detection of changes in climate state, climate variability ...
Meteorology and Hydrology
№4, 2004
DETECTION OF CHANGES IN CLIMATE STATE, CLIMATE VARIABILITY AND CLIMATE EXTREMITY Gruza G., Rankova E., Institute for Global Climate and Ecology (IGCE), Russia (e-mail: [email protected]) Abstract. Refined definitions of a climate, climatic variables, climate variability, climate changes, and climate monitoring are given that are applicable to the analysis of climate variability from observations in the case of a changing climate. The main features of global warming are described. New data are reported on the features of climate changes over the Globe and Russia territory in the second half of the 20th century including changes in climate anomality, climate extremity and climate variability. 1. World climate program and climate monitoring. Definition of climate. Global warming has been recognized since the 1970s of the 20th century. The amount of researches concerning climate oscillations, both natural and anthropogenic, has increased sharply since then. It is becoming evident that the monitoring of the current climate state is required. The first statement on a threat to the global climate was made by the World Meteorological Organization (WMO) in 1976. In 1979, after the First World Climate Conference, the WMO established the World Climate Program (WCP), which laid the foundations for international activity on climate [WMO, 1992]. WCP activities are aimed at improving the monitoring of the climate system. In 1992, the United Nations Framework Convention on Climate Change (UNFCCC) was established. The Intergovernmental Panel on Climate Change (IPCC) was established by the WMO/UNEP for scientific support of UNFCCC activities. Its goals include the comprehensive and objective estimation of observed and expected climate changes and anthropogenic impacts. The IPCC has published three scientific reports on climate change: IPCC-1990, IPCC-1995, and IPCC-2001 [IPCC, 2001]. They provide the basic data on the observed climate and its variations, climate models, and the degree of agreement between model results and observations. Definition of Climate. Let us refine the term climate as applied to the analysis of a climate and climatic variability from observed data in the case of a changing climate. Weather is defined as the physical state of the atmosphere at a given point of the globe at a given time. The state of the atmosphere is characterized, in particular, by air temperature, wind velocity, humidity, precipitation, solar radiation, clouds, and phenomena such as fog, hoarfrost, hail, and other weather variables (weather elements). In a narrow but widely used sense, the climate is a generalization of weather changes and is represented by a set of weather conditions in a given spatial area over a given time interval. A statistical description in terms of means, extremes, variability indices for certain parameters, and frequencies of events over a given time period is used for climate characterization. All of these descriptive statistics are called climatic variables. According to WMO recommendations, a period of three decades is used as a standard period for the estimation of climatic variables characterizing the current climate. Nowadays, this is the 1961-1990 period.
Meteorology and Hydrology
№4, 2004
In modern research, the term climate is also used to denote the global climate, which is characterized by a set of states of the global climate system during a given time interval. The global climate system consists of five basic components: the atmosphere, hydrosphere, cryosphere, land surface, and biosphere. Their interactions have a large effect on weather fluctuations over long time intervals and are responsible for climate formation and climate changes. Climate Variability. The spectrum of changes in meteorological and oceanic characteristics is continuous. As in most aperiodic processes, its density tends to infinity only for periodic components and their harmonics---the annual and diurnal components. The contribution of periodic processes to the total variance is finite and can be estimated for the annual and diurnal cycles if their amplitudes are known. Based on A.S. Monin's (the most complete) classification of oscillations of meteorological and oceanic parameters depending on their scales, the following scales are usually distinguished: o Micrometeorological variability: from fractions of a second to several minutes. o Mesometeorological variability: from several minutes to several hours. o Synoptic variability: from several hours to two--three weeks. Individual predictions and the description of basic synoptic objects characterizing weather and its changes are possible in this range of scales. The upper boundary of the range is frequently taken to be the predictability scale of individual synoptic processes, which is estimated as two--three weeks. o Climate variability: from three weeks to several decades. Variability on such a scale, taken here as the inner time scale of the climate system, characterizes internal climatic oscillations, or climate variability, or climate fluctuations. In many applications, the upper boundary of this range is reasonably taken equal to about three decades. In particular, a 30-year interval (from 1961 to 1990 at present) was adopted by the WMO to be the standard period for the estimation of climatic normals. Scales longer than several decades characterize climate changes. It is reasonable to introduce additional groups of time scales within this region. o Secular variability. o Variability of the Little Ice Age type. o Variability corresponding to glacial periods. As mentioned above, a special consideration should be given to diurnal and annual oscillations and their harmonics. The object of research in CLIVAR, a basic component of the World Climate Research Program, is climate variability corresponding to the following three time intervals: a) seasonal to annual, b) annual to decadal, c) decadal to centennial. Thus, CLIVAR does not include the variability corresponding to time intervals from three weeks (predictability limit) to three months (season), which is the traditional object of longterm weather forecasts or predictions of short-term climate oscillations. In our view, shortperiod climate oscillations from three weeks (one month) to three months deserve a special consideration. Short-term climate oscillations play an important role in the formation of seasonal climate anomalies in the midlatitudes. In this respect, large-scale variations in the properties of westerlies, such as the index cycle and blocking processes, are of most considerable interest.
Meteorology and Hydrology
№4, 2004
Some quantitative assessments of the surface air temperature variability are presented in [Gruza&Rankova, 1980]. Climate Changes. The definition of a climate introduced above makes it possible to use, as climatic variables, any statistical characteristics of any parameters describing the state of the climate system over a given interval of time. It is only necessary to indicate the characteristic under consideration and the corresponding time interval. A climate change in a given region or over the Globe is characterized by the difference between some climatic variables for two given intervals of time. A climate change can be considered actual if it exceeds the probable error of the corresponding calculated climatic variables or can be considered statistically significant in the framework of the stochastic climate model (hypothesis) adopted if that change exceeds a prescribed level of significance. Climate changes can be caused by both natural internal and external factors and anthropogenic activities. In Article 1 of the UNFCCC, a climate change is defined in a narrower sense as "a change of climate which is attributed directly or indirectly to human activity that alters the composition of the global atmosphere and which is, in addition to natural climate variability, observed over comparable time periods." Thus, the UNFCCC distinguishes between climate changes caused by human activities altering the atmospheric composition and climate variability caused by natural factors. One should bear in mind that only total climate changes caused by natural and anthropogenic factors can be estimated from observations. The determination of the causes of climate changes and the estimation of corresponding effects are complicated problems, which can now be solved by applying climate models. Following scientific practice, we will adhere to the above-given, more general definition of climate changes, regardless of their causes. Thus, any climatic variables can be used for scientific analysis of climate variability and changes, and any base periods (including those different from 30 years) can be used for estimating deviations from averages. It is only important to explicitly introduce exact definitions and adhere to them while describing results. 2. Global warming The most prominent feature of the climate in the 20th century is global warming, which is characterized by a rise in the surface air temperature as averaged over the globe, each hemisphere, or most large continental or oceanic regions. A general view of global warming can be given by time series of globally averaged surface temperature [Jones et al., 1999] (over the oceans, the air temperature is replaced by the sea surface temperature). Figure 1 shows such series of annual mean temperature anomalies over 1856--2001 for the globe and both hemispheres (before 1856, the temperature data in the Southern Hemisphere are sparse). Global warming in the 20th century has been recognized to be an unprecedented event over the last 1000 years: the annual mean global surface air temperature increased by 0.6 ± 0.2oC over the century. However, global warming was found to be nonuniform in time. Three intervals can be distinguished: warming over 1910-1945, weak cooling in 1946-1975, and the most intense warming starting in 1976. The warmest decade occurred in the 1990s, and the warmest year was 1998. The ten warmest years took place after 1983, and eight of them occurred starting in 1990. The year 2000 was the 22nd in the continuous sequence of years with a global mean temperature exceeding the 1961-1990 normal.
Meteorology and Hydrology
№4, 2004
Northern Hemisphere
Southern Hemisphere
Global
Figure 1. Time series of annual surface air temperature anomalies averaged over the Northern Hemisphere (a), Southern Hemisphere (b) and the Globe (c), 1856-2001. The datasets of the East England University (CRU) are used. The anomalies were calculated as deviations from the 1961-1990 means. It is necessary to note, that changes in the global air temperature in individual months (here are not presented) are substantially different from changes in the annual means.
Meteorology and Hydrology
№4, 2004
3. Climate monitoring over Russia: goals, tasks, and practical implementation Allowance for climate changes and for large climate anomalies is important in decision making concerned with power industry, industry, and nature management. Recognized by the world community, this importance has led to the need for a system of continuous observations of the climate (climate monitoring). The system must include the collection and generalization of climate data, the permanent estimation of current climate anomalies and climate changes, and the dissemination of climatic information among governing bodies and the public. In Russia, a system of climate monitoring is being elaborated under the guidance of Academician Yu.A. Izrael at the Institute of Global Climate and Ecology (IGCE), the Russian Federal Service for Hydrometeorology and Environmental Monitoring (Rosgidromet) and the Russian Academy of Sciences. The concept of environmental monitoring was developed by Izrael in 1974 in a work devoted to the foundations of monitoring [Izrael, 1974], where the term “environmental monitoring” first appeared in the Russian literature. This term meant observations or measurements with the aim of control or decision making. In [Gruza&Rankova, 1989] the basic goals of climate monitoring are stated as follows: - regular observations of the climate system state, including the collection and generalization of climatic data and the determination of the characteristics of the current earth's climate; - the probability estimation of the degree of anomaly of the climate system state; - the detection of natural and anthropogenic causes of observed anomalies. The goals of monitoring also include the estimation of the scales of probable climate changes and oscillations in the future. Regular climate monitoring over the Russian Federation is performed by scientific institutions of the Roshydromet (IGCE, the All-Russia Research Institute for Hydrometeorological Information--World Data Center (VNIIGMI--MTsD), the Hydrometeorological Center of Russia, the Main Geophysical Observatory, the Arctic and Antarctic Research Institute, the State Hydrological Institute, and the St. Petersburg Division of the State Institute of Oceanography). This makes it possible to estimate observed climate anomalies in quasi-real time. Climate monitoring is based on operational observation data (received through communication channels) and includes the preparation of monthly, seasonal, and yearly bulletins. An analysis of regional climates is based on spatially averaged anomalies of meteorological variables (deviations from the values averaged over the base period 1961-1990). They are calculated by averaging point data (station observations) over equal-area latitude--longitude boxes followed by averaging over a region. The domain of analysis (globe, hemisphere, or a region) is divided into equal-area boxes of given size, and the arithmetic mean of the station values of the variable under analysis is calculated inside each box. Next, the resulting values are averaged with weights proportional to the area of the intersection of boxes with the territory of the region. Seasonal means of anomalies of meteorological variables are calculated by averaging monthly anomalies at stations for winter (December--February), spring (March--May), summer (June-August), and fall (September--November) when data for at least two months in a season are
Meteorology and Hydrology
№4, 2004
available. Annual means are calculated for a "seasonal" year (December--November) by averaging seasonal anomalies when data for all four seasons are available. Monthly bulletins "Climate Monitoring Data" have been issued since 1984. The ideas and techniques used for preparing data for monthly climate monitoring underlay the technique for preparing yearly climate monitoring bulletins "Climate Changes over Russia," which have been issued since 1997 and include seasonal and annual anomalies of air temperature and precipitation and some climatic indices. Since 1999, bulletins "Climate Changes over Russia" have been available on the web-site http://climate.mecom.ru . Materials of the annual and seasonal generalizations "Climate Features and Climate Anomalies over the Russian Federation Territory" have been included in yearly national reports on the state of the environment in Russia since 1999 [Review, 1999]. Meteorological Data for Climate Monitoring. A climate-monitoring database has been compiled at the IGCE. It includes time series of basic climatic parameters over the available observation period, reanalyses, and climate model outputs based on data from the world's leading meteorological centers. Current data used for extending the existing time series and for implementing climate monitoring are formed from operational observations received through communication channels and are included in the database in quasi-real time. The observed data in the climate-monitoring database are contained in two data sets: station data (compiled and supported by the IGCE) and data at the nodes of a uniform grid (produced by the UK Meteorological Office and supplemented via the Internet). The set of station data contains monthly mean air temperatures and monthly precipitation totals measured at stations of the global meteorological observation network starting in 1886. All stages of compiling the data sets involved data verification and control (including statistical control with the use of temporal and spatial coupling between observations). The statistical homogeneity of the series was ensured by introducing corrections that compensate for changes in instruments, observational techniques, station sites, etc. Stations located at sites with a highly increasing population were not included in the data set. The set of gridded data is known as the combined data set of temperature anomalies over land/sea in five-degree boxes (Jones et al., 1999) and covers the period of time starting in 1856. In the IPCC report [9], this data set was qualified to be the most compatible (in terms of observation homogeneity and quality) with the requirements imposed on data used for empirical climate studies and climate change detection. The data over land were based on the basic data set of United Kingdom station reports [Jones, 1994] and were calculated by averaging the station reports within each grid box. The number of grid boxes was 2592 (1296 in each hemisphere), and the number of all stations used was greater than 2500, but their number after 1991 reduced to 800-1000. The data over sea were obtained by averaging the sea surface temperatures (over grid boxes and calendar months of each year) contained in the COADS data set [Folland & Parker, 1995]. In operational climate monitoring, climate changes are estimated by using the data set, which includes 1383 stations where the time series of air temperature and atmospheric precipitation start no later than in 1951. The longest series in the data set contain data beginning in 1986 (although data over earlier years are available for individual stations). The dataset includes data from 1383 stations (global set). The former USSR territory is represented there by 455 stations, of which 315 are located in the Russian Federation.
Meteorology and Hydrology
№4, 2004
Figure 2 reflects the history of developing these networks of stations. Continuous observations during the entire century are available only at 156 stations in the former Soviet Union. Most of these stations are located in European Russia. Most of the time series of observations at stations of the Asian part of country begin in the mid-1930s. By 1936, the number of operating stations (among the 455 chosen) had been 338. All of the 455 stations operated from 1952 to the late 1980s. Some of the stations were closed in the last decade, others appeared outside of Russia. A thorough inspection and verification of all available observations over the last 10--15 years is now being conducted in IGCE in order to maximally reduce a network destruction observed in the last years (first of all, in applied to the 1383 stations chosen).
Global (max=1383) f. USSR (max=455)
Figure 2. Variations of the number of meteorological stations over the Globe and former USSR as presented in the IGCE climatic datasets, 1886-2000. The database is supplemented with current data from CLIMAT records received through communication channels at the Roshydromet Main Computer Center and at the All-Russia Research Institute of Hydrometeorological Information – World Data Centre (system for the storage of data in archive format). Data quality control is performed by comparing data from the two sources and by comparing them with synoptic observations. Data verification and quality control are based on special procedures for checking the consistency of observations in time and space, on data published in bulletins etc. In conclusion, it should be noted once again that a stable network of stations had been formed only by the 1940s. This network has been degraded severely during the last decade. The extension of the network and the elimination of various data inhomogeneities (data homogenization) have become high-priority problems to be solved in the operating climate monitoring system over Russia. Indicators of a Changing Climate and Dynamic Normals. Statistical climate characteristics are widely different in different regions and over different time intervals. The table lists basic
Meteorology and Hydrology
№4, 2004
statistical characteristics useful for describing climate states and changes in a certain region for two time intervals: Temporal and spatial averages characterize climatic variability in time and space, respectively. In [Rankova & Gruza, 1998; Gruza et al., 1999] a number of characteristics calculated from monthly mean air temperatures and monthly precipitation measured at stations within a region are used for quantitative estimation of regional climate changes. Temporal statistics Climate state Period-1: E1 E(E1)=EE1 S(E1)=SE1 Q(E1)=CAI1
Spatial statistics E =Mean S = std Q = RMS
Climate change DE=E1-E2 E(dE)=EdE S(dE)=SdE Q(dE)=CCI
Period-2: E2 E(E2)=EE2 S(E2)=SE2 Q(E2)=CAI2
In particular, useful indicators of climatic variability and regional climate changes are the climate anomaly index (CAI) and the climate change index (CCI). The climate anomaly index is defined as the Euclidean distance between the point describing the current climate state and the point representing the temporally averaged climate (normal):
CAI
y
=
2 1 n z yi − z i n ∑ i =1 σ i
(1)
The greater the CAI, the longer the distance between the point representing an instantaneous climate state and the "center" of climate state points. The climate change index is defined as the Euclidean distance between the two points describing the climate states at times y1 and y2:
CCI =
2 z −z n 1 y1i y2 i n ∑ σi i =1
(2 )
The concepts of CAI and CCI are illustrated in the last line of the table, which helps understand the meaning of these characteristics. Another fairly informative characteristic is the climate extreme index (CEI), which is defined for a particular meteorological parameter at a particular time (period) as the fraction of the area of a region characterized by extreme current anomalies of that parameter (e.g., these are anomalies lying within the10%-intervals in both tails of the climatic distribution of the parameter). Changes in climate extremeness and anomaly and in the intensity of climate changes were assessed from surface air temperatures over 1951-2000 for the entire Russia and some of its regions. It was found that the CAI weakly correlates with the CCI, but they both (and the CEI) have increased during several last decades (see more detailed in [Gruza&Rankova, 2003]). This fact is shown in part in Figure 3, which contains a schematic diagram illustrating specifications of CAI and CCI indices in 2-dimensional space: regionally averaged annual temperature over Northern Hemisphere (Y-axis) versus that of Southern Hemisphere (X-axis).
Meteorology and Hydrology
№4, 2004
Also, it is evident that the temperature changes over the last three--four decades were much greater. 0.8
VT, Northern Hemisphere
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -0.8
-0.6
-0.4
-0.2 0.0 0.2 VT, Southern Hemisphere
0.4
0.6
0.8
Figure 3. A schematic diagram illustrating specifications of CAI and CCI indices in 2-dimensional space: regionally averaged annual temperature over Northern Hemisphere (Y-axis) versus that of Southern Hemisphere (X-axis). Scatter diagram is based on CRU-datasets for the period 1956-2001 (correspond to time series from Figure 1). Correlation coefficient is 0.72. Further it is interesting to evaluate the extent to which the observed increase in climate anomaly, extremeness, and intensity of change can be attributed to climate warming. It turns out that the traditional set of climatic variables usually included in climate handbooks is insufficient for the time intervals under study in the case of a changing climate. As a rule, when one chooses climatic variables or a time interval for their estimation, the climate is assumed to be steady, which disagrees with the view of the current climate. In particular, it is far from reasonable to use the standard 30-year interval for calculating climatic normals in all cases. On the whole, it is clear that methods and algorithms designed for nonstationary random processes and their realizations would be useful in the analysis of climatic time series. Recall that, according to the definition of climatic variables given above, they include moving averages (e.g., decadal means), filtration results containing low-frequency components of processes, and trends identified in accordance with a certain hypothesis. At every particular instant of time, all of these quantities can be regarded as dynamic climatic normals, which could give (in a sense) a better characterization of the current climate than that provided by standard normals. Dynamic normals will be analyzed elsewhere. This work is restricted to linear trends and smoothed moving averages.
Meteorology and Hydrology
№4, 2004
4. Estimates of climate changes over Russia in the second half of the 2oth century. Surface Air Temperature. Figure 4 display time series of annual air temperature anomalies regionally averaged over the Russia territory. At the left the anomaly scale on the vertical axis corresponds to deviations from the mean value over 1961-1990 base period, while at the right anomalies are shown relative to a 20-year interval in the late 19th century. It is evident that the climate in Russia in the 20th century differed from that in the 19th century. It is useful to note, the year-to-year air temperature oscillations in an individual region (European Russia) were much greater than in the Russia as a whole.
a) а) 1.0
1.5
b)б)
2.0
1.0 1.5
0.5
0.5
0.0
1.0
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
0.0
-0.5
-0.5
-1.0
-1.0
1880 1900 1920 1940 1960 1980 2000
1880 1900 1920 1940 1960 1980 2000
Figure 4. Annual air temperature anomalies averaged over the Russian Federation territory. The columns show deviations from the 1961-1990 means (at the left) and from the late preindustrial period means (at the right, 1886-1905). The curves show the 11-year moving average. Warming per century (1901-2000) over Russia averaged to 0.09oC/10 years. The maximum of warming in Russia was observed in 1995 (1.9oC deviation from the normal). In the second half of the 20th century (1951-2000), the general tendency of change in the annual mean air temperature over Russia was also characterized by a positive trend. The largest trend occurred in Transbaikalia and the Baikal region (0.35oC/10 years), in the Amur and Maritime regions, and central Siberia. Large positive temperature anomalies have been observed in these regions for the last 11-12 years. For Russia as a whole, warming was more pronounced in winter and spring (with trends of 0.47 and 0.29oC/10 years, respectively). An analysis of the spatial patterns of the surface air temperature trend coefficients over 19512000 for the year as a whole and the warm and cold half-years (Figure 5) shows that the climate changes are highly inhomogeneous over the territory of Russia. It is particularly important that, in the warm period, the temperature increase is much weaker and areas of warming alternate with areas of noticeable cooling, including many grain-producing areas.
Meteorology and Hydrology
№4, 2004
oC/10 yrs
Figure 5. Linear trend coefficients pattern for the annual surface air temperature over the territory of Russia (January-December). The estimates were obtained from the station observations over 1951-2000 and are given in oC/10 years. Atmospheric Precipitation. Atmospheric precipitation is extremely important for various aspects of human activities (agriculture, power industry, transport, hazardous events associated with floods and droughts, etc.) and for the climate system (clouds, latent heat fluxes, fresh water inflow to the ocean, accumulation/destruction of ice sheets and mountain glaciers, etc.). However, the climatology of precipitation has been investigated much more poorly than the climatology of temperature. Precipitation over the ocean has been poorly studied. О тенденциях изменения режима осадков на территории России можно судить по Рис.6, где приведено пространственное распределение оценок трендов осадков во второй половине столетия для года в целом The tendencies of changes in precipitation regime over the territory of Russia can be seen from Fig. 6, where spatial patterns of the precipitation totals trends in the second half of the 20th century are presented for the year as a whole. It can be seen that the annual and seasonal precipitation totals for the entire Russia and its eastern regions tend to decrease in the last 50 years. The most pronounced decrease in precipitation occurs in the northeastern region of the country. A weak tendency of precipitation to increase is observed over European Russia. For comparison, we recall that climate models predict an increase in globally averaged precipitation in response to an increased CO2 concentration. The winter precipitation is expected to increase in the high latitudes and, according to most models, in the midlatitudes. For most latitudes, models predict a mainly moderate increase in precipitation under doubled CO2, accompanied by a significant growth of the frequency and strength of heavy precipitation, especially in the tropics and the midlatitudes of the Northern Hemisphere. The
Meteorology and Hydrology
№4, 2004
expected rise of temperature contrasts between the continents and the ocean can intensify monsoons. In particular, increased precipitation is expected in the eastern Asian monsoon system. 0
30
60
90
120
150
180
80
80
60
60
40
40
20
20
0
0 0
%/10 %/10лет yrs
30
-20% -16% -12%
60
-8%
90
-4%
-0%
120
+0%
+4%
150
180
+8% +12% +16% +20%
Figure 6. The same as Fig. 5 but for the annual precipitation totals (%/10 years). Percents are calculated relative to 1961-1990 normals. Thus, the empirically estimated trends presented here poorly agree with model computations. We conclude that a further careful comparative analysis of simulated and empirical estimates is required and that a tendency of the annual and seasonal precipitation values to somewhat decrease in eastern regions of Russia over the last 50 years can be inferred from observations. 5. Assessments of changes in climate variability over Northern Hemisphere continents in the 20th century. For evaluating climatic variability, characteristics (measures) of variability and calculation methods for them were proposed. The characteristics of climatic variability were estimated and analyzed from surface air temperatures over the Northern Hemisphere. Root mean squared (or mean absolute) deviations from the average are used as a measure of variability (scatter). For nonstationary processes, total variability is assumed to split into variability attributed to a trend of the mean and residual variability associated with deviations from the trend. The latter is estimated as the RMS or mean absolute deviation from the trend. Assuming that the deviations from the trend are also subject to slow variations, the corresponding trend of variability is reasonably defined as the trend of absolute or squared deviations from the trends of means. It is this trend of absolute deviations from the trend of means that considered here to be the trend of variability. A comparison was performed for two data sets of monthly mean air temperatures (on stations and in 5*5 boxes) taken from the database of Northern Hemisphere climatic data described
Meteorology and Hydrology
№4, 2004
above. The variability trend was estimated in each point for two periods (1901-1998 and 1951-1998) and three seasons: the calendar year and the warm and cold half-years. Stations with sufficiently complete series (with the observation gaps not exceeding 10% of the years) were only considered here. Four models were used to estimate the trend of the mean air temperature at each station: “Y” - regression on time (usual linear trend), “C” - regression on the mean carbon dioxide concentration, “G” regression on the mean global temperature, and “M” - filtered series involving oscillations with periods more than 10 years. For all four models, the trend of variability was estimated as a linear trend of absolute deviations from the trend of means. Thus, a set of linear trends of variability (AY, AC, AG, and AM) was obtained for each of points. The results were presented on 24 patterns over the Northern Hemisphere continents (four models * two periods * 3 time scales). An example of this comparison is reproduced in Figure 7, containing some fragment from AC-results. The conclusions suggested by their analysis are mainly the following. 80
(а)
60
40
20 80
-120
-60
-120
-60
0
60
120
(b)
60
40
20
-180
0 o oC/100 C/100 лет yrs
60 -1.0 -0.5 -0.2
0
120 0.2 0.5
180 1.0
Figure 7. The patterns of linear trends in the climate variability indicators (from model AC). The variability indicators are defined here as absolute deviations (in local points) of annual temperature from its regression on CO2-concentration. Estimates are based on station observed data (upper pattern) and gridded data in 5*5 boxes (lower pattern) for the period 1951-1998. The results for AY, AC, and AG are similar. The variability over the entire century (19011998) was found to increase in the cold season over nearly the entire Eurasia and in central US regions. In the warm season, variability increased only in central Eurasia and southern US regions. The variability in Europe, northern US regions, and Japan decreased substantially. At the same time, the trends over the second half of the century (1951-1998) suggest increased variability over Eurasia within 40-60N, in Japan, eastern US regions, and Canada. A considerable decline in variability was observed manly in the cold season over individual regions of Russia and in eastern and western regions of Canada. The variability estimates AM look somewhat differently: an increase in variability in all seasons occurred only over eastern
Meteorology and Hydrology
№4, 2004
US regions. In the cold season, the variability over most of the territory declined. A spatially inhomogeneous distribution of variability estimates was observed over Europe. Thus, the characteristics of the variability of annual air temperature anomalies varied on scales of 10-100 years. The climate variability was observed to enhance (especially pronouncedly in 1951-1998) over Eurasia within 40-60N, in eastern US regions, and Japan. Although the magnitudes of these changes are frequently rather small, but they are of the same sign for different data sets and estimation methods. The stability of the results confirms that the climate variability and the frequency of extremes increased. 6. Assessment of the climatic response of surface air temperature to changes in carbon dioxide and aerosol atmospheric concentrations The estimates of climatic response of surface air temperature to changes in the carbon dioxide and atmospheric aerosol concentrations were obtained from observations and model results (produced by the transitive global ocean--atmosphere climate model HadCM2 at the Hadley Center of the UK Meteorological Office). A detailed analysis of estimates derived from observed and simulated data and their comparison are presented and illustrated in [Gruza & Rankova, 1999]. We note here only the basic result: Under the CO2 concentrations and the scales of their changes considered here (up to 50 ppmv) we can adopt that a local change in air temperature is a linear function of the change in CO2 concentration (rather than being a logarithmic function accepted in the general case for the global surface air temperature). The spatial patterns and the magnitudes of the simulated and observed climate responses are widely different (when analyzing the observed data authors have been compelled to restrict themselves to the region between 35N and 65N, and 20E and 150E, in which only it is possible to compare estimations concerning to observed and simulated climates). Conclusions Reliable estimates of the air temperature time series, averaged over large regions, may be calculated from available observation data only for the period starting in the 20th century after 1930’s. It is necessary to keep stations with long period of observation. Spatial descriptive statistics are found to be useful for the more detailed specification of the regional climate’s state and changes. It is expedient, in particular, to include the Climate Anomaly Index (CAI) and Climate Change Index (CCI) in the set of indicators of the climate variability and changes. The climate variability was observed to increase (especially in last 50-years) over Eurasia within 40N-60N, in eastern US regions, and Japan. Although the magnitudes of these changes are frequently rather small, but they are of the same sign for different data sets and estimation models. Climate change index CCI definitely increased in recent decades. The stability of results confirms the assumption of increase in climate variability and in the extremes frequency. We should investigate, whether it is connected only with the climate warming, or other causes exist as well. Acknowledgments. This work was supported by the Russian Foundation for Basic Research, project nos. 00-05-64873 and 02-05-64982.
Meteorology and Hydrology
№4, 2004
REFERENCES 1. WMO, 1992: World Climate Program1992—2001. Third WMO Long-Range Plan, part II, vol. II, WMO, no. 762. 2. Gruza, G.V. and Rankova, E.Ya., 1980: Structure and variability of the observed climate. Surface air temperature. Northern Hemisphere. – Leningrad: Gidrometeoizdat, 72 p. (in Russian). 3. Gruza, G.V. and Rankova, E.Ya., 1989: Monitoring and Probabilistic Forecast of ShortPeriod Climate Oscillations, 60th Anniversary of the Center for Hydrometeorological Forecasts, Leningrad: Gidrometeoizdat, pp. 148-170 (in Russian). 4. Gruza, G.V. and Rankova, E.Ya., 1999: Estimation of the Climate Response to Changes in Greenhouse Gas Concentration from Surface Air Temperature Observations over Russia, Izv. Akad. Nauk, Fiz. Atmos. Okeana, vol. 35, no. 6, pp. 742–749 (in Russian) [Izv., Atmos. Ocean. Phys. (Engl. Transl.), vol. 35, no. 6, pp. 671-678]. 5. Gruza, G.V. and Rankova, E.Ya., 2003: Variations and changes in climate over the Russian territory. Izv. Akad. Nauk, Fiz. Atmos. Okeana, vol. 39, no. 2, pp. 166–185 (in Russian) [Izv., Atmos. Ocean. Phys. (Engl. Transl.), vol. 39, no. 2, pp. 166-185]. 6. Izrael, Yu.A., 1974: Global System of Observations. Prediction and Assessment of Changes in the State of Environment. Principles of Monitoring, Meteorol. Hydrol., no. 7, pp. 3-8 (in Russian). 7. Review of Environmental Pollution in Russia for 1999. St. Petersburg: Gidrometeoizdat, 1999 (in Russian). 8. Rankova, E.Ya. and Gruza, G.V., 1998: Indicators of Climate Change for Russia, Meteorol. Hydrol., 1998, no. 1 (in Russian). 9. IPCC, 2001: Climate Change 2001. The Scientific Basis. Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. Van der Linden and D.Xiaosu, Eds., Cambridge University Press, 944 pp. 10. Jones, P.D., New, M., Parker, D.E., Martin, S., and Rigor, I.G., 1999: Surface Air Temperature and Its Changes over the Past 150 Years, Rev. Geophys., vol. 37, pp. 173-199. 11. Jones, P.D., 1994: «Hemispheric surface air temperature variations: a reanalysis and an update to 1993». Journal of Climate, 7, p. 1794-1802. 12. Folland, C. K., and D.E. Parker, 1995: «Correction of instrumental biases in historical sea surface temperature data». Quarterly Journal of the Royal Meteorological Society, 121: 319367. 13. Gruza, G., Rankova, E., Razuvaev, V., and Bulygina, O., 1999: «Indicators of Climate Change for the Russian Federation». Climatic Change, 42, 219-242.
№4, 2004
DETECTION OF CHANGES IN CLIMATE STATE, CLIMATE VARIABILITY AND CLIMATE EXTREMITY Gruza G., Rankova E., Institute for Global Climate and Ecology (IGCE), Russia (e-mail: [email protected]) Abstract. Refined definitions of a climate, climatic variables, climate variability, climate changes, and climate monitoring are given that are applicable to the analysis of climate variability from observations in the case of a changing climate. The main features of global warming are described. New data are reported on the features of climate changes over the Globe and Russia territory in the second half of the 20th century including changes in climate anomality, climate extremity and climate variability. 1. World climate program and climate monitoring. Definition of climate. Global warming has been recognized since the 1970s of the 20th century. The amount of researches concerning climate oscillations, both natural and anthropogenic, has increased sharply since then. It is becoming evident that the monitoring of the current climate state is required. The first statement on a threat to the global climate was made by the World Meteorological Organization (WMO) in 1976. In 1979, after the First World Climate Conference, the WMO established the World Climate Program (WCP), which laid the foundations for international activity on climate [WMO, 1992]. WCP activities are aimed at improving the monitoring of the climate system. In 1992, the United Nations Framework Convention on Climate Change (UNFCCC) was established. The Intergovernmental Panel on Climate Change (IPCC) was established by the WMO/UNEP for scientific support of UNFCCC activities. Its goals include the comprehensive and objective estimation of observed and expected climate changes and anthropogenic impacts. The IPCC has published three scientific reports on climate change: IPCC-1990, IPCC-1995, and IPCC-2001 [IPCC, 2001]. They provide the basic data on the observed climate and its variations, climate models, and the degree of agreement between model results and observations. Definition of Climate. Let us refine the term climate as applied to the analysis of a climate and climatic variability from observed data in the case of a changing climate. Weather is defined as the physical state of the atmosphere at a given point of the globe at a given time. The state of the atmosphere is characterized, in particular, by air temperature, wind velocity, humidity, precipitation, solar radiation, clouds, and phenomena such as fog, hoarfrost, hail, and other weather variables (weather elements). In a narrow but widely used sense, the climate is a generalization of weather changes and is represented by a set of weather conditions in a given spatial area over a given time interval. A statistical description in terms of means, extremes, variability indices for certain parameters, and frequencies of events over a given time period is used for climate characterization. All of these descriptive statistics are called climatic variables. According to WMO recommendations, a period of three decades is used as a standard period for the estimation of climatic variables characterizing the current climate. Nowadays, this is the 1961-1990 period.
Meteorology and Hydrology
№4, 2004
In modern research, the term climate is also used to denote the global climate, which is characterized by a set of states of the global climate system during a given time interval. The global climate system consists of five basic components: the atmosphere, hydrosphere, cryosphere, land surface, and biosphere. Their interactions have a large effect on weather fluctuations over long time intervals and are responsible for climate formation and climate changes. Climate Variability. The spectrum of changes in meteorological and oceanic characteristics is continuous. As in most aperiodic processes, its density tends to infinity only for periodic components and their harmonics---the annual and diurnal components. The contribution of periodic processes to the total variance is finite and can be estimated for the annual and diurnal cycles if their amplitudes are known. Based on A.S. Monin's (the most complete) classification of oscillations of meteorological and oceanic parameters depending on their scales, the following scales are usually distinguished: o Micrometeorological variability: from fractions of a second to several minutes. o Mesometeorological variability: from several minutes to several hours. o Synoptic variability: from several hours to two--three weeks. Individual predictions and the description of basic synoptic objects characterizing weather and its changes are possible in this range of scales. The upper boundary of the range is frequently taken to be the predictability scale of individual synoptic processes, which is estimated as two--three weeks. o Climate variability: from three weeks to several decades. Variability on such a scale, taken here as the inner time scale of the climate system, characterizes internal climatic oscillations, or climate variability, or climate fluctuations. In many applications, the upper boundary of this range is reasonably taken equal to about three decades. In particular, a 30-year interval (from 1961 to 1990 at present) was adopted by the WMO to be the standard period for the estimation of climatic normals. Scales longer than several decades characterize climate changes. It is reasonable to introduce additional groups of time scales within this region. o Secular variability. o Variability of the Little Ice Age type. o Variability corresponding to glacial periods. As mentioned above, a special consideration should be given to diurnal and annual oscillations and their harmonics. The object of research in CLIVAR, a basic component of the World Climate Research Program, is climate variability corresponding to the following three time intervals: a) seasonal to annual, b) annual to decadal, c) decadal to centennial. Thus, CLIVAR does not include the variability corresponding to time intervals from three weeks (predictability limit) to three months (season), which is the traditional object of longterm weather forecasts or predictions of short-term climate oscillations. In our view, shortperiod climate oscillations from three weeks (one month) to three months deserve a special consideration. Short-term climate oscillations play an important role in the formation of seasonal climate anomalies in the midlatitudes. In this respect, large-scale variations in the properties of westerlies, such as the index cycle and blocking processes, are of most considerable interest.
Meteorology and Hydrology
№4, 2004
Some quantitative assessments of the surface air temperature variability are presented in [Gruza&Rankova, 1980]. Climate Changes. The definition of a climate introduced above makes it possible to use, as climatic variables, any statistical characteristics of any parameters describing the state of the climate system over a given interval of time. It is only necessary to indicate the characteristic under consideration and the corresponding time interval. A climate change in a given region or over the Globe is characterized by the difference between some climatic variables for two given intervals of time. A climate change can be considered actual if it exceeds the probable error of the corresponding calculated climatic variables or can be considered statistically significant in the framework of the stochastic climate model (hypothesis) adopted if that change exceeds a prescribed level of significance. Climate changes can be caused by both natural internal and external factors and anthropogenic activities. In Article 1 of the UNFCCC, a climate change is defined in a narrower sense as "a change of climate which is attributed directly or indirectly to human activity that alters the composition of the global atmosphere and which is, in addition to natural climate variability, observed over comparable time periods." Thus, the UNFCCC distinguishes between climate changes caused by human activities altering the atmospheric composition and climate variability caused by natural factors. One should bear in mind that only total climate changes caused by natural and anthropogenic factors can be estimated from observations. The determination of the causes of climate changes and the estimation of corresponding effects are complicated problems, which can now be solved by applying climate models. Following scientific practice, we will adhere to the above-given, more general definition of climate changes, regardless of their causes. Thus, any climatic variables can be used for scientific analysis of climate variability and changes, and any base periods (including those different from 30 years) can be used for estimating deviations from averages. It is only important to explicitly introduce exact definitions and adhere to them while describing results. 2. Global warming The most prominent feature of the climate in the 20th century is global warming, which is characterized by a rise in the surface air temperature as averaged over the globe, each hemisphere, or most large continental or oceanic regions. A general view of global warming can be given by time series of globally averaged surface temperature [Jones et al., 1999] (over the oceans, the air temperature is replaced by the sea surface temperature). Figure 1 shows such series of annual mean temperature anomalies over 1856--2001 for the globe and both hemispheres (before 1856, the temperature data in the Southern Hemisphere are sparse). Global warming in the 20th century has been recognized to be an unprecedented event over the last 1000 years: the annual mean global surface air temperature increased by 0.6 ± 0.2oC over the century. However, global warming was found to be nonuniform in time. Three intervals can be distinguished: warming over 1910-1945, weak cooling in 1946-1975, and the most intense warming starting in 1976. The warmest decade occurred in the 1990s, and the warmest year was 1998. The ten warmest years took place after 1983, and eight of them occurred starting in 1990. The year 2000 was the 22nd in the continuous sequence of years with a global mean temperature exceeding the 1961-1990 normal.
Meteorology and Hydrology
№4, 2004
Northern Hemisphere
Southern Hemisphere
Global
Figure 1. Time series of annual surface air temperature anomalies averaged over the Northern Hemisphere (a), Southern Hemisphere (b) and the Globe (c), 1856-2001. The datasets of the East England University (CRU) are used. The anomalies were calculated as deviations from the 1961-1990 means. It is necessary to note, that changes in the global air temperature in individual months (here are not presented) are substantially different from changes in the annual means.
Meteorology and Hydrology
№4, 2004
3. Climate monitoring over Russia: goals, tasks, and practical implementation Allowance for climate changes and for large climate anomalies is important in decision making concerned with power industry, industry, and nature management. Recognized by the world community, this importance has led to the need for a system of continuous observations of the climate (climate monitoring). The system must include the collection and generalization of climate data, the permanent estimation of current climate anomalies and climate changes, and the dissemination of climatic information among governing bodies and the public. In Russia, a system of climate monitoring is being elaborated under the guidance of Academician Yu.A. Izrael at the Institute of Global Climate and Ecology (IGCE), the Russian Federal Service for Hydrometeorology and Environmental Monitoring (Rosgidromet) and the Russian Academy of Sciences. The concept of environmental monitoring was developed by Izrael in 1974 in a work devoted to the foundations of monitoring [Izrael, 1974], where the term “environmental monitoring” first appeared in the Russian literature. This term meant observations or measurements with the aim of control or decision making. In [Gruza&Rankova, 1989] the basic goals of climate monitoring are stated as follows: - regular observations of the climate system state, including the collection and generalization of climatic data and the determination of the characteristics of the current earth's climate; - the probability estimation of the degree of anomaly of the climate system state; - the detection of natural and anthropogenic causes of observed anomalies. The goals of monitoring also include the estimation of the scales of probable climate changes and oscillations in the future. Regular climate monitoring over the Russian Federation is performed by scientific institutions of the Roshydromet (IGCE, the All-Russia Research Institute for Hydrometeorological Information--World Data Center (VNIIGMI--MTsD), the Hydrometeorological Center of Russia, the Main Geophysical Observatory, the Arctic and Antarctic Research Institute, the State Hydrological Institute, and the St. Petersburg Division of the State Institute of Oceanography). This makes it possible to estimate observed climate anomalies in quasi-real time. Climate monitoring is based on operational observation data (received through communication channels) and includes the preparation of monthly, seasonal, and yearly bulletins. An analysis of regional climates is based on spatially averaged anomalies of meteorological variables (deviations from the values averaged over the base period 1961-1990). They are calculated by averaging point data (station observations) over equal-area latitude--longitude boxes followed by averaging over a region. The domain of analysis (globe, hemisphere, or a region) is divided into equal-area boxes of given size, and the arithmetic mean of the station values of the variable under analysis is calculated inside each box. Next, the resulting values are averaged with weights proportional to the area of the intersection of boxes with the territory of the region. Seasonal means of anomalies of meteorological variables are calculated by averaging monthly anomalies at stations for winter (December--February), spring (March--May), summer (June-August), and fall (September--November) when data for at least two months in a season are
Meteorology and Hydrology
№4, 2004
available. Annual means are calculated for a "seasonal" year (December--November) by averaging seasonal anomalies when data for all four seasons are available. Monthly bulletins "Climate Monitoring Data" have been issued since 1984. The ideas and techniques used for preparing data for monthly climate monitoring underlay the technique for preparing yearly climate monitoring bulletins "Climate Changes over Russia," which have been issued since 1997 and include seasonal and annual anomalies of air temperature and precipitation and some climatic indices. Since 1999, bulletins "Climate Changes over Russia" have been available on the web-site http://climate.mecom.ru . Materials of the annual and seasonal generalizations "Climate Features and Climate Anomalies over the Russian Federation Territory" have been included in yearly national reports on the state of the environment in Russia since 1999 [Review, 1999]. Meteorological Data for Climate Monitoring. A climate-monitoring database has been compiled at the IGCE. It includes time series of basic climatic parameters over the available observation period, reanalyses, and climate model outputs based on data from the world's leading meteorological centers. Current data used for extending the existing time series and for implementing climate monitoring are formed from operational observations received through communication channels and are included in the database in quasi-real time. The observed data in the climate-monitoring database are contained in two data sets: station data (compiled and supported by the IGCE) and data at the nodes of a uniform grid (produced by the UK Meteorological Office and supplemented via the Internet). The set of station data contains monthly mean air temperatures and monthly precipitation totals measured at stations of the global meteorological observation network starting in 1886. All stages of compiling the data sets involved data verification and control (including statistical control with the use of temporal and spatial coupling between observations). The statistical homogeneity of the series was ensured by introducing corrections that compensate for changes in instruments, observational techniques, station sites, etc. Stations located at sites with a highly increasing population were not included in the data set. The set of gridded data is known as the combined data set of temperature anomalies over land/sea in five-degree boxes (Jones et al., 1999) and covers the period of time starting in 1856. In the IPCC report [9], this data set was qualified to be the most compatible (in terms of observation homogeneity and quality) with the requirements imposed on data used for empirical climate studies and climate change detection. The data over land were based on the basic data set of United Kingdom station reports [Jones, 1994] and were calculated by averaging the station reports within each grid box. The number of grid boxes was 2592 (1296 in each hemisphere), and the number of all stations used was greater than 2500, but their number after 1991 reduced to 800-1000. The data over sea were obtained by averaging the sea surface temperatures (over grid boxes and calendar months of each year) contained in the COADS data set [Folland & Parker, 1995]. In operational climate monitoring, climate changes are estimated by using the data set, which includes 1383 stations where the time series of air temperature and atmospheric precipitation start no later than in 1951. The longest series in the data set contain data beginning in 1986 (although data over earlier years are available for individual stations). The dataset includes data from 1383 stations (global set). The former USSR territory is represented there by 455 stations, of which 315 are located in the Russian Federation.
Meteorology and Hydrology
№4, 2004
Figure 2 reflects the history of developing these networks of stations. Continuous observations during the entire century are available only at 156 stations in the former Soviet Union. Most of these stations are located in European Russia. Most of the time series of observations at stations of the Asian part of country begin in the mid-1930s. By 1936, the number of operating stations (among the 455 chosen) had been 338. All of the 455 stations operated from 1952 to the late 1980s. Some of the stations were closed in the last decade, others appeared outside of Russia. A thorough inspection and verification of all available observations over the last 10--15 years is now being conducted in IGCE in order to maximally reduce a network destruction observed in the last years (first of all, in applied to the 1383 stations chosen).
Global (max=1383) f. USSR (max=455)
Figure 2. Variations of the number of meteorological stations over the Globe and former USSR as presented in the IGCE climatic datasets, 1886-2000. The database is supplemented with current data from CLIMAT records received through communication channels at the Roshydromet Main Computer Center and at the All-Russia Research Institute of Hydrometeorological Information – World Data Centre (system for the storage of data in archive format). Data quality control is performed by comparing data from the two sources and by comparing them with synoptic observations. Data verification and quality control are based on special procedures for checking the consistency of observations in time and space, on data published in bulletins etc. In conclusion, it should be noted once again that a stable network of stations had been formed only by the 1940s. This network has been degraded severely during the last decade. The extension of the network and the elimination of various data inhomogeneities (data homogenization) have become high-priority problems to be solved in the operating climate monitoring system over Russia. Indicators of a Changing Climate and Dynamic Normals. Statistical climate characteristics are widely different in different regions and over different time intervals. The table lists basic
Meteorology and Hydrology
№4, 2004
statistical characteristics useful for describing climate states and changes in a certain region for two time intervals: Temporal and spatial averages characterize climatic variability in time and space, respectively. In [Rankova & Gruza, 1998; Gruza et al., 1999] a number of characteristics calculated from monthly mean air temperatures and monthly precipitation measured at stations within a region are used for quantitative estimation of regional climate changes. Temporal statistics Climate state Period-1: E1 E(E1)=EE1 S(E1)=SE1 Q(E1)=CAI1
Spatial statistics E =Mean S = std Q = RMS
Climate change DE=E1-E2 E(dE)=EdE S(dE)=SdE Q(dE)=CCI
Period-2: E2 E(E2)=EE2 S(E2)=SE2 Q(E2)=CAI2
In particular, useful indicators of climatic variability and regional climate changes are the climate anomaly index (CAI) and the climate change index (CCI). The climate anomaly index is defined as the Euclidean distance between the point describing the current climate state and the point representing the temporally averaged climate (normal):
CAI
y
=
2 1 n z yi − z i n ∑ i =1 σ i
(1)
The greater the CAI, the longer the distance between the point representing an instantaneous climate state and the "center" of climate state points. The climate change index is defined as the Euclidean distance between the two points describing the climate states at times y1 and y2:
CCI =
2 z −z n 1 y1i y2 i n ∑ σi i =1
(2 )
The concepts of CAI and CCI are illustrated in the last line of the table, which helps understand the meaning of these characteristics. Another fairly informative characteristic is the climate extreme index (CEI), which is defined for a particular meteorological parameter at a particular time (period) as the fraction of the area of a region characterized by extreme current anomalies of that parameter (e.g., these are anomalies lying within the10%-intervals in both tails of the climatic distribution of the parameter). Changes in climate extremeness and anomaly and in the intensity of climate changes were assessed from surface air temperatures over 1951-2000 for the entire Russia and some of its regions. It was found that the CAI weakly correlates with the CCI, but they both (and the CEI) have increased during several last decades (see more detailed in [Gruza&Rankova, 2003]). This fact is shown in part in Figure 3, which contains a schematic diagram illustrating specifications of CAI and CCI indices in 2-dimensional space: regionally averaged annual temperature over Northern Hemisphere (Y-axis) versus that of Southern Hemisphere (X-axis).
Meteorology and Hydrology
№4, 2004
Also, it is evident that the temperature changes over the last three--four decades were much greater. 0.8
VT, Northern Hemisphere
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -0.8
-0.6
-0.4
-0.2 0.0 0.2 VT, Southern Hemisphere
0.4
0.6
0.8
Figure 3. A schematic diagram illustrating specifications of CAI and CCI indices in 2-dimensional space: regionally averaged annual temperature over Northern Hemisphere (Y-axis) versus that of Southern Hemisphere (X-axis). Scatter diagram is based on CRU-datasets for the period 1956-2001 (correspond to time series from Figure 1). Correlation coefficient is 0.72. Further it is interesting to evaluate the extent to which the observed increase in climate anomaly, extremeness, and intensity of change can be attributed to climate warming. It turns out that the traditional set of climatic variables usually included in climate handbooks is insufficient for the time intervals under study in the case of a changing climate. As a rule, when one chooses climatic variables or a time interval for their estimation, the climate is assumed to be steady, which disagrees with the view of the current climate. In particular, it is far from reasonable to use the standard 30-year interval for calculating climatic normals in all cases. On the whole, it is clear that methods and algorithms designed for nonstationary random processes and their realizations would be useful in the analysis of climatic time series. Recall that, according to the definition of climatic variables given above, they include moving averages (e.g., decadal means), filtration results containing low-frequency components of processes, and trends identified in accordance with a certain hypothesis. At every particular instant of time, all of these quantities can be regarded as dynamic climatic normals, which could give (in a sense) a better characterization of the current climate than that provided by standard normals. Dynamic normals will be analyzed elsewhere. This work is restricted to linear trends and smoothed moving averages.
Meteorology and Hydrology
№4, 2004
4. Estimates of climate changes over Russia in the second half of the 2oth century. Surface Air Temperature. Figure 4 display time series of annual air temperature anomalies regionally averaged over the Russia territory. At the left the anomaly scale on the vertical axis corresponds to deviations from the mean value over 1961-1990 base period, while at the right anomalies are shown relative to a 20-year interval in the late 19th century. It is evident that the climate in Russia in the 20th century differed from that in the 19th century. It is useful to note, the year-to-year air temperature oscillations in an individual region (European Russia) were much greater than in the Russia as a whole.
a) а) 1.0
1.5
b)б)
2.0
1.0 1.5
0.5
0.5
0.0
1.0
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
0.0
-0.5
-0.5
-1.0
-1.0
1880 1900 1920 1940 1960 1980 2000
1880 1900 1920 1940 1960 1980 2000
Figure 4. Annual air temperature anomalies averaged over the Russian Federation territory. The columns show deviations from the 1961-1990 means (at the left) and from the late preindustrial period means (at the right, 1886-1905). The curves show the 11-year moving average. Warming per century (1901-2000) over Russia averaged to 0.09oC/10 years. The maximum of warming in Russia was observed in 1995 (1.9oC deviation from the normal). In the second half of the 20th century (1951-2000), the general tendency of change in the annual mean air temperature over Russia was also characterized by a positive trend. The largest trend occurred in Transbaikalia and the Baikal region (0.35oC/10 years), in the Amur and Maritime regions, and central Siberia. Large positive temperature anomalies have been observed in these regions for the last 11-12 years. For Russia as a whole, warming was more pronounced in winter and spring (with trends of 0.47 and 0.29oC/10 years, respectively). An analysis of the spatial patterns of the surface air temperature trend coefficients over 19512000 for the year as a whole and the warm and cold half-years (Figure 5) shows that the climate changes are highly inhomogeneous over the territory of Russia. It is particularly important that, in the warm period, the temperature increase is much weaker and areas of warming alternate with areas of noticeable cooling, including many grain-producing areas.
Meteorology and Hydrology
№4, 2004
oC/10 yrs
Figure 5. Linear trend coefficients pattern for the annual surface air temperature over the territory of Russia (January-December). The estimates were obtained from the station observations over 1951-2000 and are given in oC/10 years. Atmospheric Precipitation. Atmospheric precipitation is extremely important for various aspects of human activities (agriculture, power industry, transport, hazardous events associated with floods and droughts, etc.) and for the climate system (clouds, latent heat fluxes, fresh water inflow to the ocean, accumulation/destruction of ice sheets and mountain glaciers, etc.). However, the climatology of precipitation has been investigated much more poorly than the climatology of temperature. Precipitation over the ocean has been poorly studied. О тенденциях изменения режима осадков на территории России можно судить по Рис.6, где приведено пространственное распределение оценок трендов осадков во второй половине столетия для года в целом The tendencies of changes in precipitation regime over the territory of Russia can be seen from Fig. 6, where spatial patterns of the precipitation totals trends in the second half of the 20th century are presented for the year as a whole. It can be seen that the annual and seasonal precipitation totals for the entire Russia and its eastern regions tend to decrease in the last 50 years. The most pronounced decrease in precipitation occurs in the northeastern region of the country. A weak tendency of precipitation to increase is observed over European Russia. For comparison, we recall that climate models predict an increase in globally averaged precipitation in response to an increased CO2 concentration. The winter precipitation is expected to increase in the high latitudes and, according to most models, in the midlatitudes. For most latitudes, models predict a mainly moderate increase in precipitation under doubled CO2, accompanied by a significant growth of the frequency and strength of heavy precipitation, especially in the tropics and the midlatitudes of the Northern Hemisphere. The
Meteorology and Hydrology
№4, 2004
expected rise of temperature contrasts between the continents and the ocean can intensify monsoons. In particular, increased precipitation is expected in the eastern Asian monsoon system. 0
30
60
90
120
150
180
80
80
60
60
40
40
20
20
0
0 0
%/10 %/10лет yrs
30
-20% -16% -12%
60
-8%
90
-4%
-0%
120
+0%
+4%
150
180
+8% +12% +16% +20%
Figure 6. The same as Fig. 5 but for the annual precipitation totals (%/10 years). Percents are calculated relative to 1961-1990 normals. Thus, the empirically estimated trends presented here poorly agree with model computations. We conclude that a further careful comparative analysis of simulated and empirical estimates is required and that a tendency of the annual and seasonal precipitation values to somewhat decrease in eastern regions of Russia over the last 50 years can be inferred from observations. 5. Assessments of changes in climate variability over Northern Hemisphere continents in the 20th century. For evaluating climatic variability, characteristics (measures) of variability and calculation methods for them were proposed. The characteristics of climatic variability were estimated and analyzed from surface air temperatures over the Northern Hemisphere. Root mean squared (or mean absolute) deviations from the average are used as a measure of variability (scatter). For nonstationary processes, total variability is assumed to split into variability attributed to a trend of the mean and residual variability associated with deviations from the trend. The latter is estimated as the RMS or mean absolute deviation from the trend. Assuming that the deviations from the trend are also subject to slow variations, the corresponding trend of variability is reasonably defined as the trend of absolute or squared deviations from the trends of means. It is this trend of absolute deviations from the trend of means that considered here to be the trend of variability. A comparison was performed for two data sets of monthly mean air temperatures (on stations and in 5*5 boxes) taken from the database of Northern Hemisphere climatic data described
Meteorology and Hydrology
№4, 2004
above. The variability trend was estimated in each point for two periods (1901-1998 and 1951-1998) and three seasons: the calendar year and the warm and cold half-years. Stations with sufficiently complete series (with the observation gaps not exceeding 10% of the years) were only considered here. Four models were used to estimate the trend of the mean air temperature at each station: “Y” - regression on time (usual linear trend), “C” - regression on the mean carbon dioxide concentration, “G” regression on the mean global temperature, and “M” - filtered series involving oscillations with periods more than 10 years. For all four models, the trend of variability was estimated as a linear trend of absolute deviations from the trend of means. Thus, a set of linear trends of variability (AY, AC, AG, and AM) was obtained for each of points. The results were presented on 24 patterns over the Northern Hemisphere continents (four models * two periods * 3 time scales). An example of this comparison is reproduced in Figure 7, containing some fragment from AC-results. The conclusions suggested by their analysis are mainly the following. 80
(а)
60
40
20 80
-120
-60
-120
-60
0
60
120
(b)
60
40
20
-180
0 o oC/100 C/100 лет yrs
60 -1.0 -0.5 -0.2
0
120 0.2 0.5
180 1.0
Figure 7. The patterns of linear trends in the climate variability indicators (from model AC). The variability indicators are defined here as absolute deviations (in local points) of annual temperature from its regression on CO2-concentration. Estimates are based on station observed data (upper pattern) and gridded data in 5*5 boxes (lower pattern) for the period 1951-1998. The results for AY, AC, and AG are similar. The variability over the entire century (19011998) was found to increase in the cold season over nearly the entire Eurasia and in central US regions. In the warm season, variability increased only in central Eurasia and southern US regions. The variability in Europe, northern US regions, and Japan decreased substantially. At the same time, the trends over the second half of the century (1951-1998) suggest increased variability over Eurasia within 40-60N, in Japan, eastern US regions, and Canada. A considerable decline in variability was observed manly in the cold season over individual regions of Russia and in eastern and western regions of Canada. The variability estimates AM look somewhat differently: an increase in variability in all seasons occurred only over eastern
Meteorology and Hydrology
№4, 2004
US regions. In the cold season, the variability over most of the territory declined. A spatially inhomogeneous distribution of variability estimates was observed over Europe. Thus, the characteristics of the variability of annual air temperature anomalies varied on scales of 10-100 years. The climate variability was observed to enhance (especially pronouncedly in 1951-1998) over Eurasia within 40-60N, in eastern US regions, and Japan. Although the magnitudes of these changes are frequently rather small, but they are of the same sign for different data sets and estimation methods. The stability of the results confirms that the climate variability and the frequency of extremes increased. 6. Assessment of the climatic response of surface air temperature to changes in carbon dioxide and aerosol atmospheric concentrations The estimates of climatic response of surface air temperature to changes in the carbon dioxide and atmospheric aerosol concentrations were obtained from observations and model results (produced by the transitive global ocean--atmosphere climate model HadCM2 at the Hadley Center of the UK Meteorological Office). A detailed analysis of estimates derived from observed and simulated data and their comparison are presented and illustrated in [Gruza & Rankova, 1999]. We note here only the basic result: Under the CO2 concentrations and the scales of their changes considered here (up to 50 ppmv) we can adopt that a local change in air temperature is a linear function of the change in CO2 concentration (rather than being a logarithmic function accepted in the general case for the global surface air temperature). The spatial patterns and the magnitudes of the simulated and observed climate responses are widely different (when analyzing the observed data authors have been compelled to restrict themselves to the region between 35N and 65N, and 20E and 150E, in which only it is possible to compare estimations concerning to observed and simulated climates). Conclusions Reliable estimates of the air temperature time series, averaged over large regions, may be calculated from available observation data only for the period starting in the 20th century after 1930’s. It is necessary to keep stations with long period of observation. Spatial descriptive statistics are found to be useful for the more detailed specification of the regional climate’s state and changes. It is expedient, in particular, to include the Climate Anomaly Index (CAI) and Climate Change Index (CCI) in the set of indicators of the climate variability and changes. The climate variability was observed to increase (especially in last 50-years) over Eurasia within 40N-60N, in eastern US regions, and Japan. Although the magnitudes of these changes are frequently rather small, but they are of the same sign for different data sets and estimation models. Climate change index CCI definitely increased in recent decades. The stability of results confirms the assumption of increase in climate variability and in the extremes frequency. We should investigate, whether it is connected only with the climate warming, or other causes exist as well. Acknowledgments. This work was supported by the Russian Foundation for Basic Research, project nos. 00-05-64873 and 02-05-64982.
Meteorology and Hydrology
№4, 2004
REFERENCES 1. WMO, 1992: World Climate Program1992—2001. Third WMO Long-Range Plan, part II, vol. II, WMO, no. 762. 2. Gruza, G.V. and Rankova, E.Ya., 1980: Structure and variability of the observed climate. Surface air temperature. Northern Hemisphere. – Leningrad: Gidrometeoizdat, 72 p. (in Russian). 3. Gruza, G.V. and Rankova, E.Ya., 1989: Monitoring and Probabilistic Forecast of ShortPeriod Climate Oscillations, 60th Anniversary of the Center for Hydrometeorological Forecasts, Leningrad: Gidrometeoizdat, pp. 148-170 (in Russian). 4. Gruza, G.V. and Rankova, E.Ya., 1999: Estimation of the Climate Response to Changes in Greenhouse Gas Concentration from Surface Air Temperature Observations over Russia, Izv. Akad. Nauk, Fiz. Atmos. Okeana, vol. 35, no. 6, pp. 742–749 (in Russian) [Izv., Atmos. Ocean. Phys. (Engl. Transl.), vol. 35, no. 6, pp. 671-678]. 5. Gruza, G.V. and Rankova, E.Ya., 2003: Variations and changes in climate over the Russian territory. Izv. Akad. Nauk, Fiz. Atmos. Okeana, vol. 39, no. 2, pp. 166–185 (in Russian) [Izv., Atmos. Ocean. Phys. (Engl. Transl.), vol. 39, no. 2, pp. 166-185]. 6. Izrael, Yu.A., 1974: Global System of Observations. Prediction and Assessment of Changes in the State of Environment. Principles of Monitoring, Meteorol. Hydrol., no. 7, pp. 3-8 (in Russian). 7. Review of Environmental Pollution in Russia for 1999. St. Petersburg: Gidrometeoizdat, 1999 (in Russian). 8. Rankova, E.Ya. and Gruza, G.V., 1998: Indicators of Climate Change for Russia, Meteorol. Hydrol., 1998, no. 1 (in Russian). 9. IPCC, 2001: Climate Change 2001. The Scientific Basis. Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. Van der Linden and D.Xiaosu, Eds., Cambridge University Press, 944 pp. 10. Jones, P.D., New, M., Parker, D.E., Martin, S., and Rigor, I.G., 1999: Surface Air Temperature and Its Changes over the Past 150 Years, Rev. Geophys., vol. 37, pp. 173-199. 11. Jones, P.D., 1994: «Hemispheric surface air temperature variations: a reanalysis and an update to 1993». Journal of Climate, 7, p. 1794-1802. 12. Folland, C. K., and D.E. Parker, 1995: «Correction of instrumental biases in historical sea surface temperature data». Quarterly Journal of the Royal Meteorological Society, 121: 319367. 13. Gruza, G., Rankova, E., Razuvaev, V., and Bulygina, O., 1999: «Indicators of Climate Change for the Russian Federation». Climatic Change, 42, 219-242.
