Last update to this documentation: August 30, 2024
The Integrated Water Portal includes some legacy products from the original Integrated Water Portal hosted by North Carolina State University, but with an ever-improving and expanding product suite. The key input data for the products is gridded Stage IV Daily Accumulations, downloaded from https://water.noaa.gov in NetCDF format (see https://water.noaa.gov/about/precipitation-data-access). The percent of normal and fraction of normal are relative to the gridded Oregon State University PRISM normals for 1981-2010.
The SPI calculations are performed using historical accumulated precipitation probability distributions using software originally developed at NCSU (Cumbie-Ward and Boyles 2016) that implemented a technique developed by McRoberts and Nielsen-Gammon (2012). Briefly, data from COOP stations are grouped into quasi-homogeneous regions and the data pooled to calculate the moments of a Pearson type-III distribution for different durations and times of year. The higher-order moments are normalized by the location parameter and interpolated to the Stage IV grid, where they are multiplied by the OSU PRISM normals to recover distributions consistent with the spatial variations of precipitation represented by the PRISM normals. The cumulative probability of the actual precipitation for a given grid point, ending time, and duration is mapped using the Pearson type-III distribution onto a standard normal distribution, and the corresponding normal Z-score is the SPI value.
The SPI Blends use the concept of weighting kernels to cause recent precipitation to have a greater influence on drought index values than precipitation in the more distant past (Beguería et al. 2014). This minimizes the problem often encountered with the SPI in which the index value changes suddenly because precipitation that occurred near the beginning of the measurement interval is no longer included in the accumulated precipitation calculation. The specific weight used here is a partial ramp weight, whereby for a nominal duration of n days or n months, precipitation during the first n/2 is given a weight of 1, and the weight for precipitation between n/2 and 3n/2 decreases linearly from 1 to 0. Note that an alternate interpretation of this weighting procedure is to average together all the accumulated precipitation values between 0-n/2 and 0-3n/2, effectively blending the SPI indices for that range of durations. The Pearson type-III calculations and other steps proceed as with the SPI methods described above.
The color tables for low values of the SPI and SPI blends are designed to coincide with the US Drought Monitor percentile values for D0 through D4, except with higher numerical precision than is listed on the USDM web site. High values of SPI and SPI blends are color-coded analogously to represent unusually wet conditions.
Products are generated daily around 9 AM CT and are updated around 1 PM CT to capture any analyses missing from the earlier cutoff.
For technical questions, please contact the Southern Regional Climate Center Director, John Nielsen-Gammon, at n-g@tamu.edu
References
Beguería, S., S. M. Vicente-Serrano, F. Reig, and B. Latorre, 2014: Standardized precipitation evapotranspiration index (SPEI) revisited: parameter fitting, evapotranspiration models, tools, datasets and drought monitoring. Int. J. Climatol., 34, 3001-3023, https://doi.org/10.1002/joc.3887.
Cumbie-Ward, R. V., and R. P. Boyles, 2016: Evaluation of a high-resolution SPI for monitoring local drought severity. J. Appl. Meteor. Climatol., 55, 2247-2262, https://doi.org/10.1175/JAMC-D-16-0106.1
McRoberts, D. B., and J. W. Nielsen-Gammon, 2012: The use of a high-resolution standardized precipitation index for drought monitoring and assessment. J. Appl. Meteor. Climatol., 51, 68-83, https://doi.org/10.1175/JAMC-D-10-05015.1.