{"finding":{"findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","status":"published","tier":1,"rule":{"name":"forest_loss","version":"1.0","params":{"days":90,"minAlerts":2630,"minConfidence":"high","minHa":263,"minValidPct":60,"ndviDrop":0.1}},"title":"Forest loss, Velasco, Santa Cruz (Bolivia) — tile 3/4: 9699.25 ha in 90 d","summary":"819268 GFW integrated deforestation alerts (≥ high confidence) covering 9699.25 ha between 2026-06-30 and 2026-09-28; AOI alert density is 1.074× its 12 km neighbourhood. Awaiting optical confirmation.","geometry":{"type":"Polygon","coordinates":[[[-62.15,-17.41],[-60.82,-17.41],[-60.82,-15.44],[-62.15,-15.44],[-62.15,-17.41]]]},"bbox":[-62.15,-17.41,-60.82,-15.44],"aoi":{"id":"bol-8-14-3","name":"Velasco, Santa Cruz (Bolivia) — tile 3/4","tags":["forest-hotspot","discovered","bol","bolivia","santa-cruz","velasco"]},"context":{"enso":{"phase":"El Niño","oni":1.8},"events":[],"baseline":{"metric":"alert ha per deg²","ringKm":12,"aoiValue":3701.863,"regionalValue":3447.152,"ratio":1.074},"notes":["Living value (benefit transfer, order of magnitude): 9699.25 ha of tropical forest ≈ $65,159,562/yr alive (band $19,136,620–$252,442,380, 2020 USD), ≈ $6,515,956,200 over 100 yr ($2,808,269,716 NPV @ 2 %). Source: Costanza R. et al. et al. — a floor for one kind of value, not a price."]},"blindSpots":["Coverage is 30°N–30°S only — zero alerts outside the tropics means 'not monitored', not 'no loss'.","Cannot distinguish legal clearing, fire, storm damage and selective logging from illegal clearing; attribution needs concession/protected-area layers (M3).","Alerts lag reality by ~1–2 weeks; a burst just before the sweep is undercounted.","Regrowth and plantation harvest cycles look like loss; the baseline ring only partly compensates.","Persistent cloud can delay optical confirmation for months (SAR-only RADD alerts stay 'candidate')."],"observedAt":"2026-09-28T00:00:00Z","createdAt":"2026-09-28T00:00:56.529Z","updatedAt":"2026-09-28T01:29:43.492Z","createdBy":"system:forest_loss@1.0","confirmed":{"at":"2026-09-28T00:00:56.529Z","independence":"sensor","signal":{"id":"s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28","kind":"scene","source":"sentinel-2-l2a","datetime":"2026-09-28T00:00:00Z","method":{"name":"eo_compare","version":"1.0","params":{"bbox":[-62.15,-17.41,-60.82,-15.44],"composite":"median","dateA":"2026-05-16","dateB":"2026-09-28","index":"NDVI","width":256}},"summary":"Median-composite NDVI changed -0.229 (2026-05-16 → 2026-09-28); valid pixels 100% / 100%. Scenes: [object Object], [object Object], [object Object], [object Object], [object Object], [object Object]","values":{"deltaNdvi":-0.2286013821261399,"validPctA":100,"validPctB":100}}},"evidence":[{"id":"gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28","kind":"alert","source":"gfw-integrated-alerts","datetime":"2026-09-28T00:00:00Z","href":"https://data-api.globalforestwatch.org/dataset/gfw_integrated_alerts","method":{"name":"forest_alerts","version":"1.0","params":{"bbox":[-62.15,-17.41,-60.82,-15.44],"days":90,"minConfidence":"high"}},"summary":"819268 alerts (≥ high confidence), 9699.25 ha, 2026-06-30…2026-09-28.","values":{"alerts":819268,"ha":9699.25,"high_alerts":490005,"high_ha":5801.42,"highest_alerts":329263,"highest_ha":3897.82,"living_value_100y_npv2_usd":2808269716,"living_value_100y_usd":6515956200,"living_value_usd_yr":65159562}}],"attribution":null,"narration":{"v":1,"eventId":"01a0e5a1-c323-7866-a38e-30d6454d92f0","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T01:29:38.851Z","actor":"model:claude-opus-5","prev":"5e759c7a456c35c40ae494ed4a7bec2df91faa6d3a08063fb3e5953a921c6171","kind":"narrated","text":"9,699 ha of Bolivian forest flagged in Velasco over 90 days, with satellite greenness drop confirming\n\nBetween 30 June and 28 September 2026, integrated deforestation alerts recorded 819268 high-or-higher confidence detections across 9699.25 ha in tile 3/4 of Velasco, Santa Cruz, Bolivia. Of that, 5801.42 ha sits at high confidence and 3897.82 ha at the highest confidence level. An independent sensor check agrees: median-composite Sentinel-2 NDVI, a measure of live vegetation, fell -0.229 between 16 May and 28 September, with 100% valid pixels in both composites — so cloud is not hiding the picture here.\n\nWhat this is not: the alerts cannot tell legal clearing, fire, storm damage or selective logging apart from illegal clearing. That matters this season, because conditions are El Niño (ONI 1.8), when drought and fire can strip canopy without any clearing. Alert density here is only 1.074× the surrounding 12 km ring, so this tile resembles its neighbourhood rather than standing out. Plantation harvest and regrowth cycles can also mimic loss.\n\nWhat would change our mind: burn-scar mapping showing fire, concession or protected-area layers showing permitted activity, or later imagery showing rapid green-up consistent with regrowth or seasonal dieback.\n\nKey numbers:\n- Alerts (≥ high confidence): 819268 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- Area flagged (ha, 90 days): 9699.25 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- High-confidence area (ha): 5801.42 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- Highest-confidence area (ha): 3897.82 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- NDVI change (May→Sep): -0.229 [s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28]\n- Valid pixels, later composite (%): 100 [s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28]\n\nConfidence: medium\nCaveats:\n- Cause unknown: fire, legal clearing, storm damage and selective logging all look alike in these alerts.\n- El Niño (ONI 1.8) raises the chance that drought or fire, not clearing, drives part of the greenness drop.\n- Alerts lag ground events by about 1–2 weeks, so any burst just before 28 September is undercounted.\n- Alert density is only 1.074× the 12 km neighbourhood — this is a busy region, not an isolated anomaly.\n- Regrowth and plantation harvest cycles can register as loss; the baseline ring only partly corrects for this.\n- The dollar figures in the finding's notes are order-of-magnitude benefit transfers, not a market price or a loss estimate.","model":{"id":"claude-opus-5","provider":"anthropic"},"promptSha256":"895e3a1245d2b4c1c9842be49171b1c1bbbe8667d4a1bf9758f2a5aa101201ca","transcriptSha256":"d32f023f85efd1fc3531395defb497e359ac9c3fafbf50b29be1fec8eda54246","evidenceRefs":["gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28","s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28"]},"reviews":[{"v":1,"eventId":"01a0e5a1-d539-7da8-9efa-eb535fd35345","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T01:29:43.481Z","actor":"model:claude-sonnet-5","prev":"f7ab426178a50ecf172c498955b8104fc5f4bbe2b8b5ddbc1c13dbf96aa59e4b","kind":"reviewed","decision":"approve","tier":1,"note":"All key numbers (alerts, ha, high/highest confidence ha, NDVI delta, valid pixel %) match the evidence exactly.; Narration clearly acknowledges blind spots: cause ambiguity, El Niño confound, alert lag, baseline ring context, regrowth/plantation cycles, and caveats the monetary figures as benefit-transfer estimates not prices.; No individuals are named; only place/tile identifiers are used.; AOI is not tagged 'control' (tags include forest-hotspot, discovered, bol, etc.), consistent with a live monitoring finding, so controlAoi check as defined (must be tagged control) fails, but this is expected for a real finding—reviewing per schema literally.","verdict":"publish"}],"notifications":[],"replies":[],"retracted":null,"history":[{"at":"2026-09-28T00:00:56.529Z","kind":"created","status":"candidate","actor":"system:forest_loss@1.0"},{"at":"2026-09-28T00:00:56.529Z","kind":"confirmed","status":"confirmed","actor":"system:forest_loss@1.0"},{"at":"2026-09-28T01:29:38.851Z","kind":"narrated","status":"confirmed","actor":"model:claude-opus-5"},{"at":"2026-09-28T01:29:43.481Z","kind":"reviewed","status":"confirmed","actor":"model:claude-sonnet-5"},{"at":"2026-09-28T01:29:43.492Z","kind":"status_changed","status":"published","actor":"model:claude-sonnet-5"}],"eventCount":5,"lastEventHash":"47e75cdda51b04c20eee73d2f2ad243cacc884481e079cf0f743dff64847b902"},"events":[{"v":1,"eventId":"01a0e551-bae4-7b77-b915-057e27f4e726","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T00:00:56.529Z","actor":"system:forest_loss@1.0","prev":null,"kind":"created","rule":{"name":"forest_loss","version":"1.0","params":{"days":90,"minAlerts":2630,"minConfidence":"high","minHa":263,"minValidPct":60,"ndviDrop":0.1}},"title":"Forest loss, Velasco, Santa Cruz (Bolivia) — tile 3/4: 9699.25 ha in 90 d","summary":"819268 GFW integrated deforestation alerts (≥ high confidence) covering 9699.25 ha between 2026-06-30 and 2026-09-28; AOI alert density is 1.074× its 12 km neighbourhood. Awaiting optical confirmation.","tier":1,"geometry":{"type":"Polygon","coordinates":[[[-62.15,-17.41],[-60.82,-17.41],[-60.82,-15.44],[-62.15,-15.44],[-62.15,-17.41]]]},"bbox":[-62.15,-17.41,-60.82,-15.44],"aoi":{"id":"bol-8-14-3","name":"Velasco, Santa Cruz (Bolivia) — tile 3/4","tags":["forest-hotspot","discovered","bol","bolivia","santa-cruz","velasco"]},"observedAt":"2026-09-28T00:00:00Z","evidence":[{"id":"gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28","kind":"alert","source":"gfw-integrated-alerts","datetime":"2026-09-28T00:00:00Z","href":"https://data-api.globalforestwatch.org/dataset/gfw_integrated_alerts","method":{"name":"forest_alerts","version":"1.0","params":{"bbox":[-62.15,-17.41,-60.82,-15.44],"days":90,"minConfidence":"high"}},"summary":"819268 alerts (≥ high confidence), 9699.25 ha, 2026-06-30…2026-09-28.","values":{"alerts":819268,"ha":9699.25,"high_alerts":490005,"high_ha":5801.42,"highest_alerts":329263,"highest_ha":3897.82,"living_value_100y_npv2_usd":2808269716,"living_value_100y_usd":6515956200,"living_value_usd_yr":65159562}}],"context":{"enso":{"phase":"El Niño","oni":1.8},"events":[],"baseline":{"metric":"alert ha per deg²","ringKm":12,"aoiValue":3701.863,"regionalValue":3447.152,"ratio":1.074},"notes":["Living value (benefit transfer, order of magnitude): 9699.25 ha of tropical forest ≈ $65,159,562/yr alive (band $19,136,620–$252,442,380, 2020 USD), ≈ $6,515,956,200 over 100 yr ($2,808,269,716 NPV @ 2 %). Source: Costanza R. et al. et al. — a floor for one kind of value, not a price."]},"blindSpots":["Coverage is 30°N–30°S only — zero alerts outside the tropics means 'not monitored', not 'no loss'.","Cannot distinguish legal clearing, fire, storm damage and selective logging from illegal clearing; attribution needs concession/protected-area layers (M3).","Alerts lag reality by ~1–2 weeks; a burst just before the sweep is undercounted.","Regrowth and plantation harvest cycles look like loss; the baseline ring only partly compensates.","Persistent cloud can delay optical confirmation for months (SAR-only RADD alerts stay 'candidate')."]},{"v":1,"eventId":"01a0e551-e42f-7c07-b7e2-3a3b9dc44ad0","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T00:00:56.529Z","actor":"system:forest_loss@1.0","prev":"47a9ecf4e2d624b6ff50ccef9982198996da619c0f17b1ee0f25f4c66d43850b","kind":"confirmed","signal":{"id":"s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28","kind":"scene","source":"sentinel-2-l2a","datetime":"2026-09-28T00:00:00Z","method":{"name":"eo_compare","version":"1.0","params":{"bbox":[-62.15,-17.41,-60.82,-15.44],"composite":"median","dateA":"2026-05-16","dateB":"2026-09-28","index":"NDVI","width":256}},"summary":"Median-composite NDVI changed -0.229 (2026-05-16 → 2026-09-28); valid pixels 100% / 100%. Scenes: [object Object], [object Object], [object Object], [object Object], [object Object], [object Object]","values":{"deltaNdvi":-0.2286013821261399,"validPctA":100,"validPctB":100}},"independence":"sensor"},{"v":1,"eventId":"01a0e5a1-c323-7866-a38e-30d6454d92f0","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T01:29:38.851Z","actor":"model:claude-opus-5","prev":"5e759c7a456c35c40ae494ed4a7bec2df91faa6d3a08063fb3e5953a921c6171","kind":"narrated","text":"9,699 ha of Bolivian forest flagged in Velasco over 90 days, with satellite greenness drop confirming\n\nBetween 30 June and 28 September 2026, integrated deforestation alerts recorded 819268 high-or-higher confidence detections across 9699.25 ha in tile 3/4 of Velasco, Santa Cruz, Bolivia. Of that, 5801.42 ha sits at high confidence and 3897.82 ha at the highest confidence level. An independent sensor check agrees: median-composite Sentinel-2 NDVI, a measure of live vegetation, fell -0.229 between 16 May and 28 September, with 100% valid pixels in both composites — so cloud is not hiding the picture here.\n\nWhat this is not: the alerts cannot tell legal clearing, fire, storm damage or selective logging apart from illegal clearing. That matters this season, because conditions are El Niño (ONI 1.8), when drought and fire can strip canopy without any clearing. Alert density here is only 1.074× the surrounding 12 km ring, so this tile resembles its neighbourhood rather than standing out. Plantation harvest and regrowth cycles can also mimic loss.\n\nWhat would change our mind: burn-scar mapping showing fire, concession or protected-area layers showing permitted activity, or later imagery showing rapid green-up consistent with regrowth or seasonal dieback.\n\nKey numbers:\n- Alerts (≥ high confidence): 819268 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- Area flagged (ha, 90 days): 9699.25 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- High-confidence area (ha): 5801.42 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- Highest-confidence area (ha): 3897.82 [gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28]\n- NDVI change (May→Sep): -0.229 [s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28]\n- Valid pixels, later composite (%): 100 [s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28]\n\nConfidence: medium\nCaveats:\n- Cause unknown: fire, legal clearing, storm damage and selective logging all look alike in these alerts.\n- El Niño (ONI 1.8) raises the chance that drought or fire, not clearing, drives part of the greenness drop.\n- Alerts lag ground events by about 1–2 weeks, so any burst just before 28 September is undercounted.\n- Alert density is only 1.074× the 12 km neighbourhood — this is a busy region, not an isolated anomaly.\n- Regrowth and plantation harvest cycles can register as loss; the baseline ring only partly corrects for this.\n- The dollar figures in the finding's notes are order-of-magnitude benefit transfers, not a market price or a loss estimate.","model":{"id":"claude-opus-5","provider":"anthropic"},"promptSha256":"895e3a1245d2b4c1c9842be49171b1c1bbbe8667d4a1bf9758f2a5aa101201ca","transcriptSha256":"d32f023f85efd1fc3531395defb497e359ac9c3fafbf50b29be1fec8eda54246","evidenceRefs":["gfw-integrated-bol-8-14-3-2026-06-30..2026-09-28","s2-ndvi-median-bol-8-14-3-2026-05-16..2026-09-28"]},{"v":1,"eventId":"01a0e5a1-d539-7da8-9efa-eb535fd35345","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T01:29:43.481Z","actor":"model:claude-sonnet-5","prev":"f7ab426178a50ecf172c498955b8104fc5f4bbe2b8b5ddbc1c13dbf96aa59e4b","kind":"reviewed","decision":"approve","tier":1,"note":"All key numbers (alerts, ha, high/highest confidence ha, NDVI delta, valid pixel %) match the evidence exactly.; Narration clearly acknowledges blind spots: cause ambiguity, El Niño confound, alert lag, baseline ring context, regrowth/plantation cycles, and caveats the monetary figures as benefit-transfer estimates not prices.; No individuals are named; only place/tile identifiers are used.; AOI is not tagged 'control' (tags include forest-hotspot, discovered, bol, etc.), consistent with a live monitoring finding, so controlAoi check as defined (must be tagged control) fails, but this is expected for a real finding—reviewing per schema literally.","verdict":"publish"},{"v":1,"eventId":"01a0e5a1-d544-7ffc-ad49-f35dce5a59b1","findingId":"01a0e551-bae4-7b6d-9412-eaccbf96a9de","at":"2026-09-28T01:29:43.492Z","actor":"model:claude-sonnet-5","prev":"557a8e72b5fbdf6fb1ba1ef56a2cad6a15203720b63c63194dc692ab2760a131","kind":"status_changed","from":"confirmed","to":"published","reason":"autonomous publish: narrated by model:claude-opus-5, reviewed by model:claude-sonnet-5 (policy 2026-09-26-autonomous)","gates":{"narratedBy":"model:claude-opus-5","reviewedBy":["model:claude-sonnet-5"],"policy":"2026-09-26-autonomous"}}],"inclusion":{"index":689,"leafHash":"4b0587b3113fce4724057a799b2097a3fd05fe328fdc0aa956e210cfb147d250","proof":["01a3e268942ab898cbfcfb92ec199222089596b54137a25aa21b49a4e044b48f","cc3b6a93943026ec363964448148b522027c9226dda1d92f82a330f80085eafc","e73b714623ca3c51edc8dec3d398981aac9f22b1cad5f7cccd86a51c06606f14","1a4096fef3f408fb9611c7ec6bd5555c2d3e4c444b2d4d1825a326726d13c72a","89ea47a2db6f92bde86a283891634b3c2ce72e93ad607587cac01f8e71117f35","7dca4cb46e918bc2462caff163cbc3833e9d56e1231ed15deafc73303f4232da","5539526203a0f9a3fe1a73a93ff155b59f2e165868d80dfa2e6993baad67874b","4dcd70c622d414afaa3e0c21c3cf5233386d63f463994a07fff4e5e92c390f72","5f7580b7f5a7f09ff7675565fb36c19876079a6c101f721419a5a07a7a23b1c1"],"size":714,"root":"23ed27dda9a2da2abe3dbd3567303dfd6d402bbc077a4783e4e79b987c5fec16"}}