Climate Projections from Coupled Model Intercomparison Project (CMIP6)

Pollen has reprocessed publicly-available data from the Coupled Model Intercomparison Project (CMIP6), which is an international collaboration of dozens of climate labs that seeks to predict what will happen to the Earth's climate between now and the year 2100. If you've seen a simple graph in, say, The New York Times of how hot the Earth will be in a "business as usual" scenario vs "best case scenario", then you have already seen this data, except aggregated into a simple, global mean.

There's so much more than just global mean temperatures available, though. The scientific community has created projections for hundreds of variables (snow coverage, precipitation, etc) in resolutions as small as 100km squares and in time steps as small as daily, projected out into the year 2100. We've reprocessed a useful subset of this data so, instead of wrestling with its original, hard to use, Earth science specific format (netCDF), you can query it easily with SQL, directly in Snowflake, instead.

See https://www.pollen.io/snowflake/ for more information.

Samples/Tables Included:
- Monthly simulated "historical" climate data from 1850-2015
- Monthly projected climate data from 2015-2100 based on different potential carbon emission scenarios (called Representative Concentration Pathways and Shared Socioeconomic Pathways)
- All of the above for temperature, humidity, precipitation, and snow area coverage (with more on the way)
- Also, for convenience, a table of shapefiles of US states, for easy joining by state name
- Also, a table of all the Creative Commons licensing information you'll need to stay compliant with the original data's attribution requirements.

Fields Included (see also: https://www.pollen.io/snowflake/actual_variables.html)
- GIS-style polygon defining the 100km square on Earth
- square surface area of geographic polygon (for taking weighted averages)
- source (i.e. which climate model)
- experiment (i.e. best case vs worst case vs middle of road)
- month and year
- temperature in F and C
- precipitation in mm/day and in/day
- specific and relative humidity
- snow area coverage

Please note at all of the above (even for historical dates) are simulations, not actual, direct measurements.

Example Use Case:
Join your customer addresses and use temperature projections to see which users in your own customer list are going to experience the most summer temperature change between now and 2050. Use temperature projections to see which US states will experience the most heat stress during harvest season over 10, 20, 30 and 50 year intervals. Snow projections to see if you can still go skiing on Mount Shasta in December after the year 2040.

Pollen团队对耦合模型比较计划第六阶段（Coupled Model Intercomparison Project, CMIP6）的公开数据进行了再处理。CMIP6是由数十个气候实验室组成的国际合作项目，旨在预测2100年前地球气候的变化趋势。如果你曾在《纽约时报》等媒体上看到过“照常排放情景”与“最佳情景”下地球温升的简单折线图，那么你已经接触过该数据集的简化版本——仅聚合为全球平均温度数据。

不过，该数据集包含的远不止全球平均温度。科学界已针对数百个变量（积雪覆盖、降水等）生成了投影数据，空间分辨率最低可达100平方公里格点，时间分辨率最低可达日度，投影时长延伸至2100年。我们对其中的实用子集进行了再处理，让你无需与原本难以使用的地球科学专用格式（netCDF）打交道，而是可以直接在Snowflake中通过SQL轻松查询这些数据。

更多信息请访问 https://www.pollen.io/snowflake/。

包含的样本与数据表如下：
- 1850年至2015年的月度模拟“历史”气候数据
- 基于不同潜在碳排放情景（即典型浓度路径（Representative Concentration Pathways, RCP）与共享社会经济路径（Shared Socioeconomic Pathways, SSP））生成的2015年至2100年月度气候投影数据
- 上述全部数据涵盖温度、湿度、降水与积雪面积覆盖度（更多变量即将上线）
- 为方便使用，还附带了美国各州的形状文件表，可通过州名快速关联
- 同时提供了符合原始数据署名要求的完整知识共享（Creative Commons）许可信息表，确保数据使用合规。

包含的字段（详见：https://www.pollen.io/snowflake/actual_variables.html）：
- 定义地球100平方公里格点的地理信息系统（GIS）风格多边形
- 地理多边形的表面积（用于计算加权平均值）
- 数据来源（即所用的气候模型）
- 试验情景（即最佳情景、最差情景与中庸情景）
- 月份与年份
- 华氏度与摄氏度表示的温度
- 以毫米/日与英寸/日为单位的降水量
- 比湿度与相对湿度
- 积雪面积覆盖度

请注意，所有上述数据（即便为历史时期数据）均为模拟结果，而非直接实测数据。

典型应用场景：
将客户地址数据与该数据集关联，通过温度投影分析你的客户名单中哪些用户在2050年前的夏季温度变化最为显著；通过温度投影评估未来10、20、30及50年周期内，美国哪些州在收获季将面临最严重的热应激风险；通过积雪投影预测2040年后，是否仍可在12月前往沙斯塔山滑雪。

搜集汇总

背景与挑战

背景概述

该数据集基于CMIP6的公开气候预测数据，由Pollen重新处理，将原始netCDF格式转换为易于SQL查询的形式。它包含1850年至2100年的历史和模拟气候数据，涵盖温度、降水、湿度及积雪等多个变量，适用于分析未来气候变化对商业和地理的影响。