Lightning Lecture AMOS Annual Meeting and International Conference on Tropical Meteorology and Oceanography

Chellenges to homogenise historical radiosonde humidity data (#1018)

Branislava Jovanovic 1 , Robert Smalley 1 , Chris Lucas 2 , Steven Siems 3 , Bertrand Timbal 4
  1. Climate Monitoring and Prediction Section, Comunity Forecasts, Australian Bureau of Meteorology, Melbourne
  2. Science for Services, Science and Innovation, Australian Bureau of Meteorology, Melbourne
  3. School of Earth, Atmosphere and Environment, Monash University, Melbourne, Australia
  4. Centre for Climate Research, National Environment Agency, Singapore

Water vapour plays a major role in the global climate. It is a potent and abundant greenhouse gas and has a strong effect on radiative transfer. It is a key component in the formation of clouds and precipitation and, through the associated latent heating, plays a key role in the transport of energy in the atmosphere. Given its importance, understanding its historical variability and evolution is crucial for understanding the present climate and estimating any future regional climate changes.

Observing water vapour in the free atmosphere is challenging, as absolute concentrations decrease rapidly with increasing altitude. Direct measurements of humidity are made by radiosondes and were historically subject to two reporting biases - warm and dry. Respectively these were caused by cold observations (when ambient air temperature below -40 °C) and dry observations (when relative humidity less than 20%) being reported as missing. Further, there are also measurement biases in the long-term record, caused by the introduction of new radiosonde or sensor types. These biases have a limiting effect on identifying any changes in global atmospheric water vapour. Hence, it is important to develop homogeneous records with the aim of gaining greater confidence in the results of the analysis.

To improve the homogeneity of the long-term data series, daily humidity data (represented as dew point temperature, DWPT) for the period starting in 1987 (i.e. after the Vaisala radiosondes were introduced) were removed if the ambient temperature was below -40 °C or the relative humidity below 20%. This was done based on the formula proposed for the WMO Hygrometer Intercomparison.

Preliminary results related to the development of the bias-corrected Australian monthly humidity data set are presented. Adjustments of the time series were determined using historical metadata and an objective statistical test for break-point detection.