Adapted Caussinus-Mestre Algorithm for Networks of Temperature series (ACMANT)

Abstract

Any change in technical or environmental conditions of observations may result in bias from the precise values of observed climatic variables. The common name of these biases is inhomogeneity (IH). IHs usually appears in a form of sudden shift or gradual trends in the time series of any variable, and the timing of the shift indicates the date of change in the conditions of observation. The seasonal cycle of radiation intensity often causes marked seasonal cycle in the IHs of observed temperature time series, since a substantial portion of them has direct or indirect connection to radiation changes in the micro-environment of the thermometer. Therefore the magnitudes of temperature IHs tend to be larger in summer than in winter. A new homogenisation method (ACMANT) has recently been developed which treats in a special way the seasonal changes of IH-sizes in temperature time series. The ACMANT is a further development of the Caussinus-Mestre method, that is one of the most effective tool among the known homogenising methods. The ACMANT applies a bivariate test for searching the timings of IHs, the two variables are the annual mean temperature and the amplitude of seasonal temperature-cycle. The ACMANT contains several further innovations whose efficiencies are tested with the benchmark of the COST ES0601 project. The paper describes the properties and the operation of ACMANT and presents some verification results. The results show that the ACMANT has outstandingly high performance. The ACMANT is a recommended method for homogenising networks of monthly temperature time series that observed in mid-or high geographical latitudes, because the harmonic seasonal cycle of IH-size is valid for these time series only.

Benchmarking homogenization algorithms for monthly data

“…The time series are centered by subtracting the mean because homogenization aims to improve the temporal consistency of the data, not the absolute level. The first four contributions in Figure 1 -ACMANT (Spain) [18], PRODIGE monthly (Meteo France) [19], USHCN main (NOAA, USA) [4] and MASH main (Hungarian weather service) [20,21] are all direct homogenization algorithms and clearly perform better than the traditionally used SNHT method [22], here exemplified by CSNHT.…”

Tendencias recientes e índices de cambio climático de la temperatura y la precipitación en Andorra, Pirineos (1935-2008)

“…HOMER añade -gracias a la incorporación del algoritmo descrito por Pickard et al (2011)-la posibilidad de realizar la detección simultánea en el conjunto de series. Ambas técnicas se superponen y se ayudan -en el caso de la temperatura-del método de detección ACMANT (Domonkos, 2011). Finalmente, la consulta del metadato permite introducir/eliminar puntos de cambio que aparecen como artificio estadístico.…”

The joint influence of break and noise variance on the break detection capability in time series homogenization

“…Furthermore, V can be interpreted as that fraction of variance which is explained by the chosen segmentation. The decomposition into internal and external variance is applied in the state-of-the-art break search algorithms, such as PRODIGE (Caussinus and Mestre, 2004) and ACMANT (Domonkos, 2011), which both use the maximum external variance V max (k) as a break criterion that determines the true break positions. V max (k) is defined as the maximum variance attainable by any combination of break positions for a given number of breaks k. This optimum segmentation containing the maximum external variance V max is determined by using the optimal partitioning approach (Bellman, 1954;Jackson et al, 2005) separately for each number of breaks k. In this approach, the multiple re-use of solutions from truncated sub-series speeds up the search drastically (compared to a test of all combinations), which is described in more detail in Lindau and Venema (2013).…”

“…Although a simultaneous correction of all breaks is more accurate , most algorithms analyzed by Venema et al (2012) correct break by break beginning today and moving backward in time, such as PHA (Menne et al, 2009), iCraddock, (Craddock, 1979;Brunetti et al, 2006), AnClim (Štépanek et al, 2009), and the various SNHT variants (Alexandersson and Moberg, 1997) that participated. HOME recommended five homogenization methods: AC-MANT (Domonkos, 2011), PRODIGE (Caussinus and Mestre, 2004), MASH (Szentimrey, 2007(Szentimrey, , 2008, PHA, and Craddock. These methods have in common that they have been designed to take the inhomogeneity of the reference into account, either by using a pairwise approach (PRODIGE, PHA, Craddock) or by carefully selecting the series for the composite reference (ACMANT, MASH).…”