Norrant & Douguedroit 2005 Monthly and daily precipitation trends in the Mediterranean

Référence bibliographique complète

NORRANT C., DOUGUEDROIT A. Monthly and daily precipitation trends in the Mediterranean (1950-2000). Theoretical and Applied Climatology, 2005, Vol. 83, p. 89-106.

Abstract: The aim of the paper lies in the identification of possible significant linear trends at monthly, seasonal and annual timescales in the Mediterranean during the second half of the 20th century. Monthly and daily records of 63 stations have been used to elaborate several precipitation indices. For each index the stations have been gathered together by Rotated Principal Component Analyses to determine 8 sub-areas which can be considered as identical. Trends have been estimated from the scores of each eigenvector retained in all RPCAs. They are mainly non existant or non significant decreasing, even if a few monthly trends appear to be significantly diminishing, primarily during winter months. The significant trends seem mainly related to a number of rainy days RD>10 mm, which represents a high percentage of the total rainfall amount. Greece is remarkable: the sum of daily precipitation SDP>0.1mm and >10mm decrease significantly during January, winter, the rainy season and the entire year whereas SDP>95th percentile increases significantly, in accordance with the climatic change scenarios for the end of this century as does the decreasing of the total monthly and seasonal rainfall.

Organismes / Contact

(1) - Paramètre(s) atmosphérique(s) modifié(s)

(2) - Elément(s) du milieu impacté(s)

(3) - Type(s) d'aléa impacté(s)

(3) - Sous-type(s) d'aléa

Massif / Secteur

Site(s) d'étude

Période(s) d'observation

(1) - Modifications des paramètres atmosphériques

According to previous results the prevailing evolution is a non significant decrease of the precipitation or a lack of linear trend at seasonal and annual time scales, whether during the whole last century or a shorter period. For daily precipitation, a decrease of the number of rainy days and of the occurrence of very extreme precipitation has been shown (Karl and Knight, 1998; Kunkel et al., 1999; Groisman et al., 1999; Easterling et al., 2000b; Palmer and Räisänen, 2002).

On the whole, lack of trend and non-significant trends are greatly prevailing at the monthly and consequently the seasonal and annual timescales. Significant trends represent exceptions isolated in a non significant context. They take on an importance in a single sub-area, Greece.

Monthly, seasonal and annual SDP>0.1mm mainly decrease non significantly. The only significant trends are decreasing during October in the Mediterranean Spain, December in the Lions and Genoa Gulfs, March in the Atlantic region, January, winter and the year in Greece, winter and the year in Italy, winter in the Near East and increasing during April in the Lions and Genoa Gulfs. Monthly SPD>10mm trends are similar to SPD>0.1mm ones. Monthly SDP>95th percentile show more rising trends than the previous indices, even if these increases are in most cases non-significant. From November to January and during March in Greece, December in Italy and January and April in the Gulf of Valencia, SDP>95th percentile significant increasing trends are observed.

Both RD>0.1mm and RD>10mm decrease non significantly for the most part. Significant diminishing trends for both indices are detected during January in Greece and Italy and March in the Atlantic region, and for RD>10mm during October in Greece, November and December in the Gulf of Genoa and February in the Lions and Genoa Gulfs. Exceptions exist: significant increases are observed during October and April in the Lions and Genoa Gulfs and during October and November in the Near East.

Both MDP>0.1mm and MDP>10mm mainly decrease non significantly. The only exceptions where both indices significantly diminish are during December and January in the Near East, February in the Gulf of Genoa and March in the Atlantic region. Seasonal and annual results have similar aspects.

Relationships between SDP>0.1mm and the other indices mainly show a strong significant positive correlation with RD>10mm in most of the area. The number of days influences the change in SDP>0.1mm more than the mean daily precipitation. Significant SDP>0.1mm trends are highly influenced either by the two indices, the number of rainy days and the mean daily precipitation, or by a single one.

The search of linear trends which indicate, in the second half of the 20th century, the appearance of rainfall features presenting characteristics of a future climate change in the regional climate scenarios for the end of the 21st century (2071-2100) in the Mediterranean (Räisänen et al., 2004) gives only limited results: main signals are non significant. Only a few cases of significant decreasing rainfall trends have been found at the monthly, seasonal and annual timescales. A particular case detaches itself from the whole: the Greek sub-area characterized by several significant linear trends. The significant decrease of SDP>0.1mm in January, when much precipitation takes place, leads winter into a significant decrease too. As it is the main rainfall season it induces the same situation during the whole rainy season and the entire year. Similar characteristics are also found for SDP>10mm in Greece. On the other hand, extreme precipitation corresponding with SDP>95th percentile present opposite significant trends; they are significantly increasing during seven months out of the nine rainy ones, the three seasons from autumn to spring, the whole rainy season and the year.

Scenarios from recent regional models, studying Europe and the Mediterranean suggest, for the 2071-2100 period, a concentration of the precipitation marked by the increase of extreme rainfall which goes with a decrease of the precipitation totals (Räisänen et al., 2004). We will only suggest that those scenarios may possibly happen as they have occurred in Greece. But it does not induce that such a change will continue in that country until 2071-2100.

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

The study extends from September 1950 to August 2000. The different timescales used here are mainly the monthly one (limited from September to May) and secondly the seasonal one (the four usual seasons and the rainy season from September to May) and the year. All the rainfall series have been tested by the non parametric Mann-Kendall test (Sneyers, 1990) and can be considered as homogeneous. The network is made up at the monthly timescale by 63 stations including 52 also used for daily analyses. North Africa has only been studied from a monthly totals point of view because daily data are missing.

For each of the 9 months, each of the 5 seasons and the year, the following values have been calculated for the precipitation >0.1mm and >10 mm:
- the amount of precipitation: the sum of daily precipitation index (SDP),
- the frequency of events: the number of days with precipitation or rain days index (RD),
- the intensity of precipitation: the mean daily precipitation index (MDP).

The index of extreme rainfall chosen is the 95th percentile. For each timescale the following values have also been calculated:
- the sum of precipitation >95th percentile index (SDP>95),
- the ratio between the sum of the total precipitation >0.1mm and the sum of precipitation >10mm or >95th percentile.

The regionalization of each of the indices previously presented has been conducted for each month, season and year by using the eigenvectors of Rotated Principal Component Analyses (RPCAs) with a Varimax rotation (Richman, 1986) of R type with the stations as variables and the values of each index as observations. 8 sub-areas have been determined. Linear trends have been calculated on the score series of the eigenvectors corresponding with each precipitation sub-area of all the analyses and on each station used to compute the different RPCAs. They have been tested with the non-parametric Kendall test on ranks at the 0.05 level.

It is possible to estimate the amount of the change of precipitation between 1950-51 and 1999-2000. The a coefficient of each regression of the modeled trends is used to calculate the difference between the two extreme months, seasons or years 1950-51 and 1999-2000.

(2) - Effets du changement climatique sur le milieu naturel

Sensibilité du milieu à des paramètres climatiques

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

(3) - Effets du changement climatique sur l'aléa

Paramètre de l'aléa

Sensibilité du paramètres de l'aléa à des paramètres climatiques

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

(4) - Remarques générales

(5) - Syntèses et préconisations

(1) - Modifications des paramètres atmosphériques
Reconstitutions
Observations	According to previous results the prevailing evolution is a non significant decrease of the precipitation or a lack of linear trend at seasonal and annual time scales, whether during the whole last century or a shorter period. For daily precipitation, a decrease of the number of rainy days and of the occurrence of very extreme precipitation has been shown (Karl and Knight, 1998; Kunkel et al., 1999; Groisman et al., 1999; Easterling et al., 2000b; Palmer and Räisänen, 2002). On the whole, lack of trend and non-significant trends are greatly prevailing at the monthly and consequently the seasonal and annual timescales. Significant trends represent exceptions isolated in a non significant context. They take on an importance in a single sub-area, Greece. Monthly, seasonal and annual SDP>0.1mm mainly decrease non significantly. The only significant trends are decreasing during October in the Mediterranean Spain, December in the Lions and Genoa Gulfs, March in the Atlantic region, January, winter and the year in Greece, winter and the year in Italy, winter in the Near East and increasing during April in the Lions and Genoa Gulfs. Monthly SPD>10mm trends are similar to SPD>0.1mm ones. Monthly SDP>95th percentile show more rising trends than the previous indices, even if these increases are in most cases non-significant. From November to January and during March in Greece, December in Italy and January and April in the Gulf of Valencia, SDP>95th percentile significant increasing trends are observed. Both RD>0.1mm and RD>10mm decrease non significantly for the most part. Significant diminishing trends for both indices are detected during January in Greece and Italy and March in the Atlantic region, and for RD>10mm during October in Greece, November and December in the Gulf of Genoa and February in the Lions and Genoa Gulfs. Exceptions exist: significant increases are observed during October and April in the Lions and Genoa Gulfs and during October and November in the Near East. Both MDP>0.1mm and MDP>10mm mainly decrease non significantly. The only exceptions where both indices significantly diminish are during December and January in the Near East, February in the Gulf of Genoa and March in the Atlantic region. Seasonal and annual results have similar aspects. Relationships between SDP>0.1mm and the other indices mainly show a strong significant positive correlation with RD>10mm in most of the area. The number of days influences the change in SDP>0.1mm more than the mean daily precipitation. Significant SDP>0.1mm trends are highly influenced either by the two indices, the number of rainy days and the mean daily precipitation, or by a single one. The search of linear trends which indicate, in the second half of the 20th century, the appearance of rainfall features presenting characteristics of a future climate change in the regional climate scenarios for the end of the 21st century (2071-2100) in the Mediterranean (Räisänen et al., 2004) gives only limited results: main signals are non significant. Only a few cases of significant decreasing rainfall trends have been found at the monthly, seasonal and annual timescales. A particular case detaches itself from the whole: the Greek sub-area characterized by several significant linear trends. The significant decrease of SDP>0.1mm in January, when much precipitation takes place, leads winter into a significant decrease too. As it is the main rainfall season it induces the same situation during the whole rainy season and the entire year. Similar characteristics are also found for SDP>10mm in Greece. On the other hand, extreme precipitation corresponding with SDP>95th percentile present opposite significant trends; they are significantly increasing during seven months out of the nine rainy ones, the three seasons from autumn to spring, the whole rainy season and the year.
Modélisations	Scenarios from recent regional models, studying Europe and the Mediterranean suggest, for the 2071-2100 period, a concentration of the precipitation marked by the increase of extreme rainfall which goes with a decrease of the precipitation totals (Räisänen et al., 2004). We will only suggest that those scenarios may possibly happen as they have occurred in Greece. But it does not induce that such a change will continue in that country until 2071-2100.
Hypothèses

(2) - Effets du changement climatique sur le milieu naturel
Reconstitutions
Observations
Modélisations
Hypothèses

(3) - Effets du changement climatique sur l'aléa
Reconstitutions
Observations
Modélisations
Hypothèses

Pays / Zone	Massif / Secteur	Site(s) d'étude	Exposition	Altitude	Période(s) d'observation
Mediterranean					1950-2000