Standardized Precipitation Index (SPI)

Standardized Precipitation Index (SPI)

 

Index name: Standardized Precipitation Index (SPI).

Ease of use: Green.

Origins: The result of research and work done in 1992 at Colorado State University, United States, by McKee et al. The outcome of their work was first presented at the 8th Conference on Applied Climatology, held in January 1993. The basis of the index is that it builds upon the relationships of drought to frequency, duration and timescales.

In 2009, WMO recommended SPI as the main meteorological drought index that countries should use to monitor and follow drought conditions (Hayes, 2011). By identifying SPI as an index for broad use, WMO provided direction for countries trying to establish a level of drought early warning.

Characteristics: Uses historical precipitation records for any location to develop a probability of precipitation that can be computed at any number of timescales, from 1 month to 48 months or longer. As with other climatic indicators, the time series of data used to calculate SPI does not need to be of a specific length. Guttman (1998, 1999) noted that if additional data are present in a long time series, the results of the probability distribution will be more robust because more samples of extreme wet and extreme dry events are included. SPI can be calculated on as little as 20 years’ worth of data, but ideally the time series should have a minimum of 30 years of data, even when missing data are accounted for.

SPI has an intensity scale in which both positive and negative values are calculated, which correlate directly to wet and dry events. For drought, there is great interest in the ‘tails’ of the precipitation distribution, and especially in the extreme dry events, which are the events considered to be rare based upon the climate of the region being investigated.

Drought events are indicated when the results of SPI, for whichever timescale is being investigated, become continuously negative and reach a value of -1. The drought event is considered to be ongoing until SPI reaches a value of 0. McKee et al. (1993) stated that drought begins at an SPI of -1 or less, but there is no standard in place, as some researchers will choose a threshold that is less than 0, but not quite -1, while others will initially classify drought at values less than -1.

Owing to the utility and flexibility of SPI, it can be calculated with data missing from the period of record for a location. Ideally, the time series should be as complete as possible, but SPI calculations will provide a ‘null’ value if there are insufficient data to calculate a value, and SPI will begin calculating output again as data become available. SPI is typically calculated for timescales of up to 24 months, and the flexibility of the index allows for multiple applications addressing events that affect agriculture, water resources and other sectors.

Input parameters: Precipitation. Most users apply SPI using monthly datasets, but the computer programs have the flexibility to produce results when using daily and weekly values. The methodology of SPI does not change based upon using daily, weekly or monthly data.

Applications: The ability of SPI to be calculated at various timescales allows for multiple applications. Depending on the drought impact in question, SPI values for 3 months or less might be useful for basic drought monitoring, values for 6 months or less for monitoring agricultural impacts and values for 12 months or longer for hydrological impacts. SPI can also be calculated on gridded precipitation datasets, which allows for a wider scope of users than those just working with station-based data.

Strengths: Using precipitation data only is the greatest strength of SPI, as it makes it very easy to use and calculate. SPI is applicable in all climate regimes, and SPI values for very different climates can be compared. The ability of SPI to be computed for short periods of record that contain missing data is also valuable for those regions that may be data-poor or lacking long-term, cohesive datasets. The program used to calculate SPI is easy to use and readily available. NDMC provides a program for use on personal computers that has been distributed to more than 200 countries around the world. The ability to be calculated over multiple timescales also allows SPI to have a wide breadth of application. Many articles relating to SPI are available in the science literature, giving novice users a multitude of resources to rely on for assistance.

Weaknesses: With precipitation as the only input, SPI is deficient when accounting for the temperature component, which is important to the overall water balance and water use of a region. This drawback can make it more difficult to compare events of similar SPI values but different temperature scenarios. The flexibility of SPI to be calculated for short periods of record, or on data that contain many missing values, can also lead to misuse of the output, as the program will provide output for whatever input is provided. SPI assumes a prior distribution, which may not be appropriate in all environments, particularly when examining short-duration events or entry into, or exit out of, drought. There are many versions of SPI available, implemented within various computing software packages other than that found in the source code distributed by NDMC. It is important to check the integrity of these algorithms and the consistency of output with the published versions.

Resource: The SPI program can be run on Windows-based personal computers. Download at: National Drought Mitigation Center – SPI Program. Global SPI data at 3-, 6-and 12-moth scales is available at NCAR/UCAR Research Data Archive. Visualization of global SPI data at 3-, 6- and 12-month scales is provided by the Tokyo Climate Center (TCC) ClimatView tool. Additional resources are available at the Flood and Drought portal developed by GEF, UN Environment, IWA and DHI.

References: 
Guttman, N.B., 1998: Comparing the Palmer Drought Index and the Standardized Precipitation Index. Journal of the American Water Resources Association, 34: 113–121. DOI: 10.1111/j.1752-1688.1998.tb05964.x. (For more information on this paper, please contact the IDMP HelpDesk).

Guttman, N.B., 1999: Accepting the Standardized Precipitation Index: A Calculation Algorithm. Journal of the American Water Resources Association, 35: 311–322. DOI: 10.1111/j.1752-1688.1999.tb03592.x. (For more information on this paper, please contact the IDMP HelpDesk).

Hayes, M., M. Svoboda, N. Wall and M. Widhalm, 2011: The Lincoln Declaration on Drought Indices: Universal Meteorological Drought Index Recommended. Bulletin of the American Meteorological Society, 92(4): 485–488. DOI: 10.1175/2010BAMS3103.1.

McKee, T.B., N.J. Doesken and J. Kleist, 1993: The Relationship of Drought Frequency and Duration to Time Scales. Proceedings of the 8th Conference on Applied Climatology, 17–22 January 1993, Anaheim, CA. Boston, MA, American Meteorological Society.

World Meteorological Organization, 2012: Standardized Precipitation Index User Guide, (WMO-No. 1090), World Meteorological Organization, Geneva, Switzerland.

Wu, H., M.J. Hayes, D.A. Wilhite and M.D. Svoboda, 2005: The Effect of the Length of Record on the Standardized Precipitation Index Calculation. International Journal of Climatology, 25(4): 505–520. DOI: 10.1002/joc.1142.

Currently used by: Argentina, Austria, Belize, Bosnia and Herzegovina, Brazil, Bulgaria, Canada, Chile, Croatia, Cyprus, Dominican Republic, Germany, Greece, Hong Kong, Iran, Israel, Jamaica, Jordan, Kazakhstan, Libya, Lithuania, Macedonia, New Zealand, Pakistan, Peru, Slovenia, Spain, Sri Lanka, Switzerland, Tanzania, Thailand, Trinidad and Tobago, Turkey, Ukraine, USA.

70 Responses

  1. Hello everyone!!

    I have used monthly precipitation using the monthly rainfall data, the SPI index is generated used the tool on WMO website. Pls guide me as to how to convert the monthly SPI index values to an Yearly index. Please guide

    1. Dear Vidya,
      Thank you for your question. We recommend to calculate the annual SPI based on annual precipitation values.

      Kind regards,
      Valentin

  2. Hello! I have input my data and setup for generating the output. When I generated it is done completely but no output. Thus result at all.

    1. Dear Jabiru,

      thank you for your question. Can you please clarify what output is missing?

      Best regards,
      IDMP TSU

  3. This software uses a parametric or non parametric method? Which distribution is used in this software to calculate SPI?

  4. I have used the program to generate 1-month SPI values. I found that the month of July (highest monthly rainfall in study area) across multiple years has a SPI of exactly zero which seems like a strange result. Are there any specific reasons for this?

  5. Dear Belkys

    Thank you for your comment. With SPI time scale I understand that you mean the run lengths of 1,3,6,12,… months. The choice of run length depends on what time scale you want to look at – short, medium or long-term moisture conditions. Further info on this is given here: Standardized Precipitation Index User Guide. Note that a data series of monthly or weekly precipitation of at least 30 years is required in order to calculate robust probabilities.

    Best regards
    IDMP TSU

  6. Dear Yeluma
    The SPI is one of several indices used in the US Drought Monitor. The National Drought Mitigation Center has more information on the USDM: https://drought.unl.edu/Home.aspx
    Additionally, please find information and references in the following websites:
    https://www.ncdc.noaa.gov/temp-and-precip/drought/nadm/indices
    https://www.drought.gov/data-maps-tools/us-gridded-standardized-precipitation-index-spi-nclimgrid-monthly
    Publications (open source):
    https://journals.ametsoc.org/view/journals/bams/80/3/1520-0477_1999_080_0429_mtduts_2_0_co_2.xml?tab_body=pdf

    Best regards
    IDMP TSU

  7. Dear Yeluma
    The SPI is one of several indices used in the US Drought Monitor. The National Drought Mitigation Center has more information on the USDM: https://drought.unl.edu/Home.aspx
    Additionally, please find information and references in the following websites:
    https://www.ncdc.noaa.gov/temp-and-precip/drought/nadm/indices
    https://www.drought.gov/data-maps-tools/us-gridded-standardized-precipitation-index-spi-nclimgrid-monthly
    Publications (open source):
    https://journals.ametsoc.org/view/journals/bams/80/3/1520-0477_1999_080_0429_mtduts_2_0_co_2.xml?tab_body=pdf

    Best regards
    IDMP TSU

  8. Dear Shani Khan

    The SPI program uses two input parameters – the date and precipitation amount. Please consult the documentation of the SPI tool for detailed information on how to format input files. The documentation is part of the software package that can be downloaded here: https://drought.unl.edu/droughtmonitoring/SPI/SPIProgram.aspx.
    Regarding your analysis of various stations, you can generate the SPI for each station by preparing an input file for each station.

    Best regards
    IDMP TSU

      1. Dear Angad
        The website is temporarily offline, but will go online soon again. The link is correct, please try again to see the status.
        Best regards
        IDMP TSU

  9. tus datos históricos tienes q ordenarlos en dos columnas (Fecha; Precipitacion (mm)) para que puedas trabajarlos. saludos

  10. Dear Shafique,

    Thank you for your question. The severity of the SPI is more negative number. Please note the SPI is standardized. Also, you will need to compare the SPI calculations with the historical occurrences and impacts.

    Best regards,

    IDMP TSU

  11. Dear Shafique,

    Thank you for your question. Please let us know if you are still having a problem using the SPI downloadable file with Windows 10. We will check with the US National Drought Mitigation Center who are the developers of the software to see there are any recurring issues. Here is the link to their website.

    Best regards,
    IDMP TSU

  12. Dear Sachini,

    Thank you for your question. This is a detailed question. We will email you directly if there is still a problem.

    Best regards,
    IDMP TSU

  13. Hi
    i want to calculate the SPI for 6 month time scale. I have 10 different rainfall stations of one region. can anyone tell me how to arrange the data for calculation of SPI. I have the data for same time e.g 1981 to 2020 for all stations.

  14. Dear all anyone tell me how to seprate duraton and severity variables from calculated SPI.
    I have successfullu calculate the SPI values from different stations but dont know how to identiify the sevrity and duration vaiables.

  15. If I click on executable file then SPI generator file is seen and not understand about documentation how to carry input file and SPI are obtain

  16. If I click on executable file then SPI generator file is seen and not understand about documentation how to carry input file and SPI are obtain

  17. Sir how to install SPI software in pc. i am using window 10 in pc. If i download the document and programming file then no setup is seen in that folder. kindly help me about that. Thankuu

  18. I’m going to compare drought events between two cities. There I’m using only 12 years of data(2007-2018).
    For one town spi values related to February is NA(for the whole period only February data). I don’t why. But the software returned the February data for the other town. Please help me in this regard.

  19. Dear Abela Bekele

    Thank you for your comment.
    As for a model that will deliver the best SPI predictions 3 months ahead of time, this is dependent on which model works best for your local context. Your national Meteorlogical Service may have information on which model performs best in your region of Ethiopia. Once you have decided on a model, you can use the SPI generator application (https://drought.unl.edu/droughtmonitoring/SPI/SPIProgram.aspx) to run gridded projected precipitation values for your chosen time period. In this case, the program treats gridded data as station data.
    You can use the precipitation data from your local station to calculate the SPI. Please note that for a viable result, you will need good monthly precipitation data for at least the past 25-30 years. A longer data track is always better in order to get more reliable results.

    Best regards
    IDMP team

  20. which models is more appropriate to forecasting drought a head of time of three month through using SPI? specially for the semi-arid areas of Oromia pastoral areas of Ethiopia, How will I get SPI soft were? could I use the rain fall data from local station to get SPI result of my interest area?

  21. Hi
    How to take timescales in computing spi? Taking them in the cumulative probability or in precipitation data? Can you share me a theoretical step by step computation

    1. Dear Shahla,

      Thank you for your email. If we understand your request correctly, it is recommended to use 30 years of precipitation data. You can use 10 years but then the analysis will not be robust.

      Please look at the references at the bottom of the SPI web page. Specifically, look at the Standardized Precipitation Index User Guide published by WMO.

      Best regards,
      IDMP Team

    1. Dear Santoshchougale.27

      Thank you for your comment.
      The Standardized Precipitation Index (SPI) User Guide provides an explanation of the data needed in order to calculate the SPI. You can find it here: https://library.wmo.int/pmb_ged/wmo_1090_en.pdf

      This extract of the above-mentioned publication lists the key points on how the SPI works:
      • Precipitation is normalized using a probability distribution function so that values of SPI are
      actually seen as standard deviations from the median.
      • A normalized distribution allows for estimation of both dry and wet periods.
      • Accumulated values can be used to analyse drought severity (magnitude).
      • At least 30 years of continuous monthly precipitation data are needed but longer-term
      records would be preferable.
      • SPI timescale intervals shorter than 1 month and longer than 24 months may be unreliable.
      • It is spatially invariant in its interpretation.
      • Its probability-based nature (probability of observed precipitation transformed into an index)
      makes it well suited to risk management and triggers for decision-making.

      Kind regards,
      IDMP Team

  22. Hi Jaldesa
    About drought forecasting, First of all, you should obtain precipitation values of the future period via outputs of CMIP5 models, or weather generators, or etc., then you can put them in the MDM (https://agrimetsoft.com/mdm.aspx) tool or other tools for calculating rain-based drought indices.
    When you extract your specific data from CMIP5 models under RCP scenarios, you should use statistical downscaling methods (https://agrimetsoft.com/sd-gcm.aspx) or dynamical downscaling methods for finer resolutions then apply them in your drought monitoring tool for the future period. Also, be careful for implementing calibration and validation steps
    If you have further question let me know (salehnia61@gmail.com).
    Regards,
    Nasrin

  23. Hi Robert Stefanski and Hassan
    I checked the DrainC tool and compare it. Also, I focused on the formula I think the result of Drainc is not true. But the Drought Mitigation Center software for SPI is similar to DMAP and it is true.

  24. Hi Jaldesa Doyo
    If you want to predict drought so you need to future climate data. You can use weather generators such as KNN-WG.
    The best prediction for drought prediction is GCM models. because GCM models have been run by dynamic models based on RCP scenarios. RCP scenarios consider climate change.
    You can use SD-GCM tool for downscaling GCM data. If you want to use coarse scale so you can use DMAP V1.1.
    Good Luck

  25. I am to get this vital information from you, My question is that , how we can calculate future drought / drought forecast for drought monitoring by SPI and RDI?

    1. Dear Eremugo Issac,

      Here is the refernce:
      Nalbantis, I. and G. Tsakiris, 2009: Assessment of hydrological drought revisited. Water Resources Management, 23(5): 881–897. DOI 10.1007/s11269-008-9305-1.

      Best regards,
      IDMP TEAM

    1. Dear Hassan,

      Thank you for your email. It is always go to look at different formulas. Are there significant differences in the three methods? This may be an interesting journal article. However, it not a question of which one of them is true but which one best represents drought in your country and region.

      We do know that the Drought Mitigation Center software for SPI is widely used.

      Best regards,

      IDMP

  26. Hello!
    I need to have SPI software temple to download please if you have send to me. And Mann-Kendal software temple.
    Best regards,

  27. Hello!

    Can you explain me or give me reference about -0.8 SPI value as a threshold of moderate drought? It will serve to research in Peru.

    Thanks a lot.

    1. Dear Kris Correa Marrou,

      The threshold on any drought index including the SPI have to be referenced to historical events and impacts.
      Please look at the following publications:
      SPI guide at https://www.droughtmanagement.info/literature/WMO_standardized_precipitation_index_user_guide_en_2012.pdf;
      the Handbook on Drought Indices at https://www.droughtmanagement.info/find/guidelines-tools/handbook-drought-indicators-and-indices/
      and the National Drought Management Policy Guidelines at https://www.droughtmanagement.info/find/guidelines-tools/guidelines/
      Best regards,

      IDMP Team

  28. Hello!
    Can you explain me or give me a reference about -0.8 SPI value as a threshold of moderate drought? It will serve to research in Peru.

    Thanks a lot.

  29. my question regarding this SPI 1,is what are the softwares we can calculate SPI on 2, for how many year that we have to calculate that the result have to be accurate 3, what are the accurate classes for the drought in pastoralist areas e.g moderate ,severy and others or standard numbers are

    1. Dear Galgalo Wako,

      Here are the answers to your questions.

      1) Please find the SPI Guide at https://www.droughtmanagement.info/literature/WMO_standardized_precipitation_index_user_guide_en_2012.pdf

      2) According the above guide, the minimum dataset is 30 years of rainfall data. Here the exact text from page 7 of the SPI guide:

      “… recommendation of having around 50–60 years of data available. Unless one has 80–100 years of data, the sample size is too small and the statistical confidence of the probability estimates on the tails (both wet and dry extremes) becomes weak beyond 24 months. In addition, having only the minimum 30 years of data (or less) shortens the sample size and weakens the confidence. Technically, one could run the SPI on less than 30 years of data bearing in mind, however, the statistical limitations and weaker confidence pointed out above.”

      3) The SPI can be used for many applications but you will need to know the historical droughts and impacts. We will need to obtain the previous droughts impacting pastoralists then you could determine which classes or thresholds to use for your country or region.

      IDMP Team

  30. softwares about standardized preciptation index and detail statsitcal calculation and working procedure about standardized streamflow indx(ssfi), streamflow drought index(sdi), drought reconenesance index(dri) ,percent normal precipitation, deciles

    1. Dear Thabang,

      Please take a look at Tables 1 and 2 in the SPI Guide. The classification scheme can also be adjusted to your actual needs. This is where you will need to know the historical impacts on drought in your country or location.

      Best regards,
      IDMP TSU

Leave a Reply

Your email address will not be published. Required fields are marked *

Translate »