Introducing the Moisture Balance Drought Index

This article was compiled by MBDI researcher Melanie Lenart and principal investigator Andrew Ellis.

It’s less flashy than a flood and more subtle than an earthquake. Yet drought actually takes a bigger economic toll in the United States than other natural disasters. Drought losses average billions of dollars a year. In the Southwest, drought is anathema to water and fire managers, ranchers, farmers, and many others who fear lower reservoir levels and parched landscapes.

Monitoring drought, however, is not easy. Drought is not simply the absence of precipitation; if January and July receive about the same precipitation, for example, the landscape is usually drier in the summer, when temperatures are warmer and the sun shines for more hours, sucking more moisture from the landscape. Drought also varies over large regions, with active weather monitoring networks too sparsely located to provide detailed information to meet the needs of many people.

There are several products available to help managers and citizens understand drought conditions, including the Palmer Drought Severity Index (PDSI), the Standardized Precipitation Index (SPI), and the Arizona Drought Monitor. Recently, researchers at Arizona State University and the University of Arizona have developed another tool: the Moisture Balance Drought Index (MBDI), designed to suit the arid climate of the Southwest.   Among other things, the MBDI offers this advantage: Internet users can readily access drought index results at a variety of spatial scales using the tools available on the MBDI website.

In addition, the MBDI presents information at a finer scale than the PDSI, and it allows users to specify the area and time period over which to assess drought conditions. Unlike the SPI, it considers the influence of evaporative demand as well as precipitation.

The Nuts and Bolts of MBDI

The essence of the MBDI is that it assesses the difference between precipitation and potential evapotranspiration (PE), the amount of water that has the potential to move from the Earth’s surface into the air. In economic terms, this would be an analysis of the supply and demand for a good. While drought is most simplistically defined as a decline in precipitation for an area compared to its long-term average, looking solely at precipitation is not the whole story. Evaporation also has an effect.

In 2001 in Payson, Arizona, for example, 2.77 inches of precipitation fell in January, while 2.60 inches of rain fell in July. However, average temperatures were starkly different, registering 37.8 degrees F in January and 74.8 degrees F in July. Daylight also extends about four hours longer in July. These two months thus register vast differences in the amount of moisture that can potentially evaporate from the landscape and plants.

Using the MBDI approach,  the hotter and longer days caused July to have a moisture deficit of about 4.06 inches, which means there was not enough precipitation to meet the moisture demand. January, on the other hand, had a surplus of moisture to the tune of 1.97 inches. That’s the equivalent of about 6 inches difference on the moisture balance—all from the same amount of precipitation.

Supply and demand of moisture

The ultimate impact of drought depends not only on precipitation but also on evaporation rates, and thus temperature. Both precipitation rates and temperature, in turn, vary from the average across a landscape based on various factors, most notably elevation. So the first step in assessing drought level involves interpolating these climate factors across the landscape. Fortunately, the PRISM (Parameter-elevation Regressions on Independent Slopes Model) data set does this.

The PRISM data set contains estimates for monthly precipitation and temperature for every location in the United States, in some cases at a resolution as small as 800 acres. The MBDI bases its assessments on PRISM data for roughly 4,000-acre squares, with each side of the square measuring 4 kilometers, or about 2½  miles.

The formula to calculate the MBDI starts with precipitation, as logic would dictate. Capturing the influence of evaporation is slightly more challenging. Water evaporates not only from the landscape following the laws of physics but also from plants, in a biological process known as transpiration. Together, these evaporative processes are known as evapotranspiration.

The MBDI uses the Hamon method to estimate potential evapotranspiration. The method was developed based on studies of how much water evaporates from well-watered turf using established relationships among temperature and day length (Figure 2).

Even though natural landscapes in the Southwest rarely have this much available moisture, taking PE into consideration accounts for the additional stress on plants as high temperatures boost the evaporative pull (Figure 3). By accounting for PE, the MBDI also recognizes that water supplies, such as canals and reservoirs, that are exposed to air face much higher evaporation rates in summer than winter.

Dimensions of drought impacts

The complexity of drought creates the need for multiple ways to interpret its occurrence or future likelihood for management purposes. The MBDI considers the effects of drought at a variety of scales, from one month to four years.

Reservoir levels are more likely to be affected by a longer-term drought, often on the scale of years. Meanwhile, the overall greenness or health of vegetation as measured by satellite imagery tends to reflect shorter-term climatic conditions of about six months.

A recent analysis led by Andrew Ellis of Arizona State University found groundwater levels in 16 studied wells in Arizona correlated best with the drought indices at the scale of three to four years (36 to 48 months)

Reservoir levels were best predicted with medium-range values of the MBDI. These levels were considered in a case study comparing two adjacent Arizona sub-basins that together provide nearly half of the water supply for metropolitan Phoenix: the Salt and Verde reservoirs. For most months, the Verde storage was best predicted using a time scale of 12 months, while the Salt was best predicted using a time scale of 24 to 36 months, according to the study. These differences likely relate to differences in size of the two reservoirs, as the Salt system holds about seven times more water than the Verde system.

The best time frames for considering streamflow, the amount of water flowing in rivers, were shorter than for groundwater and reservoirs. Differences related in part to the geographic locations of the studied basins—eight Colorado River Basin watersheds—from 1948 to 2007.

In the southerly basins—the Upper Gila, Little Colorado, Salt, and Verde—time frames from one to 12 months work best at explaining variability year-round. In the northerly basins—the Animas, Tomichi, Yampa and Virgin—the optimum time frames for predicting river flow clustered around six to 12 months for spring and summer but 12 to 48 months for winter, the study concluded.

Another study, led by University of Arizona researchers, is testing how well MBDI values compare to greenness of the landscape. The preliminary results indicate the MBDI does best at predicting greenness using a six-month or 12-month scale, although the three-month scale also worked reasonably well.

The comparison of MBDI values to greenness is based on comparisons of 17 sites in Arizona between 1989 and 2007, with greenness values calculated from satellite imagery using the Normalized Difference Vegetation Index (NDVI). Spring greenness is the most variable in the Southwest, especially in lowland desert.

Scales of drought

Because drought operates at varying spatial and time scales, the MBDI takes different time frames into consideration using cumulative comparisons. For example, the one-month time frame would compare August conditions to conditions during every previous August of record.

Each month in the record is then given a rank that indicates where it falls in the historical line-up during the MBDI period of record, from 1895 to the present.  The driest years will fall into the lowest rank, such as the 25th percentile, while the wettest years will rank within the top 75th percentile.

These various monthly values can be tallied together into various time frames and ranked as a unit in comparison to similar time frames—for example, for all six-month periods ending in June.

This approach acknowledges that it’s possible for an area to be extremely wet for that month compared to the average amid a two-year drought, for instance. The opposite also can be true.

Using the MBDI

On the MBDI website, users can highlight the grid cells or watersheds to consider index values in their area of interest. Web users have the option of considering which time scales best characterize the drought impacts of their particular interest, such as fire occurrence, fluctuations in wildlife populations, and reservoir levels.

Because of space limitations, the website provides only the past 15 years of data, but researchers interested in considering longer time frames are encouraged to contact the MBDI developers.

The researchers who developed the index and related website, a team led by Andrew Ellis of Arizona State University and Gregg Garfin of the University of Arizona, are interested in hearing about and helping with independent efforts involving the MBDI. The hope is that the research community can use this tool to further refine the growing understanding of how drought affects land and society at a variety of scales.

More details can be found on the MBDI website at http://azclimate.asu.edu/mbdi. Researchers and managers interested in using the MBDI for comparisons of their own interest can contact Arizona State University climatologist Andrew Ellis (Andrew.W.Ellis@asu.edu) for more details and longer datasets.