Change factors to derive projected future precipitation depth-duration-frequency (DDF) curves at 242 National Oceanic and Atmospheric Administration (NOAA) Atlas 14 stations in Florida (ver 1.1, September 2023)

This data release consists of Microsoft Excel workbooks, shapefiles, and a figure (png format) related to a cooperative project between the U.S. Geological Survey (USGS) and the Florida Flood Hub for Applied Research and Innovation at the University of South Florida to derive projected future change factors for precipitation depth-duration-frequency (DDF) curves at 242 National Oceanic and Atmospheric Administration (NOAA) Atlas 14 stations in Florida. The change factors were computed as the ratio of projected future (2020-59 or 2050-89) to historical (1966-2005) extreme-precipitation depths fitted to extreme-precipitation data using a constrained maximum likelihood (CML) approach as described in https://doi.org/10.3133/sir20225093. The change factors are tabulated by duration (1, 3, and 7 days) and return period (5, 10, 25, 50, 100, 200, and 500 years). The official historical NOAA Atlas 14 DDF curves based on partial-duration series (PDS) can be multiplied by the change factors derived in this project to determine projected future extreme precipitation for events of a given duration and return period.

Various statistical, dynamical and hybrid downscaled precipitation datasets were used to derive the change factors at the grid cells closest to the NOAA Atlas 14 stations including (1) the Coordinated Regional Downscaling Experiment (CORDEX), (2) the Localized Constructed Analogues (LOCA) dataset, (3) the Multivariate Adaptive Constructed Analogs (MACA) dataset, (4) the Analog Resampling and Statistical Scaling Method by Jupiter Intelligence using the Weather Research and Forecasting Model (JupiterWRF). The emission scenarios evaluated include representative concentration pathways RCP4.5 and RCP8.5 from the Coupled Model Intercomparison Project Phase 5 (CMIP5) for the downscaled climate datasets CORDEX, LOCA, and MACA. The emission scenarios evaluated for the JupiterWRF downscaled dataset include RCP8.5 from CMIP5, and shared socioeconomic pathways SSP2-4.5 and SSP5-8.5 from the Coupled Model Intercomparison Project Phase 6 (CMIP6). Only daily durations are evaluated for JupiterWRF.

When applying change factors to the historical NOAA Atlas 14 DDF curves to derive projected future precipitation DDF curves for the entire range of durations and return periods evaluated as part of this project, there is a possibility that the resulting projected future DDF curves may be inconsistent across duration and return period. By inconsistent it is meant that the precipitation depths may decrease for longer durations instead of increasing. Depending on the change factors used, this may happen in up to 6% of cases. In such a case, it is recommended that users use the higher of the projected future precipitation depths derived for the duration of interest and the previous shorter duration.

This data release consists of three shapefiles: (1) polygons of climate regions (Climate_regions.shp); (2) polygons of Areal Reduction Factor (ARF) regions for the state of Florida (ARF_regions.shp); and (3) point locations of NOAA Atlas 14 stations in Florida for which depth-duration-frequency curves and change factors of precipitation depths were developed as part of this project (Atlas14_stations.shp).

This data release also includes 35 tables. Eight tables tables contain computed change factors for the four downscaled climate datasets for the two future projection periods: (1) CORDEX 2020-59 (CF_CORDEX_2040_to_historical.xlsx); (2) CORDEX 2050-89 (CF_CORDEX_2070_to_historical.xlsx);(3) LOCA 2020-59 (CF_LOCA_2040_to_historical.xlsx); (4) LOCA 2050-89 (CF_LOCA_2070_to_historical.xlsx);(5) MACA 2020-59 (CF_MACA_2040_to_historical.xlsx); (6) MACA 2050-89 (CF_MACA_2070_to_historical.xlsx); (7) JupiterWRF 2038-42 (CF_JupiterWRF_2040_to_historical.xlsx); and (8) JupiterWRF 2068-72 (CF_JupiterWRF_2070_to_historical.xlsx).

Twelve tables contain the corresponding DDF values for the historical and future projection periods in each of the four downscaled climate datasets: (1) CORDEX historical (DDF_CORDEX_historical.xlsx); (2) CORDEX 2020-59 (DDF_CORDEX_2040.xlsx); (3) CORDEX 2050-89 (DDF_CORDEX_2070.xlsx); (4) LOCA historical (DDF_LOCA_historical.xlsx); (5) LOCA 2020-59 (DDF_LOCA_2040.xlsx); (6) LOCA 2050-89 (DDF_LOCA_2070.xlsx); (7) MACA historical (DDF_MACA_historical.xlsx); (8) MACA 2020-59 (DDF_MACA_2040.xlsx); (9) MACA 2050-89 (DDF_MACA_2070.xlsx); (10) JupiterWRF historical (DDF_JupiterWRF_historical.xlsx); (11) JupiterWRF 2038-42 (DDF_JupiterWRF_2040.xlsx); and (12) JupiterWRF 2068-72 (DDF_JupiterWRF_2070.xlsx).

Twelve tables contain quantiles of change factors at 242 NOAA Atlas 14 stations in Florida derived from downscaled climate datasets considering: (1) all models and all emission scenarios evaluated for 2020-59 (CFquantiles_2040_to_historical_all_models_allRCPs.xlsx); (2) all models and all emission scenarios evaluated for 2050-89 (CFquantiles_2070_to_historical_all_models_allRCPs.xlsx); (3) all models and only the RCP4.5 and SSP2-4.5 emission scenarios for 2020-59 (CFquantiles_2040_to_historical_all_models_RCP4.5.xlsx); (4) all models and only the RCP4.5 and SSP2-4.5 emission scenarios for 2050-89 (CFquantiles_2070_to_historical_all_models_RCP4.5.xlsx); (5) all models and only the RCP8.5 and SSP5-8.5 emission scenarios for 2020-59 (CFquantiles_2040_to_historical_all_models_RCP8.5.xlsx); (6) all models and only the RCP8.5 and SSP5-8.5 emission scenarios for 2050-89 (CFquantiles_2070_to_historical_all_models_RCP8.5.xlsx); (7) best models and all emission scenarios evaluated for 2020-59 (CFquantiles_2040_to_historical_best_models_allRCPs.xlsx); (8) best models and all emission scenarios evaluated for 2050-89 (CFquantiles_2070_to_historical_best_models_allRCPs.xlsx); (9) best models and only the RCP4.5 and SSP2-4.5 emission scenarios for 2020-59 (CFquantiles_2040_to_historical_best_models_RCP4.5.xlsx); (10) best models and only the RCP4.5 and SSP2-4.5 emission scenarios for 2050-89 (CFquantiles_2070_to_historical_best_models_RCP4.5.xlsx); (11) best models and only the RCP8.5 and SSP5-8.5 emission scenarios for 2020-59 (CFquantiles_2040_to_historical_best_models_RCP8.5.xlsx); and (12) best models and only the RCP8.5 and SSP5-8.5 emission scenarios for 2050-89 (CFquantiles_2070_to_historical_best_models_RCP8.5.xlsx).

Finally, three tables contain miscellaneous information: (1) information about downscaled climate datasets and National Oceanic and Atmospheric Administration (NOAA) Atlas 14 stations used in this project (Datasets_station_information.xlsx); (2) best models for each downscaled climate dataset and for all downscaled climate datasets considered together (Best_model_lists.xlsx); and (3) areal reduction factors by region in Florida (Areal_reduction_factors.xlsx).

An R script is provided which generates boxplots of change factors at a NOAA Atlas 14 station, or for all NOAA Atlas 14 stations in an ArcHydro Enhanced Database (AHED) basin or county (create_boxplot.R). A Microsoft Word file documenting code usage and available options is also provided within this data release (Documentation_R_script_create_boxplot.docx).

Disclaimer: As a reminder, projected future (2020-59 and 2050-89) and historical (1966-2005) DDF curves fitted to extreme-precipitation data from models in each downscaled climate dataset are provided as part of this data release as a way to verify the computed change factors. However, these model-based projected future and historical DDF curves are expected to be biased and only their ratio (change factor) is considered a reasonable approximation of how historically-observed DDF depths might be multiplicatively amplified or muted in the future periods 2020-59 and 2050-89. Some very high outlier change factor values may occur due to the presence of a single very large extreme event in the future period selected for analysis (which causes a very long tail in the fitted distribution) or they may be due to non-convergence of the fitting process. In general, these very high outliers occur extremely rarely in the more than two million change factors computed as part of this project. For example, only 0.008% of the change factors are greater than 8, while only 0.26% of the change factors are greater than 4. The very high outlier values are also more common in the MACA dataset and for the longest return periods which are more uncertain.