Drought Termination and Amelioration
Drought in the U.S.
The incidence of drought in the United States has varied greatly over the past century. From the dust bowl years of the 1930s to the major droughts of 1988 and 2000, much of the U.S. has suffered from the effects of drought during the past century. While annual and seasonal precipitation totals have generally increased in the United States since 1900, severe drought episodes continue to occur.
The nation's most devastating drought occurred in the 1930s during what many refer to as the 'Dust Bowl' years. The drought affected almost the entire Plains and covered more than 60% of the US during its peak in July 1934. It brought devastating economic impacts to many and caused the migration of millions of people from the Plains to other parts of the country, many to the Western US. Major drought episodes have recurred since then (e.g., those of the 1950s, 1988, and 1998-2014), with the 2012 drought comparable to the drought of the 1930s in terms of size and severity, with serious economic and societal impacts.
Although a variety of weather related phenomena have the potential to cause great economic and personal losses in the US, drought has historically had the greatest impact on the largest number of people. Of all the weather and climate-related disasters that have caused at least $1 billion in economic losses since 1980, drought represents the second most costly event type and the second highest average event cost.
Persistent above-normal temperatures and below-normal precipitation brought drought conditions to parts of the US from 1998 into the second decade of the 21st century, with drought affecting over half of the country in 2000, 2002, 2006, and 2012-13. Heavy agricultural losses, water rationing, and severe wildfire seasons occurred during these times. Drought coverage peaked at nearly two-thirds of the US during July 2012, ranking this drought as the second most expansive drought after July 1934.
The wide variety of disciplines affected by drought, its diverse geographical and temporal distribution, and the many scales drought operates on make it difficult to develop both a definition to describe drought and an index to measure it. Common to all types of drought is the fact that they originate from a deficiency of precipitation resulting from an unusual weather pattern. If the weather pattern lasts a short time (e.g., a few weeks or a couple months), the drought is considered short-term. But if the weather or atmospheric circulation pattern becomes entrenched and the precipitation deficits last for several months to several years, the drought is considered to be a long-term drought.
Many quantitative measures of drought have been developed in the United States, depending on the discipline affected, the region being considered, and the particular application. The most frequently used indicators of drought are those developed by Wayne Palmer in the 1960s. These include the Palmer Drought Severity Index (PDSI), the Palmer Hydrological Drought Index (PHDI), the Palmer Z Index and the Crop Moisture Index (CMI). These indices have been used in countless research studies as well as in operational drought monitoring during the past 50+ years. The Palmer drought index has proven to provide one of the best indications of drought for much of the United States. It is superior to other drought indices in many respects because it accounts not only for precipitation totals, but also for temperature, evapotranspiration, soil runoff and soil recharge.
The Z Index measures short-term drought on a monthly scale while the CMI measures short-term agricultural drought on a weekly scale. The PDSI measures drought duration and intensity of long-term drought-inducing circulation patterns and responds fairly quickly as meteorological patterns often quickly change from one regime to another. However, a reflection of the long-term effects of drought on systems affected by long-term precipitation deficits is measured by the PHDI, a measure of the hydrological impacts of drought. These impacts, such as reservoir levels, groundwater levels, etc., take longer to develop and it takes longer to recover from them. It is from this index, the PHDI, that we calculate the precipitation amounts and probabilities of ending or ameliorating drought.
The Evaporative Demand Drought Index (EDDI; Hobbins et al. 2016, McEvoy et al. 2016) looks solely at the "thirst" of the atmosphere using a physically-based evaporative demand driven by near-surface temperature, wind speed, humidity, and solar radiation. Rapid agricultural drought, or "flash" drought, development is often initiated by increased evaporative demand and EDDI can provide early warning before drought impacts emerge on the ground. EDDI has the advantage of being multiscalar and serves as a good flash drought indicator at short time scales (i.e., weeks to one month) and can detect hydrologic drought at longer time scales (i.e., greater than six months).
Current Drought Reduction
Because of the far-reaching societal and economic impacts of drought, there is considerable interest in determining how much precipitation is required to end a drought as well as the probability that a region may receive the necessary amount of precipitation. Ending a hydrological drought requires that the moisture needs associated with recharge, demand and runoff have been brought back to normal or above normal. The Palmer Hydrological Drought Index (PHDI) is one tool that provides this information.
Many factors affect the quantity of precipitation required to end or ameliorate (reduce the severity of) a drought. Knowledge of the severity of the drought, as defined by the PHDI, is the essential starting point for determining the needed precipitation. The typical conditions that a region experiences during each month and season of the year (i.e., that region's climatology) is also essential. Given a drought of equal magnitude in a dry and wet climate, the wetter region requires more precipitation to end the drought.
The season in which the precipitation falls can also greatly influence the quantity of precipitation required to end a drought. During a typically moist month (such as those experienced in the winter and spring along the West Coast) more precipitation may be required to end a drought than during the typically dry months of the summer. Because soil moisture conditions are generally lower in the dry months, the precipitation needed to bring soil conditions back to normal may be less than that required to return soil moisture conditions to normal during a generally wetter season. Nevertheless, regardless of a region's climate, over a sufficiently long period of time, near-normal precipitation is often sufficient for ending a drought with moisture conditions gradually returning to normal.
However, the quantity of precipitation needed to end a drought says nothing about the probability that a region will actually receive that amount of precipitation. A region, such as the West Coast, that does not typically experience excessively heavy precipitation during the summer season, may be less likely to receive a quantity sufficient for ending a drought than a region which has a record of experiencing extreme precipitation events during the same season. The months which have the greatest probability of receiving substantially more precipitation than normal would be those with precipitation distributions with the largest positive skew (that is, those subject to more extreme precipitation events), not necessarily those months that normally receive the greatest amount of precipitation.
The technical details associated with the calculation of precipitation totals needed to end or ameliorate drought and the probability of receiving the required precipitation can be found in "Drought Termination and Amelioration: It's Climatological Probability" by Tom Karl et al. 1987.
The end of a drought is defined by a PHDI value of -0.5 while drought amelioration is achieved when a PHDI value of -2.0 is reached. Current drought reduction scenarios show the precipitation needed over periods from 1 month to 12 months as well as the associated probability of receiving that quantity of precipitation to end or ameliorate current drought within the contiguous U.S. (assuming climatological conditions for the remainder of the month). The data are updated daily with the values reflecting conditions for the current month.
The PHDI for the current month is an estimate. It is based on observed temperature and precipitation through the current day; climate normals are used to estimate the temperature and precipitation for the rest of the month. The resulting monthly temperature and precipitation data are used to compute the PHDI for the current month. This process is explained in "Computing the monthly Palmer Drought Index on a weekly basis: A case study comparing data estimation techniques" by Richard Heim, 2005.
Drought termination and amelioration information is also provided for a "worst case scenario". The PHDI for the current month is estimated from observed temperature and precipitation through the current day and estimates for the rest of the month; climate normals are used to estimate the temperature for the rest of the month, but a "worst case scenario" of no precipitation is assumed for the rest of the month. These resulting monthly temperature and precipitation data are then used to compute the PHDI for the current month. It is expected that the "worst case scenario" PHDI will always be lower (more negative, or worse) than the PHDI computed using climate normals for the rest of the month. Drought termination values for the current PHDI and "worst case scenario" PHDI, used together, provide a range of expected drought termination precipitation and probabilities which can be used for planning purposes.
The NOAA National Weather Service's Climate Prediction Center (CPC) produces official monthly and seasonal temperature and precipitation outlook products. This information was converted into projected monthly temperature and precipitation values for future months, then the projected values were used to compute projected future PHDI values. Specifically, the median forecasts of the CPC 3-month seasonal temperature and precipitation outlooks were converted into seasonal anomalies for the 102 CPC forecast regions using the CPC climatologies. The seasonal anomalies for the CPC regions were converted into seasonal anomalies for the NCEI climate divisions using a spatial analysis technique. These inferred seasonal forecast anomalies for the climate divisions were then combined with the 1981-2010 climate division normals to yield 3-month projections of temperature and precipitation for each climate division. Monthly projections were computed from the seasonal projections. For temperature, the monthly projected temperature anomalies for each of the three months of the season were set equal to the seasonal projected temperature anomaly, then monthly projected temperature values were computed by adding the monthly projected anomalies to the 1981-2010 monthly climatology. For precipitation, the seasonal projected precipitation was divided by three and distributed equally amongst the three months. The resulting monthly temperature and precipitation values for the three future (projected) months were appended to the historical temperature and precipitation data base and this dataset was then processed through the Palmer program to compute projected Palmer drought indices for the three future months. The projections for the third month of the season are provided.
The Climate Forecast System version 2 (CFSv2) is the operational seasonal forecast model used by the CPC and is used to generate EDDI and evaporative demand probability forecasts. Temperature, wind speed, humidity, and solar radiation are used from CFSv2 to compute reference evapotranspiration (one flavor of evaporative demand that is used to compute EDDI). Each month a new reference evapotranspiration ensemble is created from CFSv2 forecasts that are initialized every five days and four times a day. This results in a 24-28 member forecast ensemble each month. The ensemble forecast is then combined with the CFSv2 reforecast (hindcasts covering the period 1982-2009) to create distributions for EDDI calculations and terciles for the probability forecasts. A new forecast is provided on the first of each month for the next six-month period.