Assisi Forest 2019 Updates
Updated: Jan 4, 2022
A list of the more important changes made to Assisi Forest during 2019.
Smaller bug fixes and interface enhancements are not generally listed.
Add Upload to Server for FPS and FIA Import
For users of SQL Server, imports of FPS or FIA now upload the MDB files to the server running SQL Server so that importing can occur much faster.
Assisi Software can Build Cloud Based System for Clients
Assisi Software can build a cloud based server for anyone wanting to run Assisi remotely. The cloud offers both remote desktops and/or remote apps. The cost is around $250/mo for the base system.
Max SDI Defaults to Major Species Max SDI if no BA in Stand
Max SDI will now default to a stand’s major species Max SDI if there are no trees or BA in a stand. Previously Max SDI would default to 0. This should have no effect on growth because stands with no trees do not process growth or mortality. The change corrects for an where a divide by 0 occurs when a stand has trees but no BA and RDI was being calculated.
Added Heteroscedastic Weighting to Linear Regressions
2019.2.6 Heteroscedasticity is a term that describes residuals of a linear regression having non constant variance over their measurement range. Ordinary least squares regression methods assume there is no heteroscedasticity or that the variability of errors is constant over the range of measurements. If the variance is not constant, then a model of the error variance itself can be used to correct for heteroscedasticity. Applying a model for error does not bias estimates of linear regression coefficients. For example, the error in tree height measurements is expected to increase with height because taller trees are more difficult to measure accurately than shorter trees. As a result, height measurements of taller trees show greater variability of error than do smaller trees. When estimating a scaling factor for a height model, (called sub sampling in Assisi), the residuals often exhibit heteroscedasticity. A weighting factor such as 1/DBH can be used to decrease error variance for taller trees. The result is a statistically similar scaling factor, but with much smaller standard error.
In the example above, the weighted linear regression (right) shows a large improvement to SE: 23 vs 3 so we can have more confidence in the corrected linear regression’s slope. The difference in the slopes themselves, .88 vs .86, is statistically insignificant because correcting for heteroskedasticity does not itself bias a result. However, if the sample errors have a relationship to DBH, then applying 1/DBH would bias the result. Height of course is dependent on DBH, but the sample error itself is assumed to be independent of DBH. This version of Assisi applies a 1/DBH weighting factor for lowering linear regression heteroscedasticity for total, merch, sawlog and crown height model calibrations.
Regress Scaling Factors Relative to BH
(2019.2.6) Many height equations estimate heights relative to breast height, not ground. For these equations it’s appropriate to regress scaling factors during Assisi’s Sub Sampling process also relative to breast height. Assisi has been regressing all heights relative to ground up to this version. Regressing scaling factors relative to breast height will produce slightly different scaling factors. The results are predicted heights different in this version than earlier versions by generally less than +-1’ for 100’+ trees with smaller differences for shorter trees. This will cause small volume differences when merch heights differ more than the scaling multiple. For example, if we assume merch is about 80% of total and the scaling multiple is 1 ft, then only a few trees with greater than 1’ differences would be expected to have a very small difference in volume. When a difference does occur, it should be considered a more accurate prediction, however. Of course, when no sub sampling is done or when scaling factors equal 1 (not significant), then there will be no difference at all with this version. Here is the difference in total heights for trees in the sample stand 070704:
Merch Height in Progress Viewer not Reflecting Assisi Added Logs
(2019.2.6) When logs called in the field sum to less than merch height, Assisi will add logs up to merch. This is common, for example, when heights are called to foreign sort specs but not to domestic sort specs. The compiler is used to estimate the remaining height from top of foreign to a domestic merch. Assisi performs this calculation correctly, but within the Progress Viewer merch height only reflected the called heights, not the added domestic height. This was a display issue only: computations and database results include the domestic heights. Merch height is now reported in the Progress Viewer as the sum of both field and Assisi added logs.
Added “Stand or Type” Option for Broken Top Regressions
(2019.2.1) Added a broken height estimation option to first use stand level regression if available then type level regressions. If broken top regressions by species and type are defined in a calibration, Assisi can use them to estimate missing broken tops. However, if broken tops are measured in the field, Assisi will run regressions and use those results. For species with no field measurements, Assisi will default to the calibration’s type level broken tops regressions.
Added Check for Stratified Host and Target Calibrations
(2019.2.1) When performing a stratified calibration, the target and host calibration cannot be the same. If a calibration (the “host”) is set to save its results (regressions and scaling factors) to another calibration (the “target”), the two calibrations cannot be same, or results will overwrite the host calibration’s models. A stratified calibration is designed to save its results by species and type to a different calibration called the target so that the target calibration can later be used on stratified sample data.
Change “Default” Defect by Species to “Additional” Defect by Species
(2019.2.0) The species level defect, set in Calibration Editor has always been a “default” rather than an “additional” defect… despite the editor label of “Add Def”. In other words, the defect entered was only used when no defect was measured in the field. If a defect was measured in the field, then the “Add Def” value was not used. This has been changed to be a true “additional defect”. Now the value entered as “Add Def” is added to any defect measured in the field. Furthermore, “Add Def” was only applied to log volumes, not tree volumes. In other words, species where Volume Method = Volume Equations or Volume Tables did not have the Add Def value applied. IE, there was no default defect unless volumes were calculated from logs. Now, Add Def is applied to Volume Equations and Volume Tables as well.
Added FPS and FVS Damage Codes
(2019.2.0) Two new fields to hold FPS and FVS damage IDs were added to the AllowedDamage and AllowedDamageSeverity tables. Assisi can new easily map FPS or FVS damage codes to its own definitions of damage codes during FPS or FVS import.
Added Save Stands with Errors and Save Stands with Warnings Options to Compile Ribbon
(2019.2.0) An option to save stands that fail or that have warnings during compile was added.
Stand Age from Non-Site Index Trees
(2019.2.0) Age measurements on trees not flagged as site index trees can now be used for stand age.
Include Broken Top Snags in Regressions
(2019.2.0) Option to include broken top heights from snags in broken height regressions was added. When broken top heights are being sub sampled, you can choose to include or exclude measured broken heights from broken top snags in the regression. Broken top heights from live trees are always included in the regression, but if you feel broken top snags are similar enough to live broken tops, you may want to include their broken heights in the regression as well.
(2019.2.0) Parsing was treating a subtraction that followed an “IF()” or “SI()” as a negation.
(2019.2.0) If all plots in a cruise were empty, compile results were treated as not cruised. I.e., the results were null. This has been changed to results being 0. Although a cruise with all empty plots is rare, a clearcut action does empty out plots until a plant action occurs. These stands will have 0, not null volume now and will average as 0’s in stand averages.
(2019.2.0) The ability to truncate field called merch or sawlog log tops to calculated merch or sawlog heights is now an option set in calibration. Truncating merch has been available, but this option renames and adds sawlog truncation.
Young Growth DBH and Total Height Model Settings
(2019.2.0) The settings for the young growth model were separated into DBH growth and total height growth fields in the Calibration Editor.
Section Lengths by 3rds
(2019.2.0) Sections can now be called by 3rds to support FPS methods. When Cruise Design Section Length is set to “Thirds”, Assisi will interpret log IDs 1, 2, and 3 as 1/3 sections of total tree height rather than a fixed section length.
Tree & Count Tree Stats Added
(2019.2.0) PSE and CV for both measure and count trees added to summary tables by stand, species and species group.
IsInPlot Changed To IsOffPlot
(2019.2.0) Trees.IsInPlot has been changed to Trees.IsOffPlot.
CashFlow, Companies Added to Operations
CashFlow and Companies tables added to Operations.
(2019.1.0) A BrokenDiameter field was added to Trees table for storing measured broken diameters along with the existing broken heights.
(2019.1.0) A TaperDiameter field was added to Trees table for storing measured taper diameter along with the existing taper heights.