Research ArticleCLIMATOLOGY

Multidimensional risk in a nonstationary climate: Joint probability of increasingly severe warm and dry conditions

See allHide authors and affiliations

Science Advances  28 Nov 2018:
Vol. 4, no. 11, eaau3487
DOI: 10.1126/sciadv.aau3487
  • Fig. 1 Historical changes in warm year probability, dry year probability and joint probability.

    (A to C) Change in warm year probability, dry year probability, and joint probability, calculated as the trend over the period (1931–2015) using Sen’s slope estimator times the length of the period. Maps show results for the 50th percentile of the Bayesian sampling. Dark gray indicates no change in the location parameter through time (see Materials and Methods). (D to F) Global average of the warm year probability, dry year probability, and joint probability aggregated across the grid points in the National Oceanic and Atmospheric Administration (NOAA) observations. The bold line shows the posterior mean, and the color envelope shows the 2.5th to 97.5th percentile range (see Materials and Methods). (G to I) Global average of the warm year probability, dry year probability, and joint probability aggregated across the grid points in the Historical and Natural forcing experiments of the CMIP5 global climate model ensemble. The bold line shows the ensemble-mean posterior mean, and the color envelope shows the ensemble-mean 2.5th to 97.5th percentile range (see Materials and Methods).

  • Fig. 2 Historical changes in joint probability of years that are both warm and dry occurring simultaneously in different regions of the world.

    (A and B) Maps showing global cropland and pasture areas in the year 2000 [redrawn from (32)], along with the regions used in our analysis. (C to E) Thickness of lines shows the probability (based on the Bayesian posterior mean) that both regions of a respective region-region pair experience simultaneous warm and dry conditions in the same year during the 1931–1950, 1961–1980, and 1996–2015 periods, based on NOAA observations. Each region pair shares a single joint probability. The color of each region-region joint probability is shown as the color of the first region on the circular plot, starting with Canada and moving clockwise around the circular plot. Thus, all region-region joint probabilities involving Canada are shown in red, and all involving US_West are shown in dark gray except for Canada-US_West, etc. The values of the region-region joint probabilities are shown in (F) to (H). (F to H) Colors show the probability values depicted by lines in (C) to (E). Symbols show the P value of the difference in joint probability between the CMIP5 Historical and Natural simulations for the 1931–1950, 1961–1980, and 1986–2005 periods (see Materials and Methods). The absence of a symbol indicates P value less than 0.01, a gray circle indicates P value between 0.01 and 0.05, and a black circle indicates P value greater than 0.05.

  • Fig. 3 Historical change in joint probability of simultaneous warm+dry conditions of varying severity.

    Colors show the probability (based on the Bayesian posterior mean) that both regions of a respective region-region pair experience simultaneous warm and dry conditions in the same year during the 1931–1950, 1961–1980, and 1996–2015 periods, based on NOAA observations. (A) Joint probability for years in which the temperature anomaly is at least 2 SDs warmer than the baseline mean and the precipitation anomaly is drier than the baseline mean. (B) Joint probability for years in which the temperature anomaly is at least 2 SDs warmer than the baseline mean and the precipitation anomaly is at least 1 SD drier than the baseline mean. (C) Joint probability for years in which the temperature anomaly is at least 4 SDs warmer than the baseline mean and the precipitation anomaly is drier than the baseline mean. The P value of the difference in joint probability between the CMIP5 Historical and Natural simulations is indicated as in Fig. 2.

  • Fig. 4 Projected future change in joint probability of simultaneous warm+dry conditions of varying severity.

    Colors show the probability (based on the Bayesian posterior mean) that both regions of a respective region pair experience severe conditions in the same year during the 2020–2050 period of the RCP8.5 simulations, expressed as the absolute difference from the probability in the CMIP5 historical simulations. (A) Joint probability for years in which the temperature anomaly is at least 2 SDs (left column) or 4 SDs (right column) warmer than the baseline mean. (B) Joint probability for years in which the precipitation anomaly is drier than the baseline mean (left column) or at least 1 SD drier than the baseline mean (right column). (C) Joint probability for years in which the temperature anomaly is at least 2 SDs warmer than the baseline mean and the precipitation anomaly is drier than the baseline mean (left column), the temperature anomaly is at least 4 SDs warmer than the baseline mean and the precipitation anomaly is drier than the baseline mean (center column), or the temperature anomaly is at least 4 SDs warmer than the baseline mean and the precipitation anomaly is at least 1 SD drier than the baseline mean (right column).

  • Fig. 5 Sensitivity of projected future change in joint probability of simultaneous warm+dry conditions to the level of climate forcing.

    Colors show the probability (based on the Bayesian posterior mean) that both regions of a respective region pair experience severe conditions in the same year during the 2020–2050 period of the RCP8.5 simulations, expressed as the percent difference from the probability in the 2020–2050 period of the RCP2.6 simulations. (A) Joint probability for years in which the temperature anomaly is at least 4 SDs warmer than the baseline mean. (B) Joint probability for years in which the temperature anomaly is at least 4 SDs warmer than the baseline mean and the precipitation anomaly is drier than the baseline mean. (C) Joint probability for years in which the temperature anomaly is at least 4 SDs warmer than the baseline mean and the precipitation anomaly is at least 1 SD drier than the baseline mean.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/11/eaau3487/DC1

    Section S1. Datasets

    Section S2. Methodology

    Fig. S1. Schematic flowchart of the methodology to calculate temporal probability of warm, dry, and warm+dry years.

    Fig. S2. A C-vine copula with four dimensions, three trees, and six edges.

    Fig. S3. P values for the time trend of the residual time series.

    Fig. S4. Comparison of joint probability in the NOAA observations and CMIP5 Historical simulations.

    Table S1. List of climate model realizations for temperature variable used to calculate warm year probability for the CMIP5 Historical and Natural forcing experiments and also for future projections based on RCP2.6 and RCP8.5.

    Table S2. List of climate model realizations for precipitation variable used to calculate dry year probability for the CMIP5 Historical and Natural forcing experiments and also for future projections based on RCP2.6 and RCP8.5.

    Table S3. List of climate model realizations available and overlapped for temperature and precipitation variables used to calculate joint warm and dry year probability for the CMIP5 Historical and Natural forcing experiments and also for future projections based on RCP2.6 and RCP8.5.

    Table S4. Elliptical and Archimedean copula functions used in the present study.

    References (3340)

  • Supplementary Materials

    This PDF file includes:

    • Section S1. Datasets
    • Section S2. Methodology
    • Fig. S1. Schematic flowchart of the methodology to calculate temporal probability of warm, dry, and warm+dry years.
    • Fig. S2. A C-vine copula with four dimensions, three trees, and six edges.
    • Fig. S3. P values for the time trend of the residual time series.
    • Fig. S4. Comparison of joint probability in the NOAA observations and CMIP5 Historical simulations.
    • Table S1. List of climate model realizations for temperature variable used to calculate warm year probability for the CMIP5 Historical and Natural forcing experiments and also for future projections based on RCP2.6 and RCP8.5.
    • Table S2. List of climate model realizations for precipitation variable used to calculate dry year probability for the CMIP5 Historical and Natural forcing experiments and also for future projections based on RCP2.6 and RCP8.5.
    • Table S3. List of climate model realizations available and overlapped for temperature and precipitation variables used to calculate joint warm and dry year probability for the CMIP5 Historical and Natural forcing experiments and also for future projections based on RCP2.6 and RCP8.5.
    • Table S4. Elliptical and Archimedean copula functions used in the present study.
    • References (3340)

    Download PDF

    Files in this Data Supplement:

Navigate This Article