates can be summarized by attribute values assigned from the previous visit (time 1), current visit (time 2), midpoint between visits, or by the previous and current visits combined. The accounting method identifies the technique in which change estimates are summarized by the combination of previous (time 1) and current (time 2) attribute values. The accounting method assigns a negative to the time 1 estimate and classifies it with the time 1 attribute value. A positive is assigned to the time 2 estimate, and it is classified with the time 2 attribute value. The accounting method is covered in more detail further below.

An analyst must determine whether to summarize estimates using previous, current, midpoint, or accounting attribute values. In EVALIDator, this means choosing a classification temporal basis. The information that follows is not meant to address every analysis but provides some guidance when choosing a classification temporal basis.

If the values for an attribute should not change, are usually only pertinent at time 2, or very rarely change over time (e.g., State, county, species, evalid, aspect), then summarize by the current values for most analyses. Sometimes these attributes do actually change or the assigned value varies by inventory even though no real change occurred. Even if there is an invalid difference in the attribute value over time, the most current attribute value is more than likely the correct or best value to use. For example, an incorrectly assigned tree species is likely to be correctly identified in the more recent inventories.

Tree removals and mortality are events. These events can significantly change the condition, so it is generally acceptable to presume the removal or mortality occurred while that attribute held its time 1 value. For example, consider a forest condition that was a large stand-size class at time 1 then a harvest occurred prior to time 2. After the harvest at the time 2 visit, the condition is a small stand-size class. In such a scenario, it is logical that the removals were from the large stand-size class recorded at time 1, and not from the small class as recorded at time 2.

At times, inventories (evaluations) will be available for a given change estimate but links from the inventory to the previous inventory will not be available. In these cases, previous temporal basis choices will be available in EVALIDator but summarizing by previous attribute values will provide no estimates (nulls in report), provide estimates classified with non-descriptive classification values (e.g., other), or the summary will fail and report an error. Inventories that have re-measurement estimates where both plot visits used the annual inventory design will usually have current and previous information and summaries by previous attribute values should be available. The inventories where estimates can be successfully summarized by previous attribute values varies by region, State, and land basis (forest, timberland or non-forest). Valid summaries by previous attribute values are available for most Southern Research Station inventories, even those prior to the 2000s. As early as 2004, successful summaries can be reported for some Northern Research Station inventories and success with Pacific Northwest and Rocky Mountain Research Stations varies.

Tree-level information was usually not collected on non-timberland conditions (less productive forest land, reserved forest land, or non-forest) for inventories prior to 1999 and some later inventories. As a result, some summaries for forest land change estimates by previous attribute values may not include non-timberland conditions. Summarizing by previous and current land use will identify these situations. At present, many condition attributes and all tree attributes are not available for non-forest land in EVALIDator.

Land use change should be considered in most analyses (see **Land use change description for more info). Due to the influence of diversion on removals, the influence of reversion on net growth, and the influence of both on net change, it is important to include summaries of removals, net growth, and net change by previous and current land use or land class.

Summarizing attributes by the midpoint in time between inventories is another option. However, this option is only valid, at this time, for tree diameter and stand age classes. The midpoint in time between inventories is when the removal or mortality event is presumed to have occurred so summaries by the midpoint tree diameter and midpoint stand age are appropriate. Midpoint diameter is calculated from previous diameter (available in FIADB table) and midpoint stand age is calculated by dividing the re-measurement period between plot visits by 2 and adding the resulting quotient to the previous stand age. EVALIDator uses these calculated midpoints when Mid-point if available else previous is chosen for the temporal basis with diameter and stand age classes. At this time, choosing any other summary attribute in EVALIDator with the midpoint temporal basis results in summaries by previous attribute values (when previous values are available). When links to the previous inventory are not available, as described earlier, midpoint attribute values will not be available even though EVALIDator offers the midpoint temporal basis.

Even though it's logical to summarize removals and mortality by previous or midpoint attribute values (diameter and stand age at this time), summarizing by current attribute values may make the most sense even when the attribute routinely changes. For example, determining a location for a new mill or wood-using facility may require summaries by current owner, current land use, and other current conditions.

EVALIDator has choices current if available else previous, previous if available else current, and midpoint if available else current. The else * clause could be masking important information about the analysis. For example, if there were no previous plot visits associated with an inventory and previous if available else current is chosen then the report will show estimates classified by attribute values but all the classifications will be based on the time 2 plot visits.

The accounting temporal basis is usually available for net growth, net change, and gross growth estimation where the time 1 and time 2 visits used the annual inventory design. These associated inventories are identified with a net growth accounting flag (pop_eval.growth_acct = Y). EVALIDator ensures these inventories are the only ones available when the radio button associated with Limit evaluations to only those that can do growth accounting is chosen.

Net growth is change over time minus mortality. Net change is net growth minus removals. Gross growth is net growth plus mortality. In general, the accounting method measures shifts in an estimate across a classification attribute between the previous and current inventory. For example, to account for biomass that moves into and out of a single diameter class between time 1 and time 2, the accounting method assigns the time 1 (a negative value) and time 2 (a positive value) biomass to the initial and final diameter classes. This accounts for biomass shifts into and out of each diameter class, biomass change by diameter class. The midpoints for stand age and tree diameter are used in place of time 2 for mortality and removal trees in EVALIDator accounting estimation. Normally, use the midpoint temporal basis for summaries of removals and mortality by tree diameter or stand age when comparing to estimates generated with the accounting method.

Example : The following is an example of the accounting method for a single tree assigning biomass net growth by diameter class in the MI 2017 inventory:

In the current plot visit there is a 7-inch d.b.h. red maple (7-8.9 inch diameter class) with a dry merchantable biomass of 150 pounds. In the previous plot visit, approximately 6 years ago, the tree was 6.8 inches d.b.h. (5-6.9 inch diameter class) at 140 pounds. The accounting method assigns -140 pounds to the 5-6.9 inch diameter class and +150 pounds to the 7-8.9 inch diameter class.

The accounting method is usually used to summarize attributes that change on a continual basis such as tree diameter, stand age, tree size, stand size, and height. Many tree and condition attributes are not temporally continuous. For example, ownership changes, but it is not temporally continuous. Normally, analysts should use the previous temporal basis for attributes that are not temporally continuous or use current attribute values as described earlier. However, the accounting method will run and can be appropriate with any attribute depending on the analysis. The analyst will have to use best judgement when working with attributes such as stocking class which may or may not be temporally continuous.

The most logical temporal basis for summarizing net growth, net change, and gross growth may also depend on the other estimates and temporal choices available or used in an analysis. As previously stated, only stand age and diameter class midpoints are available for removal and mortality trees. For other attributes, summaries for removals and mortality are only available by past or current values depending upon the situation. So analysts may wish to report net growth by the same past or current attributes they are using for the mortality and removal estimates. For example, to calculate a net growth to removals ratio by tree size class one would need to summarize by previous tree size (there is no current tree size on a removals tree and none of the FIA tools calculate a tree size class at the midpoint in time).

** Land use change description

The removals component of change comprises harvest and diversion (other) removals. When the land basis is forest land, as opposed to timberland, diversion includes trees over area that diverted from forest to non-forest between the previous and current inventory and is presumed to have happened at the midpoint in time between inventories. Tree size at the midpoint is modeled from the tree size at the previous inventory. When the land basis is timberland, diversion also includes land diverted either to reserved forest land (removed from timber production by statute or administrative designation) or to less productive (other) forest land.

The tree biomass at the midpoint between inventories is counted in removals and there are relatively few plots with removals, so even the biomass attributed by one diversion plot can significantly contribute to total removals or net change (net growth minus removals) estimates. When a tree is harvested and used for a product on diversions, the tree is counted in harvest removals and not as diversion or other removals.

Net growth includes reversion. When the land basis is forest land, reversion includes trees over area that reverts from non-forest to forest between the previous and current inventory. When the land basis is timberland, reversion includes trees over area that reverts from non-forest, reserved forest land or other (less productive) forest land to timberland. Reversions are presumed to have happened at the midpoint between inventories and tree size is modeled at this midpoint based on the tree size in the current inventory.

Since reversion counts the tree biomass at the modeled midpoint between inventories, reversion trees can have a substantial impact on net growth estimates as compared to the impacts from growth on survivor trees (only counts change in biomass on tree from previous to current inventory), given equal tree biomass.