Development of external Regeneration Models for FVS

Development of external Regeneration Models for FVS – another wrench in the toolkit Don Robinson ESSA Technologies Vancouver, Canada

Development History A collaborative effort involving • • Alan Ager Umatilla National Forest Duncan Wilson, Doug Maguire OSU Nick Crookston Rocky Mountain Research Station Abdel-Azim Zumrawi BC Ministry of Forests and Range • Valerie Le. May, Peter Marshall UBC • ESSA Technologies Ltd. Vancouver, BC

Have you noticed… Despite over 30 years of development, new extensions and variants, there are very few full-featured regeneration models in FVS Ferguson, Stage & Boyd Ferguson & Crookston Ferguson & Carlson 1986 1984 1991 1993 For. Sci. INT 161 INT 279 INT 467 Why is that?

Regeneration is a Hard Problem • Inventory measured on-the-ground, not remotely • Can require numerous samples (12, 000 : : 500) • Disturbances may be slow, history may be fuzzy • Regen cues may be subtle, fine-scale, site specific • Regen may be delayed or stutter unevenly Conceptual model development takes time

Who Cares Anyway? Past - emphasis on predicting timber values over single rotation Present – the times, they are a changing - multiple ecosystem values - natural disturbances: insects, fire, disease - longer time periods Future – the times, they are a changing; again - complex, largely unknown relationships between climate, trees, disturbance, invasive species …

What Are The Options? 1. Simple Manual You are the all knowing expert: thin and plant

What Are The Options? 2. Logical Conditions You understand rules governing what will happen

What Are The Options? 3. Complex rules You use the Event Monitor to create a complex model Duncan Wilson & Doug Maguire, OSU (2002) • mimics Ferguson model: habitat types, site effects, level of disturbance; P(stocking), P(species x); predicts seedlings/ac • implemented through Event Monitor keywords • >950 keyword lines for each stand in PPE • very hard to debug

A New Approach 4. Create a “non-FVS” external model and use FVS to run it • To drive the external model, create a new keyword in the Establishment family – Add. Trees • Text files pass information each way – ES 1, ES 2 • Simplifies and localizes changes to FVS code • External model can be in any language; could be a batch file, web-service, R-script … • 2 examples …

A New Approach – Ex 1 Logistic-probabilistic model (Wilson & Maguire) Patterned after the Ferguson model … predicts seedlings/ac • habitat types • site effects • intensity of harvest disturbance

A New Approach – Ex 1 ES 1 BM-FFE-PPE Regeneration Parse user preferences Input P(stocking) Harvesting & thinning Total regen stems Diameter, height growth Growth cycles Number species Mortality P(species x) Regeneration Regen stems species x Crown Summary results Constrain regeneration observations ES 2 BM-FFE-PPE Regeneration System BM-ESTAB

A New Approach – Ex 1 – 0 – years delay after Condition becomes true 5 – years delay to schedule the activities (planting) 1 – flag to indicate which set of information to pass

A New Approach – Ex 1 *. ES 1 – from FVS out to the external model –

A New Approach – Ex 1 *. ES 2 – from the external model back to FVS – • uses pre-existing FVS code to process activity schedules • 431 = NATURAL planting keyword … other values are the usual fields: date, species, stems/ac, etc.

A New Approach – Ex 2 MSN – Regeneration database model (Le. May, Marshall, Zumrawi, Hassani, Froese, Lencar, Froese, Boisvenue) • • compare simulated stand after disturbance against a database of regeneration observed in actual disturbed stands canonical selection of closest match assign matching stand’s regeneration currently >1, 000 sites, >36, 000 regen measurements (2006)

A New Approach – Ex 2 Prognosis. BC Regeneration ES 1 Prognosis. BC Version 3 (SIBC 3) Input Parse user preferences Regeneration DB Harvesting & thinning Diameter, height growth Growth cycles Regeneration System IB-ESTAB Query database Construct current stand summary N regeneration stand summaries Construct MSN input files X-variables file Y-variables file Mortality Run MSN – user preferences Regeneration Crown Regeneration DB Query database 1 - or k- plot regeneration observations Constrain regeneration observations Summary results ES 2 Prognosis. BC Regeneration MSN For-use file Fitting stats diagnostic output file

A New Approach – Ex 2 – • • slope position and site preparation 0 – years delay to schedule the activities (planting) 2 – flag to indicate which set of information to pass program to call and arguments: location of db, location of MSN, distance algorithm to use

A New Approach – Ex 2 MSN input files are automatically generated and run; output files parsed; database is re-queried and results returned to FVS

A New Approach – Ex 2 Challenges … • Reconciliation of stand structure of selected stand with ‘known’ structure of modeled stand • Change in stand structure between disturbance and later overstory inventory (snags, falldown, growth) • Research team now looking for better prediction approaches … perhaps through linkage to a detailed process model that is sensitive to fine structure in the stand

Challenges : : Opportunities • Simpler models may gloss over details • Development of conceptual models will always be iterative and messy • Conceptual and empirical models are mutually dependent … what to measure, how to measure, when to measure • Empirical models will still require substantial field work and validation

Challenges : : Opportunities • Interface between FVS and regeneration models is simple to create and requires only minor code changes in one file • External regen model specs also modest – any program in any language that reads an input file of FVS state information and returns a file of planting (or other) activities in the expected format • Prototyping is simple … FVS not required