For reanalysis data (as opposed to weather observed at the gauge stations), see Auffhammer et al. (2013).
Hijmans et al. (2005) compile the climate data for the whole land surface on earth (excluding Antarctica) at a spatial resolution of 30 arc seconds (about 1km). Max and min temperature and monthly rainfall are available. This is NOT the time-series weather data, but useful for learning the general climate at the very disaggregated level of areas.
Downloadable at this website.
Used by Dell, Jones, and Olken (2009) and Gennaioli et al (2013).
Global Historical Climatology Network
The dataset includes monthly records by rainfall stations and the location of such stations. Used by Maccini and Yang (2006) for monthly precipitation data in Indonesia.
CRU TS 2.1
Monthly climate dataset by 0.5 degree grid for 1901-2002.
Variables available include precipitation, daily mean temperature, monthly average daily maximum temperature, monthly average daily minimum temperature, diurnal temperature range, vapour pressure, cloud cover, wet day frequency, frost day frequency.
To obtain other weather variables such as dew points from the CRU data, see its FAQ no. 5.
See Mitchell and Jones (2005) for detail.
The CRU data is the best estimate of spatial distribution of weather at each point in time. To use the CRU data for time-series analysis, however do read this. Especially, bear in mind that if no station data is available, the average value for the month from 1960-1990 is imposed. Also, changes in weather over time may reflect not only actual weather changes but also changes in the availability of station data.
The original file is downloadable at Mitchell's website (but you need to use Unix to browse the data file).
The aggregate data at the country level is also available as "TYN CY 1.1".
The CGIAR Consortium for Spatial Information (CGIAR-CSI) provides the GIS version of CRU TS 2.1 data. Even if you do not intend to use the GIS software, this dataset is useful because weather data files (in the comma delimited ascii format) are split into 6 20-year periods so that you can read these files in Excel, which does not allow you to read more than 256 columns.
These data files consist of columns entitled "value" (0.5 by 0.5 degree grid identifier) and "M
y " (weather value for of ). The mapping between "value" and geographic coordinates is available in "coordinates.txt". See CRU_21-readme.doc for details.
Monthly temperature data at a 5-degree grid spatial resolution from 1850 to present. Downloadable here. Used by Bluedorn et al. (2009).
Global Six Century Temperature Patterns
This dataset provides annual temperature at a 5 degree spatial resolution from 1730-1980. ASCII files are downloadable here. For the documentation, see Mann et al. (1998). Used by Bluedorn et al. (2009).
Terrestrial Air Temperature and Precipitation: Monthly and Annual Time Series (1950-1999)
Constructed by Center for Climatic Research, University of Delaware.
Only the average values during 1950-1999 are available.
Used by Seema Jayachandran (2006) "Air Quality and Early-Life Mortality: Evidence from Indonesia's Wildfires".
For possible concerns to use this dataset for Africa, see footnote 7 of Brückner and Ciccone (2009).
Climate Prediction Center Merged Analysis of Precipitation (CMAP)
Daily average rainfall data at 2.5 by 2.5 degree grid level, available monthly since 1979.
Used by Thomson et al. (2005) to estimate the impact of rainfall on malaria incidence in Botswana.
Global Precipitation Climatology Project (GPCP)
Rainfall data similar to CMAP.
Used by Miguel et al. (2004) to instrument per capita income growth in relation to civil war incidence.
Rainfall data at the daily frequency at 1.0 by 1.0 degree grid level is also available since October 1996.
Cramer and Leemans's CLIMATE data
"Monthly averages of mean temperature, temperature range, precipitation, rain days and sunshine hours for the terrestrial surface of the globe, gridded at 0.5 degree longitude/latitude resolution" are available for the period 1930-1960.
"The CLIMWAT database includes data from a total of 3262 meteorological stations from 144 countries."
Monthly data for 28,100 stations on evapotranspiration, precipitation, sunshine, temperature, vapour pressure, and wind speed.