3.2 Temperature Climate Extreme Indices Estimated with Uncertainties from U.S. Climate Reference Network (USCRN) Measurements

Monday, 29 January 2024: 2:00 PM
Key 10 (Hilton Baltimore Inner Harbor)
Fabio Madonna, Univ. of Salerno, Fisciano, SA, Italy; and Y. H. Essa, F. Serva, T. Gardiner, F. Marra, and M. Rosoldi

1. Introduction

The relationship between the intensity and frequency of extremes and climate change and their attribution to human activities is fundamental for improving the assessment of risk and elaborating adaptation strategies. Temperature extremes are often reported and estimated using indices, which are handy tools for decision-makers. The development of these indices has also increased our ability to compare models and observations, without the need to handle the raw underlying measurements. However, extremes are quantified neglecting measurement uncertainties. In climate studies, these are often assumed to be smaller than climate variability, especially when individual high-resolution measurements are combined to obtain spatially or temporally aggregated products. On the other hand, It is difficult to estimate uncertainties for historical data records, mainly due to the scarcity of metadata.

The advent of reference measurement networks, as well as the overall increase in observational data quality due to recent technological improvements, allows us to quantify measurement uncertainties in detail. In this paper, we discuss the importance of using measurement uncertainties in the estimation of temperature extremes. Starting from the U.S. Climate Reference Network (USCRN) temperature data and the uncertainty estimation derived for this network data and metadata within the Copernicus Climate Change Service (C3S, cds.copernicus.climate.eu) activities, an extensive assessment of the uncertainty for the climate four climate extreme indices (Frost Days, Ice Days, SUmmer days and TRopical days) and is presented in the this paper (Madonna et al., 2023).

2. Dataset and methodology

The U.S. Climate Reference Network (USCRN, Diamond et al., 2013) stations provide measurements of NST, precipitation, wind speed, soil conditions, and other ancillary variables (https://www.ncdc.noaa.gov/crn/). Stations are equipped with high-quality redundant sensors to measure temperature and precipitation.

Within an activity of the Copernicus Climate Change Service (C3S) aiming to facilitate access to international reference and baseline networks (https://climate.copernicus.eu/access-observations-baseline-and-reference-networks), the USCRN data are made available through the Climate Data Store (CDS) along with the uncertainty budget.

The main error sources in the USCRN NST product (Palecki et al., 2013) relate to (in brackets it is reported if random or systematic at the raw measurement resolution):

  • the instrument used (random),
  • the data logger used (systematic),
  • the interface between the instrument and the data logger (systematic), and
  • external effects that result from sources such as local weather (systematic).

In the USCRN NST dataset from CDS, measurement uncertainties are asymmetric as they include several systematic error sources that cannot always be properly estimated and adjusted.

3. Results and conclusions

In the top panel of Figure 1, the values of the FD with and the corresponding uncertainty range obtained from USCRN are shown for the station in Baker, Nevada. As clarified above, being the USCRN measurement uncertainties asymmetrical, an asymmetry also occurs for the uncertainties on the extreme climate indices. The negative uncertainty on the values of FD for the period 2006-2021 is up to 8 days while the positive uncertainty is up to 13 days.

Figure 1: Top panel, number of Frost days (FD) and the related 2-sigma uncertainty range for the USCRN station of Baker 5 W (Identifier: 53138, 39.01000, -114.2000) calculated using the near-surface temperatures provided by USCRN and the measurement uncertainties estimated for the Copernicus Climate Change Service (C3S); bottom panel, minimum and maximum cumulative number of TRopical nights (TR) calculated for the USCRN stations below 50N in the period 2006-2021 obtained considering negative and positive uncertainties.

The bottom panel of Figure 1 shows the minimum and maximum cumulative number of TR for the USCRN station between 50N and 25N, over the period 2006–2021. The minimum and maximum numbers are obtained by accumulating over 15 years the value at ±2-sigma of the TR index. Therefore, the minimum number is the most conservative estimation of an index including all the values which have a probability higher than 95.4% to be an extreme value, while the maximum number includes all the values with a probability higher than about 5%. The cumulative values show the latitude dependence of the uncertainties on the TR, behavior shared with the other indices. The minimum cumulative value is smaller than 75 days while the maximum value is smaller than 175 days. This result offers an example of the importance to consider the uncertainty in the number of days with the most extreme NST conditions for any long term adaptation strategy.

In conclusion, uncertainties are indispensable to compare any physical quantity with a predefined threshold or theoretical constant/value, as occurs in the calculation of climate indicators. Their estimation allows us to reliably quantify the range of values the climate extreme indices may assume. Quantifying measurement uncertainties in the evaluation of climate indices can improve the validation of any dataset and increase confidence in the quantification of temperature extremes frequency, with an enhanced capability to improve the assessment of risks. It is important to clarify that the measurement uncertainties for USCRN are, in most cases, smaller than those attributable to measurements provided by other baseline networks, such as the GCOS Surface Network (GSN), and reflected in the homogenized datasets. This is due to the high-quality redundant equipment and the ancillary sensors available at each USCRN site, as well as to the collection of detailed metadata, poor or missing in baseline or comprehensive data sets. This implies larger uncertainties for the indices estimated from other measurement networks

4. References

Diamond H., and Coauthors, 2013: U.S. climate reference network after one decade of operations: Status and assessment. Bulletin of the American Meteorological Society, 94(4), 485–498. https://doi.org/10.1175/BAMS-D-12-00170.1

Madonna and Coauthors., 2023: Uncertainties on climate extreme indices estimated from U.S. Climate Reference Network (USCRN) near-surface temperatures. Journal of Geophysical Research: Atmospheres, 128, e2022JD038057. https://doi.org/10.1029/2022JD038057

Palecki, M. A., and Coauthors, 2013: U.S. climate reference network products [Near Surface Temperature]. NOAA National Centers for Environmental Information. https://doi.org/10.7289/V5H13007

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner