AccuWeather: Historical Forecast Weather Data

AccuWeather's Historical Forecast gives data science, quantitative research, and analytics teams the proprietary AccuWeather forecast record, captured daily and preserved exactly as it was issued, so models train on the same signal that drives them in production.

This is not a snapshot of publicly available model output. It is the proprietary AccuWeather forecast, shaped by AccuWeather meteorologists and proven to deliver Superior Accuracy™. It is the same forecast delivered to over 1 billion users, the forecast that influenced what consumers bought, when they bought it, and where demand surged.

With 90+ daily forecast parameters across nearly 90,000 global locations going back to 2020, including day/night part and hourly breakdowns out to 120 hours and daily forecasts out to 15 days, teams can train predictive models on the forecast that actually drove behavior, and validate prescriptive decisioning systems against real forecast confidence and lead-time behavior before going to production.

<p><br/></p>

## **What’s included:**

- 90+ daily historical forecast weather parameters
- 15 days of forecasts snapshotted each day
- Filtered to your specific locations of interest
- Records dating back to 2020, updated every day

<p><br/></p>

## **How it works:**

1. Activate your free trial to access the most comprehensive weather database available.
2. Collaborate with AccuWeather data experts to build a custom dataset aligned to your goals.
3. Receive secure access to your custom dataset through a Snowflake private listing.

<p><br/></p>

## **Why AccuWeather**

**The forecast consumers actually saw.** AccuWeather's forecast reach over 1 billion users. Consumer purchase timing, store visits, and replenishment demand respond to AccuWeather forecasts. Training on the AccuWeather historical forecast record means modeling the same signal that shaped behavior, rather than observed weather or a public model snapshot.

<p><br/></p>

**Proprietary forecast, not a public model snapshot.** AccuWeather's Historical Forecast is the proprietary AccuWeather forecast, refined by AccuWeather meteorologists and drawn from over 170 forecast inputs. It is not a snapshot of publicly available model output and cannot be reproduced by re-running a public model.

<p><br/></p>

**Preserved as issued.** Each snapshot is the forecast as it existed at issue time, including lead time and parameters available at decision time. Training and production share the same information environment, which matters for quantitative trading, automated operations, supply planning, and other decision systems that require defensibility.

<p><br/></p>

**Superior Accuracy™.** AccuWeather's forecast accuracy is verified as superior to other sources in independent studies. More than half of the Fortune 500 use AccuWeather, with a 97% annual renewal rate.

原始信息汇总

数据集：AccuWeather: Historical Forecast Weather Data

描述

该数据集提供 AccuWeather 专有历史预报 的快照记录，旨在为数据科学、量化研究和分析团队提供与实际业务中驱动行为的相同预报信号。数据按日捕获并保持发布时的原始状态，并非公开模型输出的快照，由 AccuWeather 气象学家优化，经证明具有 Superior Accuracy™ 准确性。该预报服务于超过10亿用户。

关键特征

• 数据来源：AccuWeather 专有预报，融合超过170个预报输入，由气象学家优化。
• 时间范围：数据记录可追溯至2020年，每日更新。
• 地理覆盖：覆盖全球近90,000个地点。
• 预报粒度：
  • 日/夜分段预报及逐小时预报，最长至120小时。
  • 每日预报最长至15天。
• 参数数量：超过90个每日历史预报天气参数。
• 个性化：可筛选至用户指定的具体关注位置。

包含内容

• 超过90个每日历史预报天气参数。
• 每日存档的15天预报快照。
• 可根据用户需求过滤地点。
• 数据从2020年开始，每日更新。

数据字典

• 核心视图示例：VW_CITY_DAYNIGHT_IMPERIAL_V1_0
• 列数量：共81列。
• 示例列：
  • LOCATION_NAME, POSTAL_CODE, ADMIN_CODE, COUNTRY_CODE (位置信息)
  • LATITUDE, LONGITUDE (地理坐标)
  • DATETIME_SNAPSHOT, DATE_METEOROLOGICAL, DAY_FLAG (时间信息)
  • CLOUD_COVER_PERC_AVG, HUMIDITY_RELATIVE_AVG, TEMPERATURE_AVG, TEMPERATURE_DEW_POINT_AVG (气象要素)
  • HAS_PRECIPITATION, HAS_RAIN, HAS_SNOW, PRECIPITATION_LWE_TOTAL, SNOW_TOTAL (降水相关)
  • DEGREE_DAYS_COOLING, DEGREE_DAYS_HEATING, DEGREE_DAYS_GROWING (度日指数)
  • SOLAR_IRRADIANCE_TOTAL, INDEX_UV_AVG (太阳与紫外线)
  • PHRASE_LONG, PHRASE_SHORT (天气描述短语)
  • 及 ICE_LWE_TOTAL, WIND_SPEED_AVG 等更多参数。

业务需求与应用

• 需求预测：基于消费者实际看到的预报训练需求、产量和结果模型。
• 库存管理：在天气驱动的需求激增前进行库存调配、促销策划和广告投放。
• 风险分析：基于历史预报记录部署天气驱动的交易、对冲和风险策略。
• 自动化决策：针对真实历史预报行为校准置信度阈值，驱动推荐系统。
• 警报与操作触发：在投入生产前，利用多年历史预报验证规则，减少误报。

使用示例

• 查询需求：查询2025年2月1日亚特兰大（邮政编码30303）白天时段的历史预报如何随时间变化。
• SQL示例： sql SELECT * FROM VW_CITY_DAYNIGHT_IMPERIAL_V1_0 WHERE DATE_METEOROLOGICAL = 2025-02-01 AND DAY_FLAG = D AND POSTAL_CODE = 30303 ORDER BY LOCATION_NAME, DATE_METEOROLOGICAL, DATETIME_SNAPSHOT;

定价与访问

• 定价模式：联系销售 (Contact Sales) 获取定价。
• 交付方式：安全共享 (Secure Share)。
• 刷新频率：每小时 (Hourly)。
• 时间覆盖：按小时 (By hour)。
• 试用：提供免费试用，激活后与 AccuWeather 数据专家合作构建定制数据集，并通过 Snowflake 私有列表安全访问。

提供商信息

• 提供商：AccuWeather® Data Suite
• 联系邮箱：
  • 销售：[e-mail]
  • 支持：[e-mail]
• 类别：需求预测、库存管理、风险分析、天气。
• 法律条款：标准 (Standard)。