Hydrology and Earth System Sciences (Sep 2021)

A 10 km North American precipitation and land-surface reanalysis based on the GEM atmospheric model

  • N. Gasset,
  • V. Fortin,
  • M. Dimitrijevic,
  • M. Carrera,
  • B. Bilodeau,
  • R. Muncaster,
  • É. Gaborit,
  • G. Roy,
  • N. Pentcheva,
  • M. Bulat,
  • X. Wang,
  • R. Pavlovic,
  • F. Lespinas,
  • D. Khedhaouiria,
  • J. Mai

DOI
https://doi.org/10.5194/hess-25-4917-2021
Journal volume & issue
Vol. 25
pp. 4917 – 4945

Abstract

Read online

Environment and Climate Change Canada has initiated the production of a 1980–2018, 10 km, North American precipitation and surface reanalysis. ERA-Interim is used to initialize the Global Deterministic Reforecast System (GDRS) at a 39 km resolution. Its output is then dynamically downscaled to 10 km by the Regional Deterministic Reforecast System (RDRS). Coupled with the RDRS, the Canadian Land Data Assimilation System (CaLDAS) and Precipitation Analysis (CaPA) are used to produce surface and precipitation analyses. All systems used are close to operational model versions and configurations. In this study, a 7-year sample of the reanalysis (2011–2017) is evaluated. Verification results show that the skill of the RDRS is stable over time and equivalent to that of the current operational system. The impact of the coupling between RDRS and CaLDAS is explored using an early version of the reanalysis system which was run at 15 km resolution for the period 2010–2014, with and without the use of CaLDAS. Significant improvements are observed with CaLDAS in the lower troposphere and surface layer, especially for the 850 hPa dew point and absolute temperatures in summer. Precipitation is further improved through an offline precipitation analysis which allows the assimilation of additional observations of 24 h precipitation totals. The final dataset should be of particular interest for hydrological applications focusing on transboundary and northern watersheds, where existing products often show discontinuities at the border and assimilate very few – if any – precipitation observations.