5-Year Forest Vigour Change
+15.2% NDVI
2021-2022 mean 0.401 → 2024-2025 mean 0.462
Vigour is up — a management-effect signal.
1. Observation results
5-year cumulative ΔCO₂
+41,566t-CO₂
NDVI → AGB empirical
Water yield change
+0.3%
~ +362,707 m³/yr
Observations
71points
Summer peak × 5 years
AOI
12,398ha
Asahi range, Himi City
2. Year-by-year NDVI trend
| Year |
Summer NDVI mean |
Obs. count |
ΔNDVI vs prev. yr |
% change |
On year-to-year variability:
Single-year comparisons fluctuate ±20-40% because of climate-year differences (heatwaves,
delayed monsoon end, etc.). A 2-3-year moving average extracts the true
management / forest-state signal. Here we use the 2021-2022 mean vs 2024-2025 mean
as the long-term trend.
3. Calculation basis
// NDVI → CO2 (empirical, Pettorelli et al. family)
ΔAGB (t/ha) = ΔNDVI × 30
ΔC (t-C/ha) = ΔAGB × 0.5
ΔCO₂ (t/ha) = ΔC × 3.67
ΔCO₂ (AOI) = ΔCO₂/ha × forest_area_ha
// NDVI → water yield (conservative)
Δwater (%) = ΔNDVI × 5 (NDVI +0.1 ≈ +0.5% yield)
Δwater (m³) = baseline_yield × Δ%
// Long-term trend
NDVI_early = mean(2021, 2022)
NDVI_late = mean(2024, 2025)
ΔNDVI_long = NDVI_late − NDVI_early
4. Compared with credit-programme MRV
A low-cost alternative to MRV:
Maintaining a J-Credit forest-absorption project requires annual monitoring, reporting, and
third-party verification, costing hundreds of thousands to millions of yen. Our method
derives the same signal automatically from public data, so the marginal
cost is near zero. Project creators, corporate buyers, and local governments all share
the same numbers — and the data lineage stays consistent.
5. Sentinel-1 SAR: all-weather, all-season monitoring
Sentinel-2 NDVI is optical, so it cannot observe through cloud, fog, or at night.
In monsoon and typhoon seasons — common in Japan, India, and much of Southeast Asia — this
is a serious gap. Combining Sentinel-1 (C-band SAR / synthetic-aperture radar)
closes it.
Cloud-penetrating
6-12days
All weather, all seasons
VH backscatter
Biomassproxy
Correlates with canopy structure / standing volume
Harvest detection
Real-timepossible
Clear-cuts drop ~-6 to -10 dB
NDVI × SAR hybrid evaluation:
Combining NDVI (optical / vigour) with VH backscatter (radar / biomass) lets us independently
estimate (1) leaf health and (2) structural standing volume.
When weather knocks out one, the other fills in — there is no monitoring gap.
6. Structural error compression
Method-improvement roadmap:
- NDVI → AGB is an empirical formula → integrate Forestry Agency LiDAR for per-stand measured biomass
- Climate-year variability → moving average + AMeDAS correction to extract the pure vegetation signal
- Large-scale harvest / natural disturbance → Sentinel-1 SAR integration for cloud- and season-independent detection (§5)
- Understorey / young-stand change → combine multi-band indices (NDMI, SAVI, etc.)
- Tier 2 → Tier 3 compresses uncertainty from ±30% to ±10-15%
7. Next steps
- Per-stand time-series differencing (Shizuoka 1,102 stands as the next batch)
- Integrate the Forestry Agency LiDAR (publicly opened for Tochigi, Hyogo, Kochi) to move to Tier 3
- A dashboard so local governments and forest cooperatives can monitor their own forests
- AMeDAS integration to remove climate-year confounds
- Automated SAR time-series differencing (harvest alerts)
8. Data sources
- Sentinel-2 L2A (NDVI time series) — CDSE Sentinel Hub Statistical API
- Sentinel-1 GRD (C-band SAR, VH polarization) — CDSE Sentinel Hub Process API
- Forestry Agency Simplified Evaluation Method Ver.1.0 (water yield baseline)
- IPCC AFOLU 2006 (CO₂ conversion factors)
- Pettorelli et al. family (empirical NDVI → AGB)
※ This report is a third-party estimate built on public satellite data. It is not an official
J-Credit / VCS certification. Actual credit issuance or renewal requires monitoring reports
prepared per the Forestry Agency / programme secretariat methodologies.