# Places to check first — Koh Phangan, Surat Thani, Thailand

Generated by `scripts/11_field_priority.py` from the 2019 vs 2026 Sentinel-2 NDVI change
detection around Than Sadet–Ko Pha-ngan National Park. **Every entry is a possible change, not
confirmed clearing.** The signal rating (High or Medium) refers to the strength and size of the
satellite signal, not to confirmation that clearing occurred. The purpose of this list is to
decide where a person looks first, not to assert what happened.

**Ranking rule, in one sentence:** locations more than 250 m inside the mapped boundary
come first, High signal before Medium, largest first; everything within 250 m of the
line is kept separate until the boundary question is settled.

**Why that rule:** no reliable public digital map of the legal park boundary could be found, so the
boundary in use is an approximation and is the main source of error in this study. The list is
therefore ordered by how reliable each location is to act on, rather than by size alone.

**This ranking prioritises signal strength and distance from the uncertain boundary, not
severity.** A ranger could reasonably choose a different order depending on whether the
operational priority is certainty, severity, accessibility or something else. The rule is stated
here so it can be questioned or changed.

**Desktop check before any field visit:** each location's coordinates open directly in Google
Earth from the interactive map, so higher-resolution imagery and surrounding context can be
checked first. The map also lets you add your own pins — things the analysis did not detect — and
export them as CSV, KML or GPX.

**Taking it with you:** `sites.kml` holds all 18 locations as named placemarks grouped by tier —
open it in Google Earth or QGIS, or load it onto a handheld GPS. The map exports the same thing
plus your own pins.

## Tier 1 — check first (7 sites, 13.1 ha)
Clearly inside the mapped area, High signal. More than 250 m inside the mapped boundary,
so these remain inside the mapped area even if the boundary is wrong by that much.

| # | site | area (ha) | signal | NDVI drop | lat, lon | m from boundary |
|---|---|---|---|---|---|---|
| 1 | #2 | 5.39 | High | 0.31 | `9.71177, 100.05391` | 360 |
| 2 | #4 | 1.72 | High | 0.271 | `9.70716, 100.05839` | 682 |
| 3 | #5 | 1.54 | High | 0.333 | `9.72686, 100.04917` | 988 |
| 4 | #6 | 1.49 | High | 0.254 | `9.7077, 100.05671` | 489 |
| 5 | #7 | 1.48 | High | 0.284 | `9.73883, 100.05734` | 465 |
| 6 | #12 | 0.79 | High | 0.33 | `9.72005, 100.06834` | 398 |
| 7 | #14 | 0.72 | High | 0.345 | `9.73728, 100.05895` | 644 |

**Subtotal: 7 sites, 13.1 ha**

## Tier 2 — check next
Clearly inside the mapped area, Medium signal.

| # | site | area (ha) | signal | NDVI drop | lat, lon | m from boundary |
|---|---|---|---|---|---|---|
| 8 | #11 | 0.93 | Medium | 0.271 | `9.7611, 100.03489` | 859 |
| 9 | #15 | 0.7 | Medium | 0.306 | `9.75651, 100.05133` | 273 |

**Subtotal: 2 sites, 1.6 ha**

## Tier 3 — boundary uncertain
Within 250 m of a boundary known to be approximate. These are kept separate because
their status could change if the true park boundary is different; settle the boundary question
before acting on them.

| # | site | area (ha) | signal | NDVI drop | lat, lon | m from boundary |
|---|---|---|---|---|---|---|
| 10 | #1 | 7.41 | High | 0.571 | `9.76594, 100.0413` | 61 |
| 11 | #3 | 2.61 | High | 0.373 | `9.74162, 100.05246` | 233 |
| 12 | #8 | 1.38 | High | 0.321 | `9.69381, 100.05642` | 146 |
| 13 | #9 | 1.34 | High | 0.259 | `9.71183, 100.04942` | 134 |
| 14 | #10 | 1.05 | High | 0.336 | `9.7227, 100.04081` | 36 |
| 15 | #13 | 0.72 | High | 0.395 | `9.75946, 100.05131` | 163 |
| 16 | #16 | 0.68 | Medium | 0.298 | `9.69131, 100.05914` | 167 |
| 17 | #17 | 0.65 | High | 0.366 | `9.75513, 100.05021` | 89 |
| 18 | #18 | 0.65 | Medium | 0.269 | `9.708, 100.05433` | 228 |

**Subtotal: 9 sites, 16.5 ha**

## What to record at each site
GPS location · photos in four directions · whether there is visible clearing and what it looks
like · whether the change appears recent · cut stumps, burning, bare soil, new planting or
structures · roads or tracks giving access · who appears to be using the land · and most
importantly one of three outcomes: **confirmed / not confirmed / cannot tell**. A "not confirmed"
result is useful too: it tells us that the satellite signal did not match what was found on the
ground, which helps us understand how reliable the method is.