Research ArticleATMOSPHERIC SCIENCE

Future precipitation increase from very high resolution ensemble downscaling of extreme atmospheric river storms in California

See allHide authors and affiliations

Science Advances  15 Jul 2020:
Vol. 6, no. 29, eaba1323
DOI: 10.1126/sciadv.aba1323
  • Fig. 1 Spatial distributions of moisture fluxes from the 60 historical (left) and future (right) AR events (WRF 81 km) under the RCP8.5 forcing scenario.

    (A) Composite hourly instantaneous IVT map showing the moisture flux transport pattern averaged over each of the 60 ARs for each period at the time of maximum hourly precipitation over California. (B) Geographical distributions of the AR events, with each open circle denoting the locations at 12-hourly time intervals. The location is defined as the grid box with maximum IVT value; color shading denotes IVT magnitude.

  • Fig. 2 Mean latitude-height cross section of zonal (left) and meridional (right) wind (near the coast at 130°W) for maximum AR precipitation days over California in CESM-LENS.

    (A) Wind profiles for historical AR events. (B) Changes in wind profiles during future AR cases. The two white vertical dashed lines in each panel denote the lower and upper bounds of the latitude range of California.

  • Fig. 3 Precipitation and thermodynamic changes in simulated ARs, present versus future (WRF 3 km).

    The first two rows show precipitation results zoomed in on the Sierra Nevada region (35.0°N to 41.0°N and −122.5°W to −117.5°W): (A, B, E, and F) for historical and future averaged event total and event maximum hourly precipitation rate from all 60 AR events in each period; (C, D, G, and H) for absolute and relative future changes in event total precipitation and event-maximum hourly precipitation intensity. Stippling in (D) and (H) denotes regions where changes are statistically significant at the P < 0.1 level. Sierra Nevada watershed boundaries are overlaid in all panels, denoted by black outlines. The bottom row illustrates thermodynamic scaling of water vapor in extreme ARs: (I) Near-surface (2 m) warming over California and surrounding region at 9-km resolution; (J and K) relative (%) change in near-surface specific humidity per degree of warming and relative (%) change in IVT per degree of warming. Scaling is calculated using event-averaged quantities.

  • Fig. 4 Linking large-scale forcing (WRF 81 km) and local vertical motion to dynamically downscaled fine-scale precipitation in 3-km WRF.

    (A) Predictors used are as follows: On top is a scatter plot of the daily mean IVT_u (zonal moisture flux, kg/m per second) and IVT_v (meridional moisture flux, kg/m per second) averaged from the near-coastal grid boxes (see the rightmost column for the grid box locations), and below is the local vertical motion Wij (m/s) at 3 km from WRF, averaged over all events. Middle column (B): Variance in precipitation explained by single-predictor regression models (see the main text) and by the four-parameter MLR model. (Here, the location factor refers to the grid box location with maximum IVT for each day from the seven coastal grid boxes). (C): Predictand: i.e., daily mean precipitation at fine-scale (3 km) in both periods. Sierra Nevada watersheds are overlaid, as in Fig. 3.

  • Fig. 5 Statistically predicted precipitation using MLR method and comparison to WRF 3-km precipitation output.

    (A) Predicted precipitation using four-parameter MLR for historical and (B) future conditions. (C and D) Predicted precipitation changes from MLR and the relative difference (%) compared to the WRF simulated changes. (E and F) The respective contribution from changes in specific humidity and changes in zonal and meridional winds to future precipitation extremes. (G) Ratio of the contribution to precipitation changes from thermodynamic factors [as in (E)] to that from dynamic factors (U/V winds) [as in (F)]. (H) The precipitation changes added by including Wij as an additional predictor in the MLR model.

  • Fig. 6 IVT distributions from 40 CESM-LENS ensemble members for near-coastal grid boxes over California from historical (1996–2005) and future (2071–2080) periods under the RCP8.5 emissions scenario.

    Left and middle: Red dots represent the 5-day running mean IVT intensity starting for the entire 10-year historical and future periods. Note that this essentially includes all ARs not just the most extreme events. On the plot, IVT is truncated at the lower bound at 250 km/m per second. Each red horizontal band is a collection of points representing IVT values from each coastal grid box along the California coast. Within each band, values from each of the 40 individual CESM-LENS members are stacked one on top of the other. The corresponding IVT values for each of the 60 extreme AR events selected for downscaling during each period are denoted by black circles. Right: Centroid locations of the corresponding near-coastal grid boxes, with 3-km topography represented by the color contours over land. (Sierra Nevada watershed boundaries are also overlaid with black lines.)

Supplementary Materials

  • Supplementary Materials

    Future precipitation increase from very high resolution ensemble downscaling of extreme atmospheric river storms in California

    Xingying Huang, Daniel L. Swain, Alex D. Hall

    Download Supplement

    This PDF file includes:

    • Supplementary Text
    • Figs. S1 to S7

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article