1. Quick Summary
Drought is defined by what is missing and for whom. Meteorological drought is a precipitation deficit; agricultural drought is a soil moisture shortfall affecting crops; hydrological drought is reduced streamflow, reservoir and groundwater levels. They occur in sequence with lags between them.
Because each matters to different users, monitoring uses several indicators rather than one number. Precipitation records, soil moisture, snowpack, streamflow, groundwater levels and vegetation condition each describe a different part of the cascade.
Satellites broadened coverage dramatically, providing soil moisture, vegetation health and snow water equivalent over regions where ground instrumentation is sparse. They measure different quantities than ground stations, which is a feature and a complication.
Forecasting skill is genuinely limited and varies with lead time and season. What forecasting does well is characterise risk, not deliver certainties about a specific month.
2. What It Means
The lag structure is the key concept. A dry month is meteorological drought almost immediately, agricultural drought within weeks if soils run down, and hydrological drought months later as rivers and aquifers respond.
This is why a wet week can end a meteorological drought while reservoirs remain low, and why different agencies can legitimately describe the same situation differently.
3. Why It Happens
Indices exist to make comparisons possible, and each encodes assumptions. Some compare precipitation to a historical norm over a chosen period, which is convenient but depends on the reference period and assumes the distribution is stable — an assumption a changing climate undermines.
Soil moisture is the variable that matters most for agriculture, and it is difficult to measure at scale. Satellite retrievals sense only the top few centimetres, so models are needed to estimate what roots actually see deeper down.
Snowpack acts as natural reservoir in many regions, storing winter precipitation and releasing it in spring. Measuring water stored in snow, not just snow depth, is what makes seasonal forecasts useful in snowmelt-dominated basins.
Groundwater is the slowest and least observed component. Declining wells and satellite gravity measurements both indicate change, but attribution to climate versus pumping requires local knowledge.
4. Real Examples
Standardised precipitation indices: compare accumulated precipitation at several timescales to the historical record, revealing short and long-term deficits separately.
Vegetation indices from satellite imagery: show how plants are actually responding, which integrates soil moisture, heat and timing.
Streamflow gauges: the ground truth for water supply, though gauge networks have been shrinking in many regions.
Seasonal ensemble forecasts: run many simulations and report probabilities, which is the honest form of a long-range drought outlook.
5. How It Affects Us
Water managers use different indicators than farmers. Reservoir operators care about inflow weeks to months ahead; irrigators care about soil moisture now and in the next few weeks.
Early warning is most valuable when it triggers prepared action. The most useful monitoring systems are tied to defined responses rather than published as standalone numbers.
Reference-period choice matters politically. As the climate shifts, an index calibrated to the last century will increasingly flag conditions as abnormal even when they are becoming typical.
For individuals, local water restrictions are the practical signal, and they are set by supply system operators based on exactly this kind of monitoring.
6. Key Takeaways
- Drought has at least three definitions — meteorological, agricultural and hydrological — unfolding in sequence.
- No single index describes it; each embeds assumptions about timescale and reference period.
- Satellites fill coverage gaps but measure surface quantities that must be modelled into root-zone or streamflow terms.
- Long-range forecasts are probabilistic and should be read as risk, not prediction.
- Monitoring is most useful when linked to pre-agreed response actions.