North American historical monthly spatial climate dataset, 1901-2016

Area of Science:

• Climatology
• Environmental Science
• Geospatial Modeling

Background:

• Historical climate data are crucial for understanding long-term environmental changes and impacts.
• Previous spatial models of temperature and precipitation had limitations in accuracy and scope.
• The National Oceanic and Atmospheric Administration's National Centres for Environmental Information (NCEI) provides essential climate datasets.

Purpose of the Study:

• To develop and present improved historical monthly spatial models of temperature and precipitation for North America.
• To provide reliable climate data for the continental United States and Canada spanning 1901-2016.
• To compare the accuracy of the new models against previous spatial modeling efforts.

Main Methods:

• Utilized the North American dataset version "j" from NOAA NCEI.
• Modeled monthly minimum/maximum temperature and precipitation data from 1901 to 2016.
• Employed spatial modeling techniques, assessing predictive error using Root Generalized Cross Validation (RTGCV).

Main Results:

• The new models demonstrate reduced error compared to Natural Resources Canada (NRCAN) models from 2006.
• Average RTGCV was 0.94°C for maximum temperature, 1.3°C for minimum temperature, and 25.2% for precipitation.
• Mean absolute predictive errors were 0.7°C (max temp), 1.02°C (min temp), and 13.3 mm (precipitation).

Conclusions:

• The developed spatial models offer enhanced accuracy for historical temperature and precipitation data in North America.
• These models provide a valuable resource for climate research, impact assessment, and environmental planning.
• The findings highlight advancements in geospatial climate modeling capabilities.