Remove stale EMOS documentation
This commit is contained in:
2569718930@qq.com
+1
-1
@@ -49,7 +49,7 @@
|
||||
- 一键部署脚本：deploy.sh + deploy.ps1
|
||||
|
||||
### 移除
|
||||
- 删除 LGBM 全部代码和模型文件，EMOS 简化为纯 legacy 高斯分桶
|
||||
- 删除 LGBM 全部代码和模型文件，概率路径收口为 legacy 高斯分桶
|
||||
- 删除 Polymarket 价格拉取与 UI 层（MarketDecisionLine）
|
||||
- 删除 Groq、Meteoblue、NMC、俄罗斯 pogodaiklimat 数据源
|
||||
- 删除预热（prewarm）系统
|
||||
|
||||
@@ -40,14 +40,13 @@ Public docs center: `/docs/intro` on the main site (bilingual product documentat
|
||||
- Hong Kong uses CoWIN station `6087` (Po Leung Kuk Choi Kai Yau School) as the 1-minute reference-station curve, with HKO 10-minute observations kept as the official meteorological layer.
|
||||
- Telegram airport/runway pushes are bilingual by default and use settlement-endpoint runway temperatures for slope/current/summary copy.
|
||||
- Runtime state, cache, and core offline training/backfill flows now use SQLite as the primary path; legacy JSON/JSONL files remain only for migration, export, and explicit fallback input.
|
||||
- EMOS/CRPS calibration is wired and trainable, but production should stay on `legacy` or `emos_shadow`; `emos_primary` is only for candidates that pass local offline evaluation and manual rollout.
|
||||
- Intraday analysis is now positioned as a professional meteorology read: headline, confidence, base/upside/downside paths, next observation point, evidence chain, failure modes, and confirmation rules.
|
||||
- Intraday modal now blocks stale cached detail during refresh, so users do not briefly trade off old city/date data before full detail arrives.
|
||||
- Terminal chart/detail workflow now combines settlement observations, DEB hourly consensus, model context, probability distribution tooltips, and market-bucket mapping without blocking the chart on AI text generation.
|
||||
- Terminal data uses page memory cache, browser `localStorage`, backend short-TTL cache, SSE patch replay, and foreground refresh so returning from another tab restores the latest visible chart state quickly.
|
||||
- Market bucket matching now uses the full `all_buckets` surface and strict exact / range / or-higher / or-lower direction checks, reducing bad matches to unreasonable tail buckets.
|
||||
- The market-signal difference means `model probability - market-implied probability`; positive values indicate weather probability above market pricing, while negative values indicate the YES is already priced more fully.
|
||||
- Calibrated model probability is now the primary probability panel. It shows the active production probability engine (legacy Gaussian or EMOS), while model consensus remains a secondary reference.
|
||||
- Calibrated model probability is now the primary probability panel. It shows the active legacy Gaussian probability engine, while model consensus remains a secondary reference.
|
||||
- Non-Hong Kong airport cities now ingest `TAF` and parse `FM / TEMPO / BECMG / PROB30/40`.
|
||||
- Temperature chart now overlays `TAF Timing` markers near the expected peak window.
|
||||
- Trade cue now combines upper-air structure, `TAF`, market crowding, and `edge_percent`.
|
||||
@@ -71,7 +70,7 @@ See: [AGPL-3.0 & Commercial Boundary](docs/OPEN_CORE_POLICY.md)
|
||||
- Aggregates observations and forecasts for 51 monitored cities.
|
||||
- Uses DEB (Dynamic Error Balancing) to blend multi-model highs.
|
||||
- Builds a DEB-weighted hourly consensus path for peak-window logic and chart display.
|
||||
- Generates settlement-oriented calibrated probability buckets (`mu` + bucket distribution) via legacy Gaussian or EMOS/CRPS calibration.
|
||||
- Generates settlement-oriented calibrated probability buckets (`mu` + bucket distribution) via the legacy Gaussian calibration path.
|
||||
- Adds terminal chart/detail workflows that combine live observations, DEB-centered high-temperature context, market-bucket mapping, and model-market difference.
|
||||
- Shows calibrated Gaussian context in chart tooltips as `mu` plus the full temperature-range probability distribution, without reintroducing probability bands into the main temperature view.
|
||||
- Reuses one analysis core across web dashboard and Telegram bot.
|
||||
@@ -139,7 +138,7 @@ npm run dev
|
||||
- Hong Kong keeps `HKO` official readings in dashboard and history, without falling back to airport METAR lines.
|
||||
- Intraday analysis now separates meteorology conclusion, evidence chain, invalidation rules, confirmation rules, calibrated probability, and market reference.
|
||||
- `TAF` is used as an airport-side confirmation layer, not as the main temperature model.
|
||||
- Calibrated probability uses legacy Gaussian (default) or EMOS/CRPS when evaluated; model vote counts remain an explanatory consensus line, not the final probability.
|
||||
- Calibrated probability uses the legacy Gaussian path; model vote counts remain an explanatory consensus line, not the final probability.
|
||||
- Browser extension remains a lightweight monitoring + basic-bias product, while the site holds the full analysis experience.
|
||||
- Realtime terminal charts use SSE patches plus replayable event storage; full HTTP detail remains the authoritative snapshot.
|
||||
- Chart observations are shown in the city's local time, not the browser timezone.
|
||||
@@ -160,19 +159,6 @@ POLYWEATHER_REDIS_REQUIRED=true
|
||||
|
||||
For local development or a strict single-process fallback, keep `POLYWEATHER_EVENT_STORE=sqlite`.
|
||||
|
||||
## EMOS Local Training
|
||||
|
||||
Do not run full EMOS retraining on a small VPS. The VPS should collect data and load approved calibration files; training should run on a local/dev machine using a copied production SQLite database:
|
||||
|
||||
```powershell
|
||||
scp root@38.54.27.70:/var/lib/polyweather/polyweather.db E:\web\PolyWeather\data\polyweather-prod.db
|
||||
$env:POLYWEATHER_DB_PATH="E:\web\PolyWeather\data\polyweather-prod.db"
|
||||
$env:POLYWEATHER_RUNTIME_DATA_DIR="E:\web\PolyWeather\artifacts\local_runtime"
|
||||
python scripts\auto_retrain_probability_calibration.py --verbose --snapshot-limit 50000
|
||||
```
|
||||
|
||||
Promote a generated `default.json` only when `auto_retrain_report.json` has `ready_for_promotion=true`, and prefer `emos_shadow` before enabling `emos_primary`.
|
||||
|
||||
## Ops Verification
|
||||
|
||||
### Health / system status / metrics
|
||||
|
||||
+2
-21
@@ -36,7 +36,6 @@
|
||||
- 香港默认展示 CoWIN `6087`（保良局陈守仁小学）1 分钟参考站曲线，HKO 10 分钟实测保留为官方气象层。
|
||||
- Telegram 机场/跑道推送默认中英文双语，并统一使用结算端点跑道温度计算当前值、15 分钟趋势和文案。
|
||||
- 运行态状态、缓存与核心离线训练/回填链路已完成 SQLite 主路径收口；legacy JSON/JSONL 仅保留给迁移、导出与显式回退输入。
|
||||
- EMOS/CRPS 校准链路已接通，但生产主概率保持 `legacy` 或 `emos_shadow`；`emos_primary` 只在本地离线评估通过并手动灰度后启用。
|
||||
- 官方增强站网已统一接入：
|
||||
- `MGM`（土耳其）
|
||||
- `CMA/NMC`（中国内地）
|
||||
@@ -53,7 +52,7 @@
|
||||
- 终端数据同时使用页面内存缓存、浏览器 `localStorage`、后端短 TTL 缓存、SSE patch replay 和前台恢复刷新；从其他选项卡切回时会优先恢复最新可见图表状态。
|
||||
- 市场温度桶匹配已改为完整 `all_buckets` 映射，按 exact / range / or higher / or lower 方向严格匹配，避免把天气中枢错配到不合理尾部桶。
|
||||
- 市场信号中的“模型-市场差”口径为 `模型概率 - 市场隐含概率`，正值表示天气概率高于市场报价，负值表示市场已经更充分计价。
|
||||
- 概率区已改为”校准模型概率”；默认展示生产概率引擎输出（legacy 高斯或 EMOS），模型共识作为辅助参考。
|
||||
- 概率区已改为“校准模型概率”；默认展示 legacy 高斯概率引擎输出，模型共识作为辅助参考。
|
||||
- 今日日内结构解读以规则与结构化信号为主，AI 文案只作为可降级辅助层，不替代实测、DEB、TAF 或结算逻辑。
|
||||
- 前端设计系统全面重构：统一 CSS token 体系、消除 !important 滥用（134→49）、合并断点（18→10）、数百处硬编码颜色迁移至 CSS 变量、添加 ARIA 无障碍属性和键盘导航。完整审查记录见 `docs/frontend-ui-design-review.md`。
|
||||
|
||||
@@ -72,7 +71,7 @@
|
||||
- 聚合 51 个监控城市的实测与预报数据。
|
||||
- DEB（Dynamic Error Balancing）融合多模型最高温。
|
||||
- 构建 DEB 加权小时共识曲线，用于峰值窗口判断和图表默认 DEB 展示。
|
||||
- 输出结算导向校准概率分布（`mu` + 温度桶），通过 legacy 高斯或 EMOS/CRPS 校准引擎。
|
||||
- 输出结算导向校准概率分布（`mu` + 温度桶），通过 legacy 高斯校准路径。
|
||||
- 天气决策台把结构化实况、DEB 高温路径、完整市场温度桶和模型-市场差放进图表/详情工作流。
|
||||
- 图表 tooltip 展示校准高斯上下文：`mu` 加完整温度区间概率分布，不把概率温度带重新放回主图。
|
||||
- Web 仪表盘与 Telegram Bot 复用同一分析内核。
|
||||
@@ -149,24 +148,6 @@ POLYWEATHER_REDIS_REQUIRED=true
|
||||
|
||||
本地开发或严格单进程兜底可使用 `POLYWEATHER_EVENT_STORE=sqlite`。
|
||||
|
||||
## EMOS 本地训练流程
|
||||
|
||||
低配 VPS 只负责采集、服务和加载已通过评估的参数，不建议在 VPS 上跑 EMOS 全量训练。训练前先从 VPS 拉 SQLite 副本到本地：
|
||||
|
||||
```powershell
|
||||
scp root@38.54.27.70:/var/lib/polyweather/polyweather.db E:\web\PolyWeather\data\polyweather-prod.db
|
||||
```
|
||||
|
||||
本地训练：
|
||||
|
||||
```powershell
|
||||
$env:POLYWEATHER_DB_PATH="E:\web\PolyWeather\data\polyweather-prod.db"
|
||||
$env:POLYWEATHER_RUNTIME_DATA_DIR="E:\web\PolyWeather\artifacts\local_runtime"
|
||||
python scripts\auto_retrain_probability_calibration.py --verbose --snapshot-limit 50000
|
||||
```
|
||||
|
||||
只有 `auto_retrain_report.json` 中 `ready_for_promotion=true` 时，才允许把候选 `default.json` 传回 VPS，并优先以 `emos_shadow` 观察。
|
||||
|
||||
## 运维验收
|
||||
|
||||
### 健康与系统状态
|
||||
|
||||
@@ -518,7 +518,7 @@
|
||||
}
|
||||
|
||||
/* ══════════════════════════════════════════════════════════════
|
||||
SCAN TERMINAL OVERRIDES — EMOS DESK
|
||||
SCAN TERMINAL OVERRIDES
|
||||
══════════════════════════════════════════════════════════════ */
|
||||
|
||||
.root :global(.scan-terminal) {
|
||||
|
||||
@@ -40,10 +40,10 @@ export function runTests() {
|
||||
|
||||
const allDocsZh = publicDocSlugs.map((slug) => pageText(slug, "zh-CN")).join("\n");
|
||||
const allDocsEn = publicDocSlugs.map((slug) => pageText(slug, "en-US")).join("\n");
|
||||
for (const staleTerm of ["从地图进入", "机会榜", "日历", "EMOS", "LGBM", "城市决策卡", "付费判断台", "刷新锁"]) {
|
||||
for (const staleTerm of ["从地图进入", "机会榜", "日历", "LGBM", "城市决策卡", "付费判断台", "刷新锁"]) {
|
||||
assert(!allDocsZh.includes(staleTerm), `public Chinese docs should not expose stale term: ${staleTerm}`);
|
||||
}
|
||||
for (const staleTerm of ["map-launched", "opportunity board", "calendar", "EMOS", "LGBM", "city decision card", "paid decision workspace", "refresh lock"]) {
|
||||
for (const staleTerm of ["map-launched", "opportunity board", "calendar", "LGBM", "city decision card", "paid decision workspace", "refresh lock"]) {
|
||||
assert(!allDocsEn.includes(staleTerm), `public English docs should not expose stale term: ${staleTerm}`);
|
||||
}
|
||||
|
||||
|
||||
@@ -54,7 +54,7 @@ def test_probability_snapshot_repository_recent_rows(tmp_path, monkeypatch):
|
||||
'raw_sigma': 1.1,
|
||||
'max_so_far': 14.9,
|
||||
'peak_status': 'before',
|
||||
'probability_mode': 'emos_shadow',
|
||||
'probability_mode': 'legacy',
|
||||
'prob_snapshot': [{'v': 15, 'p': 0.6}],
|
||||
'shadow_prob_snapshot': [{'v': 15, 'p': 0.4}],
|
||||
})
|
||||
|
||||
Reference in New Issue
Block a user