Sentinel-2 multi-temporal analysis for PNW bark beetles & defoliators, Midwest white oak decline
with drought index regression, and Oregon white oak encroachment + MOB.
Bark Beetle Mortality — Year-over-year NDMI/NBR decline detects Douglas-fir beetle,
mountain pine beetle, and fir engraver. Red/grey attack = strong signal (NDMI drops 0.15-0.30).
TreeMap host matching identifies likely beetle species.
📡 Jul-Aug imagery (YoY) • NDMI + NBR change • Host species filter
Early Stress / Green Attack — Pre-visual bark beetle detection using NDRS index
(Huo et al. 2021) with CRE red-edge confirmation. Captures stress before crown symptoms appear.
📡 Jun-Aug imagery (YoY) • NDRS + CRE change • 80-88% accuracy (literature)
⚡ Advisory only — high false positive risk from drought. Flags areas for ground-checking.
White Oak Decline — Multi-factor syndrome: drought + Armillaria root rot + Hypoxylon canker
+ two-lined chestnut borer. 200K+ acres severe in Mark Twain NF alone. Detects progressive
crown thinning over 2-5 years in oak-hickory stands, regressed against GRIDMET drought index (PDSI).
📡 Aug-Sep imagery (5yr) • NDVI + NDMI trend • PDSI drought regression • Stand risk scoring
🌡️ Drought-amplified: PDSI < -2 accelerates decline. Wet/dry whiplash is especially damaging.
Oregon White Oak (Q. garryana) — Two parallel threats to declining ecosystem (<15% of historic extent).
1. Conifer encroachment: Doug-fir overtopping oaks due to fire suppression. Detected via
phenological contrast (deciduous oak vs evergreen conifer in winter imagery).
2. Mediterranean Oak Borer (MOB): Exotic invasive beetle (*Xyleborus monographus*) carrying
*Raffaelea montetyi* fungus. Detected in Multnomah/Marion/Clackamas/Washington counties OR.
📡 Winter (Jan-Mar) + Summer (Jul-Sep) • Phenological amplitude • Oak-pixel NDMI trend
🌿 Conservation value: 200+ vertebrate species depend on oak savanna habitat.
🏔️ Whitebark Pine (ESA Threatened) — Keystone high-elevation species under converging threats.
51% of standing whitebark pines are dead (USFWS 2022). Three-threat detection:
1. White Pine Blister Rust: Non-native *Cronartium ribicola* — progressive crown dieback
detected via 5-year NDVI/NDMI decline in WBP pixels. No eradication possible.
2. Mountain Pine Beetle: Targets large seed-producing trees. NDMI/NBR year-over-year change.
3. Succession: Subalpine fir/spruce overtopping WBP due to fire suppression.
📡 Jul-Aug imagery (5yr trend) • TreeMap FORTYPCD 367 + elevation >1800m • GRIDMET drought
🌱 Resistant survivor detection: Green WBP near mortality = seed collection candidates
for USFS Dorena GRC blister rust resistance screening. 1,500+ parent trees tested to date.
⚠️ ESA Section 7 consultation required for any management activity.
🌲 Lodgepole Pine MPB Survivors — Detecting trees that survived the 2000-2015 mountain pine beetle epidemic.
MPB killed ~60% of mature lodgepole across the Rockies. Surviving trees in high-mortality stands
may carry genetic resistance (thicker phloem, aggressive resin, bark chemistry).
Method: Map grey-attack mortality zones (low NDMI) in lodgepole habitat,
buffer 300m, find green canopy within buffer. Height filter (8-28m) excludes regen and Doug-fir/spruce.
📡 Jul-Sep imagery • TreeMap FORTYPCD 281 • ETH canopy height filter
🌲 GPS points exported for field verification and seed collection.
Coordinate with USFS Regional Geneticist for provenance documentation.
🌳 Legacy Tree Detector — Identifies candidate old-growth trees
in fire-adapted ecosystems. Works nationally — adapts to local forest type.
Method: Multi-tier selection using ETH canopy height (10m):
Path A (Height Dominant): ≥ site mean × 1.15 + crown peak — tallest trees in stand
Path B (Crown Shape): ≥ 20m + flat crown (SD < 3.0m) — old-growth morphology even if not tallest
Either path → candidate. LCMS rejects harvest since 1985, Landsat rejects regrowth.
Species ID — Three tiers:
🌲 Confirmed: Western larch via winter NDVI deciduous signal |
🔬 TreeMap: FIA FORTYPCD species lookup |
🏔️ Topo proxy: DEM aspect + elevation fallback
Output classes:
🌲 Tall Tree Candidate (green — passes selection, needs field verification) |
🏆 Probable Legacy (gold — converging evidence: height + crown shape + LCMS + Landsat)
📡 ETH 10m CHM (2020) • LCMS 1985-present • Landsat 1984-present • SRTM
📍 GPS candidate points exported with height, evidence score, entry path,
and species. Load into OnX, Avenza, or Google Earth for field verification.
⚠️ Screening tool — NOT a definitive old-growth map. Cannot distinguish 150yr from 250yr.
Mark candidates with GPS before any mechanical thinning operation.
🎄 Christmas Tree Farm Health — Multi-year time series analysis for planted conifer farms.
Detects progressive decline, needle retention loss (Swiss needle cast, Rhizosphaera,
Dothistroma), root rot (Phytophthora), and insect damage (balsam woolly adelgid, Diplodia).
Method: Summer NDVI (peak vigor) + Winter NDVI (needle retention) + NDMI (moisture stress)
compared across 5 years. Neighborhood z-score detects spatial anomalies. Harvest blocks discriminated
from disease mortality via contiguous-bare-area analysis.
Regions: PNW (Douglas-fir, noble fir) | SE (Fraser fir) | NE (balsam fir, spruce) | MW (Scotch pine, spruce)
📡 Jun-Aug + Dec-Feb imagery (5yr) • Region-specific disease priors • 10m resolution
📍 GPS scouting waypoints exported for declining areas — load into OnX, Avenza, or Google Earth.
Disease probability scored by region and spectral signal match.
🦋 Eastern Spruce Budworm — Choristoneura fumiferana — Most destructive
forest insect in North America. Current outbreak from Quebec (15M+ ha since 2006) is
crossing into northern Maine (3,000 ac confirmed 2024, 178K at risk 2025).
Method: Year-over-year NDMI decline (Jul-Aug) as primary index
(90% accuracy per Rahimzadeh-Bajgiran et al. 2018). Red-edge IRECI change
catches light defoliation that NDVI misses. Multi-year cumulative tracking
for mortality risk (balsam fir dies after 4-5 consecutive years).
Hosts: Balsam fir (most vulnerable) > White spruce > Red spruce > Black spruce
International: Works in Canada using ESA WorldCover + winter NDVI evergreen proxy
📡 Jul-Aug imagery • NDMI + IRECI change • TreeMap spruce-fir codes
📍 GPS scouting points exported for field verification.
Contact Maine Forest Service or SOPFIM (Quebec) for spray coordination.
Progressive Decline — 5-year peak-season NDVI + NDMI trend analysis for chronic agents:
root diseases (Armillaria, Heterobasidion), balsam woolly adelgid, Swiss needle cast.
Detects gradual canopy deterioration that single-year analysis misses.
📡 Jul-Aug imagery (5yr trend) • Linear slope analysis • All forest types
📉 Slope < -0.01/yr concerning, < -0.02 significant, < -0.03 severe
Oak Wilt — Detects CCI (Chlorophyll/Carotenoid Index) anomalies in deciduous forest.
Red oaks die within weeks; expanding pockets spread via root grafts (~12m/yr).
📡 Jun-Aug imagery • NLCD deciduous forest mask • 80-84% accuracy
Hemlock Woolly Adelgid — Tracks winter NDVI decline in evergreen canopy.
HWA causes progressive canopy thinning (4-15 years). Winter imaging isolates hemlock signal.
📡 Dec-Feb imagery • NLCD evergreen/mixed mask • Multi-year trend
🌱 Resistant survivor detection — Flags green patches near mortality for seed/cone collection
Western Conifer Defoliators — Within-season greenup anomaly detects defoliation first,
then discriminates likely agent using temporal trajectory, elevation, and severity pattern.
🦋 Species discrimination: Budworm (slow 10yr decline, mid-high elev) vs Tussock Moth (2-3yr crash, lower elev)
📡 Apr-May vs Jul-Aug greenup • Red-edge CRE confirmation • Elevation + host filter
⚠️ Tussock moth urgency: Can kill trees in 1-2 years. Check B.t.k. spray window.
Southern Pine Beetle — Detects expanding pockets of pine mortality.
SPB kills pines in characteristic spots that grow outward. Trees fade green → yellow → red → gray.
📡 Jul-Sep imagery • Year-over-year NDVI decline • SE US pine belt
⚠️ Early detection critical — Remove infested trees within weeks. Salvage logging + buffer strips.
🌲 Conifer Encroachment — Detects native conifers expanding into meadows and grasslands
due to fire exclusion. Compares Landsat baseline (~2000) to current using fall NDVI +
summer NDVI cross-check. TreeMap species composition identifies encroaching species
(juniper, Doug-fir, ponderosa, etc.).
Output classes:
🌲 Established (green — forest in both periods) |
🔄 Densification (yellow — sparse → dense) |
🔴 New encroachment (red — meadow converted to forest)
📡 Landsat 5/7/8/9 • 30m • 25-year change detection • TreeMap species ID
⚠️ Treatment: mechanical removal, mastication, or prescribed fire. Prioritize meadow edges.
💧 Water Yield Prioritization — Identifies where conifer removal will recover the most
water for aquifer recharge, spring recovery, and streamflow restoration.
Layers combined:
🌲 40yr encroachment severity (meadow→forest weighted 2×) |
💧 OpenET actual evapotranspiration vs meadow reference |
❄️ Snow sublimation model (canopy interception loss) |
🏔️ TWI + slope + aspect (recharge position) |
🗺️ NHD/NWI stream & wetland proximity |
📉 Wetland decline detection (concurrent drying signal)
Output: Priority heat map (red = highest water recovery per acre) +
protection overlay (wetlands & stream buffers for equipment exclusion) +
recoverable water estimate in gallons/acre/year.
📡 Landsat 40yr + OpenET 2016+ + PRISM + 3DEP + NHD + NWI + MODIS snow
💧 Use case: NRCS EQIP applications, municipal source-water protection, watershed restoration grants.