The Standardized Precipitation-Evapotranspiration Index, or SPEI, is a drought index that measures how wet or dry a location is relative to its own historical normal, using both rainfall and the atmosphere’s demand for moisture. Unlike simpler drought metrics that look only at precipitation, the SPEI factors in evapotranspiration, which is the water lost from soil and plants into the air. That addition makes it especially sensitive to the warming climate, where rising temperatures pull more moisture from landscapes even when rain patterns have not changed much. The result is a versatile, standardized number that researchers, agencies, and farmers use to compare drought conditions across different climates, seasons, and time horizons.
How the SPEI Is Calculated
At its core, the SPEI starts with a simple water balance: precipitation minus potential evapotranspiration. Potential evapotranspiration is the amount of water that would evaporate and transpire from plants if water supply were unlimited. Once that balance is computed for a given location and time window, the raw numbers get normalized against a long reference period, typically 30 years or more, so that the resulting index value tells you how unusual the current conditions are compared to what that spot has experienced historically.1První konference PERUN. Comparison of SPEI drought indexes with three different statistical distributions based on data from Slovakia A value near zero means conditions are roughly average. Negative values signal drier-than-normal conditions: a value around −1 indicates moderate drought, −1.5 severe drought, and below −2 extreme drought. Positive values mean wetter-than-normal conditions.
Because the values are standardized, a SPEI of −1.5 in the Amazon means the same relative departure from normal as −1.5 in the Sahel, even though the two places receive wildly different amounts of rain. That comparability is one of the index’s biggest selling points for global-scale research and monitoring.
Why Not Just Track Rainfall
The most widely used predecessor to SPEI is the Standardized Precipitation Index, or SPI, which works much the same way but uses only precipitation. In humid climates, where temperature-driven moisture demand is relatively modest, the two indices tend to agree closely. A global comparison found that in humid regions, SPI and SPEI correlate strongly, but the agreement falls apart in arid and semi-arid zones. In drylands, the SPEI detects roughly 18 to 27 percent more drought events than the SPI, and drought severity readings can differ by more than a full standardized unit.2Atmospheric Research. Comparisons of SPI and SPEI in capturing drought dynamics: A Global assessment across arid and humid regions
The reason is straightforward. In dry, hot regions, evapotranspiration plays a major role in the water budget. A modest dip in rainfall that might not trigger a drought signal in the SPI can register as serious when you also account for the atmosphere sucking moisture away from the soil. A separate analysis found that whether including evapotranspiration actually improves the match to observed soil-moisture conditions depends on the location and the time scale of the drought being measured, suggesting the two indices are best seen as complementary rather than one being universally superior.3CrossRef / Journal of Geophysical Research: Atmospheres. Uncertainties in Drought From Index and Data Selection
Timescales and What They Capture
One of the SPEI’s most useful features is that you can calculate it over different accumulation periods. A 1-month SPEI captures short-term moisture anomalies, the kind that stress rainfed crops in a single growing window. A 3-month SPEI smooths out weekly noise and is often used for agricultural monitoring. A 6-month or 12-month SPEI picks up longer droughts that affect reservoirs, river flows, and groundwater.4Natural Hazards. Multi-scale approach for different type of drought in temperate climatic conditions Even longer windows of 24 or 48 months exist for tracking multi-year dry spells that can deplete deep aquifers.
The choice of timescale is not just academic. A reservoir manager cares about different conditions than a wheat farmer, and the SPEI lets each user dial in the window that matches their concern. At longer accumulation periods, the difference between SPEI and SPI tends to grow, because temperature effects compound over time.2Atmospheric Research. Comparisons of SPI and SPEI in capturing drought dynamics: A Global assessment across arid and humid regions
The Evapotranspiration Method Debate
Because the SPEI depends on an estimate of potential evapotranspiration, the method used to calculate that estimate matters. The gold-standard approach is the Penman-Monteith equation, which accounts for temperature, humidity, wind speed, and solar radiation. The trouble is that many weather stations, especially in developing countries, do not record all four of those variables. So researchers often fall back on simpler formulas that rely mainly on temperature, such as the Thornthwaite or Hargreaves methods.
Studies comparing these approaches have found that the Hargreaves method generally agrees better with Penman-Monteith-based SPEI than Thornthwaite does, especially at shorter accumulation periods.5Journal of Hydrology: Regional Studies. Sensitivity of Standardized Precipitation and Evapotranspiration Index (SPEI) to the choice of SPEI probability distribution and evapotranspiration method A China-wide analysis that tested eight different evapotranspiration formulas found that the range of estimates can vary by up to two and a half fold, though both Penman-Monteith and Thornthwaite matched observed drought records well at the national scale. In arid regions during periods of rapid warming, however, the Thornthwaite method tended to paint a more pessimistic drought picture than Penman-Monteith.6Atmospheric Research. Choice of potential evapotranspiration formulas influences drought assessment: A case study in China The practical takeaway is that if you are reading a SPEI-based drought study, the choice of evapotranspiration formula can shift the results, and it is worth checking which method was used before taking the numbers at face value.
SPEI and Rising Temperatures
This is where the SPEI really earns its keep. As the planet warms, precipitation patterns shift in complex ways: some places get wetter, some get drier, and many see both extremes more often. But nearly everywhere gets warmer, and warmer air increases evaporative demand. An index that tracks only rainfall misses that signal entirely.
In the Middle East and North Africa, researchers found that SPEI showed a markedly more pronounced increase in drought frequency and severity than SPI over recent decades, driven primarily by rising air temperatures rather than decreasing rainfall.7Earth Systems and Environment. A Multidecadal Assessment of Drought Intensification in the Middle East and North Africa: The Role of Global Warming and Rainfall Deficit In critical agricultural zones globally, SPEI-detected drought intensification has been proceeding roughly twice as fast as SPI-based trends.2Atmospheric Research. Comparisons of SPI and SPEI in capturing drought dynamics: A Global assessment across arid and humid regions A study using SPEI-based distributions to compare observed drought patterns against natural-only climate simulations found strong increases in drought frequency, duration, and intensity attributable to anthropogenic forcing when evaluating net water availability.8Nature Communications. Evidence of anthropogenic impacts on global drought frequency, duration, and intensity
In other words, if you rely on precipitation-only indices in a warming world, you are likely underestimating how bad droughts have already become and how much worse they are projected to get.
Crop Yield Prediction and Agricultural Planning
For farmers and agricultural planners, the SPEI offers a direct link between climate data and crop outcomes. A recent study in the southeastern United States used county-level SPEI time series to train machine-learning models that predicted crop yields, achieving accuracy up to about 75 to 77 percent at regional and state scales depending on the model and the crop.9Farming System. Early-season crop yield prediction using machine learning and drought conditions in the southeastern United States Because SPEI values are available at different timescales and can be computed from relatively simple climate data, they lend themselves well to early-season forecasting. A negative SPEI trend during the first months of a growing season can signal trouble before visible crop damage appears.
SPEI is also being integrated into early-warning frameworks at the national level. A UK study developed a system that uses hydrometeorological indices including SPEI to forecast drought impacts operationally, concluding that even a generalized approach based on these indices provides an effective foundation for decision-making in drought risk management.10Natural Hazards and Earth System Sciences. Toward early warning of drought impacts: a framework for predicting drought impacts in the UK
Forests and Tree Mortality
Drought kills trees, and the SPEI has become a key metric for understanding where and when that happens. A global analysis of field-observed tree die-off events found a consistent “hotter-drought fingerprint” in the climate conditions during mortality years, with temperatures significantly above average and moisture significantly below. The researchers confirmed that this pattern held across multiple drought metrics, including SPEI at 3-, 6-, 12-, and 24-month timescales.11Nature Communications. Global field observations of tree die-off reveal hotter-drought fingerprint for Earth’s forests
For specific species, researchers have tried to identify SPEI thresholds that predict die-off. In one study of a dryland conifer species in the American Southwest, the SPEI threshold predicted the largest proportion of tree die-off relative to other climate metrics, though classification accuracy hovered around 54 percent, indicating that while SPEI captures the broad signal, pinpointing exact mortality is still challenging.12Environmental Research Letters. Dead again: predictions of repeat tree die-off under hotter droughts confirm mortality thresholds for a dryland conifer species Trees respond to a mix of factors including soil depth, microclimate, and species physiology, so no single index captures everything. Still, SPEI consistently emerges as among the most informative metrics for ecosystem drought stress.
Groundwater and River Flows
Meteorological drought, the kind that SPEI directly measures, eventually propagates underground and into rivers, but not instantly. A study in eastern Romania that correlated SPEI with groundwater levels and river discharge found that the strongest relationships appeared when the hydrological data were lagged by 6 to 12 months behind the SPEI signal. At a 12-month lag, correlations between SPEI and groundwater sometimes exceeded 0.6.13Earth Systems and Environment. Drought Over Time: an Investigation of the Impact of Meteorological Drought on Groundwater and Surface Water in the East of Romania The implication for water managers is that a sustained negative SPEI today is a preview of lower groundwater and reduced river flows months from now, giving time to act if the signal is tracked consistently.
Flash Droughts
Not all droughts develop slowly over months. Flash droughts can set in within days, driven by extreme heat waves, low humidity, and high winds that strip moisture from soil and vegetation before anyone sees it coming. Researchers have adapted the SPEI to very short timescales, computing it over 30-day rolling windows at daily resolution, to identify flash drought onset within about five to seven days.14ScienceDirect. Characteristics and drivers of the flash drought: Based on the daily resolution SPEI from the MERRA-2 between 1980 and 2023 This approach helps disentangle whether a given flash drought was driven primarily by a precipitation deficit or by a spike in evaporative demand, which in turn informs the kind of response that is needed.
Future Projections Under Different Warming Levels
Projections of future drought under climate change are where the SPEI and SPI diverge most dramatically. Under a 3°C warming scenario, SPEI projects increased drought severity across about 92 percent of global land grids, compared to only 28 percent when using SPI. Similarly, drought frequency increases show up over roughly 87 percent of land grids with SPEI but only 33 percent with SPI.15Journal of Hydrology. Global droughts in a warming climate: Evaluation of SPI and SPEI under 1.5°, 2°, and 3 °C global warming The gap is enormous, and it is almost entirely due to the rising evaporative demand that SPI cannot see.
Regional projections underscore the point. In Tehran, for instance, historical moderate-or-worse drought probability (SPEI below −1) sat around 0.15. Under future climate scenarios, that figure jumps to between 0.61 and 0.95. For severe drought (SPEI below −1.5), the historical probability of about 0.05 rises to between 0.39 and 0.90, depending on the emissions pathway.16PubMed Central. Projected changes to drought characteristics in Tehran under CMIP6 SSP-RCP climate change scenarios The researchers attributed this increase almost entirely to rising temperatures rather than declining precipitation, which is exactly the kind of signal the SPI would miss.
Known Limitations
The SPEI is not without blind spots. One of the most actively debated involves the way rising carbon dioxide affects plants. Higher CO₂ concentrations cause many plants to partially close their stomata, the tiny pores on leaves through which water vapor escapes. This means that in a high-CO₂ world, actual water loss from vegetation may be lower than what a standard evapotranspiration formula predicts. A systematic review found that nearly 90 percent of evapotranspiration-based drought projections omit this CO₂-driven vegetation feedback, and neglecting it can overestimate future drought-affected areas by up to about 17 percent under high-emissions scenarios. In some regions, including that feedback actually reverses the projected direction of change from drying to wetting.17IOP Publishing. Neglecting plant physiology: systematic overestimation of drought projections This does not mean future drought risk is overhyped, but it does mean some projections using SPEI may lean pessimistic in regions where vegetation response to CO₂ is strong.
Another limitation arises in cold regions. The standard SPEI does not account for snow. In high-latitude and high-elevation areas, much of the annual precipitation falls as snow and is locked up for months before melting into the water supply. A drought index that treats a snowfall month the same as a rainfall month misunderstands the actual water availability. Researchers have developed modified indices that incorporate snow accumulation and melt, and these outperform both SPEI and SPI in capturing hydrological drought over snow-influenced basins.18PubMed Central. A Framework for Global Multicategory and Multiscalar Drought Characterization Accounting for Snow Processes
Mapping Drought in Data-Scarce Regions
Computing the SPEI requires climate data, and in many parts of the world weather station networks are thin. Central Asia is a prime example. To fill the gap, researchers have built high-resolution gridded SPEI datasets using satellite-derived precipitation and modeled evaporation data, achieving spatial resolutions as fine as 5 kilometers across the region from 1981 to 2018.19PubMed Central. Development and application of high resolution SPEI drought dataset for Central Asia These gridded products allow drought monitoring in places where ground observations alone would leave large blind spots. Similar efforts are under way in sub-Saharan Africa, South and Southeast Asia, and other regions where drought vulnerability is high but monitoring infrastructure is sparse.
The availability of these datasets has opened up new applications. Crop insurance programs, humanitarian early-warning systems, and national adaptation plans in data-limited countries increasingly rely on satellite-informed SPEI grids rather than on raw precipitation records. The trade-off is that satellite-derived data introduces its own uncertainties, particularly in mountainous terrain where microclimate variation is large. But for broad regional monitoring and trend detection, the approach has proved its worth.