Geospatial Data

This is a geospatial data repository for agricultural economists interested in climate, production or soil data. Everything listed here is openly accessible. Most of it is global or quasi-global, with a few Africa-specific products. For development economists with interest in other open source data sets, I would refer you to a major and comprehensive data set collection effort on DEVECONDATA.

I add datasets to this list as I find them (sometimes with a bit of a time lag…), so if you’re aware of a useful data source not included here, please send me an email! Also, I’ve only worked with a selection of these datasets, and don’t claim to be an expert on any of them. This is more like a personal, jotting-down-for-memory’s-sake list turned public. So if I’m getting anything wrong, please do write and correct me!

Re-analysis temperature and precipitation datasets: 

Interpolated temperature and precipitation datasets:

  • Global Precipitation Climatology Center (GPCC) data on precipitation. Version 7 spans 1901 to 2013, and is based on data from from 67,200 stations world-wide. Contains monthly totals at 0.5° x 0.5°, 1.0° x 1.0°, and 2.5° x 2.5° latitude by longitude.
  • Willmott and Matsuura’s Gridded Monthly Time Series V 4.01. These datasets provide monthly, interpolated temperature averages and monthly precipitation totals for the entire world, from 1900 to 2014, in grids size 0.5° latitude by 0.5°.
  • Climate Research Unit (CRU) time series datasets from the University of East Anglia. Contains monthly averages for precipitation, temperature and a few other variables. A global dataset excluding Antarctica, covering 1901-2014 and gridded at 0.5°. Data here, after applying for (instantaneous) access. More info here. The CRU also creates/provides other datasets, listed here.

Flood and Drought Indices/Data: 

  • Palmer Drought Severity Index (PDSI) from Aiguo Dai and co-authors. Modeled using NCEP climate prediction precipitation data and surface temperature data from CRU as inputs, PDSI captures atmospheric moisture (i.e. meteorological drought) through a standardized index ranging from -10 (dry) to 10 (wet). The effect of temperature on atmospheric moisture, or potential evapotranspiration, is calculated through Thornthwaite’s (1948) formula. Four years of lagged temperature and precipitation data contributed to the PDSI index of each grid-month, capturing the “build up” of drought. While atmospheric moisture is correlated with soil moisture, or agricultural drought, it is not identical; more details can be found here. The PDSI data is global, at a 2.5° spatial resolution, in monthly time-steps, and the most recent scPDSIpm data covers 1950 to 2014. Interpretation of a PDSI value depends on local mean climate conditions; each grid-month value essentially compares moisture over the last 4 years to the historical grid mean. Thus, a value of +4 might imply floods in the central US but only moderate rainfall in northern Africa.
  • Standardized Precipitation Index (SPI) from NCAR/UCAR. The SPI is the number of standard deviations by which precipitation (they use CRU) lies above or below a long-term mean. Temperature data is not incorporated. Data is global, at 1° spatial resolution, in monthly time steps, and available with “long-term mean” defined around 3-month, 6-month, and 12-month intervals. Interpretation of index values, as with PDSI, changes with mean rainfall. For example, the 6-month SPI value for each grid-month compares a moving 6-month precipitation record against the long-term (since 1948) distribution for the same 6-month period. More info here.
  • Standardized Precipitation Evapotranspiration Index (SPEI). This index is available in multiple datasets, each with “long-term mean” defined by different month-intervals (1 mo, 6 mo, etc.), like SPI. Unlike SPI, but like PDSI, SPEI also allows for temperature to effect drought conditions through potential evapotranspiration (PET). (SPEI version 1 used the Thornthwaite equation of PDSI to calculate PET; the current version uses the supposedly superior Penman-Montheith equation.) Datasets should be chosen according to analysis intent: shorter month-intervals will predict soil water content and river discharge, medium time scales relate to reservoir storage/discharge, and long time scales should predict groundwater storage. Data is global, with 0.5° spatial resolution, covering 1901-2014, and long-term means defined as anything between 1 and 38 months in the various datasets. More info here.
  • The Africa & Latin America Flood and Drought Monitors, run out of Princeton University. This effort contains a range of historical/monitored and forecasted data (hydrologic, soil moisture, precipitation, etc.) at daily, weekly or monthly time scales, gridded at 0.25° resolution. Info on data construction here.

Gridded soil/land datasets and databases: 

  • Harmonized World Soil Database from FAO, IIASA, ISRIC, ISSCAS, and JRC. This massive database provides interpolated soil quality estimates for the entire world, including nutrient availability and a number of other variables, in grids spaced at 30 arc sections (approximately 1 km).
  • African SoilGrids from AfSIS/ISRIC. This source provides data on a number of interpolated soil quality indicators (soil pH, sand, soil organic matter, cation exchange capacity, etc.) in 250-meter grids, for the African continent.
  • Soil Map of the World from FAO/UNESCO. The link is down as of March 2016, but generally if “Digital Soil Map of the World (Geonetwork)” will lead to this ESRI shapefile of soil types across the world, as well as Erdas and IDRISI files.
  • Global Land Surface Model from the Terrestrial Hydrology Research Group at Princeton. A global dataset of land surface hydrology, created via multiple land surface simulations.

Gridded crop production/suitability datasets: 

List of Further Climate Lists: 

Bonus datasets: 

  • High-resolution 2015 settlement data from the Center for International Earth Science Information Network. These layers provide human population distributions at a resolution of 1 arc-second (approximately 30 m in most areas) for the year 2015, in both rural and urban areas. At the moment, layers exist only for Burkina Faso, Ghana, Haiti, Ivory Coast, Madagascar, Malawi, Mexico, the Philippines, Rwanda, South Africa, Sri Lanka, Thailand, and Uganda.
  • Copernicus global land cover map for Africa in 2015. Apparently years 2016 and 201 are coming. More details here.