The Possible Impact of Future Climate Changes on the Occurrence of Droughts

For predicting the effect of droughts on agriculture in a given region, one needs to have a completely reliable and detailed climatic forecast for that

Figure 34.3 Annual variability in wheat yields in some European countries; bold solid lines are the trends.

region and a completely adequate yield model. If these two requirements are satisfied, the technique of agroclimatic analysis can be developed as follows.

At the first stage of the analysis, an available model for the prediction of climate changes is used to carry out the calculations of meteorological and soil dynamics. The second stage of the analysis involves using a crop productivity model with regard to the agrotechnology used. The historical data can be used to give reliable and useful information about the agricultural production losses caused primarily by unfavorable climatic condi-

Figure 34.4 Annual variability in wheat yields in some Near East countries; bold solid lines are the trends.

tions. Such an analysis would enable one to see the characteristic tendencies in changes under global warming, which began to be appreciably manifest in regional climate changes. The final result of these two stages should be to provide an estimate of the crop yield and the value of permissible losses due to the unfavorable environmental conditions.

As far as the surface air temperature is concerned, there is a measure of agreement between the results obtained with the use of different GCMs

Figure 34.5 Annual variability in wheat yields in some countries of Southern Hemisphere; bold solid lines are the trends.

and the actual data, but when it comes to atmospheric precipitation the agreement is much less satisfactory. This limits the development of complete techniques for analyzing the agricultural consequences of expected climate changes in the manner we have just described. Therefore, any estimates of crop productivity changes that are carried out with the use of different model scenarios will have a rather low accuracy because of the limited reliability of the forecast used.

Trend Analysis

Instead of using climate model predictions, another method of climate change prediction is the empirical extrapolation of trends in modern climatic parameters to the forthcoming decade. In comparison with the methods based on climate models, there is no reliable physical basis for such forecasts. In this case one is assuming that the current global warming masks any natural climatic variability and thus completely controls the main processes of future climate formation. Because of this assumption, such forecasts only have a weak scientific basis, and results obtained with this technique should only be used for the very near future. We would not expect such forecasts to provide reliable results, for example, on the daily dynamics of the meteorological parameters in the vegetation growing season. For this reason the results of calculations with detailed crop productivity models using such a climate change scenario cannot be considered to be reliable. Using parameterized techniques for agrometeorological assessments based on integrated indices would be more justified for this purpose.

Paleoclimatic Reconstruction

The third method of future climate forecasting is paleoclimatic reconstruction, which has been used mainly by Russian researchers. The essence of this method is that the characteristic regularities of the earth's climatic regime in the last warm geological epoch can be assumed to apply to future climatic conditions. Such analogs, reconstructions of temperature and precipitation, are used in the Holocene climatic optimum (about 7000 years ago), in the Riss-Wuerm (Eemian) Interglacial period (about 125,000 years ago), and in the Pliocene Optimum (3-4 million years ago). It is known that the globally averaged surface air temperature in these epochs was about 1.2°C higher (Holocene optimum), about 2°C higher (Riss-Wuerm Interglacial), and 3.5°C higher (Pliocene optimum) than in the middle of the last century (Borzenkova et al., 1987; Budyko et al., 1994). According to modern forecasts based on climate models, these values of the mean surface air temperature can be expected to be reached in 2010, 2030, and 2050, respectively (Houghton et al., 2001). If the paleoclimatic analogs are to be used in agroclimatic assessments, it is necessary to appreciate that they cannot supply the high accuracy of spatial and temporal resolution. At the best, for temperature, paleoreconstructions can provide only semiannual temperature resolution, and for atmospheric precipitation we can only use them to estimate the changes in its mean annual values. Therefore, the agroclimatological techniques using the complex dynamic crop productivity models, which require the day-to-day information for the input meteorological parameters, cannot be directly based on paleoanalog scenarios.

The difficulties involved in all these three methods of climate modeling lead us to the conclusion that, at the present time, the technique of agro-climatological assessments should be to use established submodels based on empirical agrometeorological indices. This is concerned with the assessments of their mean values and in particular the estimation of the abnormal agroclimatic phenomena with which droughts are associated. A major part of such empirical techniques should be the analysis of the actual information about the variation of crop productivity and the relevant meteorological factors, principally the surface air temperature and the precipitation.

Empirical Techniques

With the purpose of developing a forecasting technique for predicting the national wheat production for the leading U.S. states using the 11-year smoothed mean anomaly, nii, the correlation graphs between nii and the Sii index and between nil and the HTC11 index have been analyzed (figure 34.6).

The calculations have shown that in each specific case both the 11-year smoothed means, S11, and HTC11, can be used as a predictor of this empirical forecasting technique; the choice between them should be determined by the value of the associated correlation coefficient. For wheat production in Colorado State shown in figure 34.6a, the correlation coefficient was .69. These graphs can be used to calculate the changes in the number of agriculturally abnormal years using the forecast of the regional climate changes in any future decade, from which it is possible to obtain the estimate for the change in the S11 index. The corresponding estimated change in nii can then be determined from the straight line in figure 34.6a. For example, if in the state of Colorado the S11 index becomes equal to -0.1, it corresponds to a change in the decade-mean anomaly of the relative productivity from 0 to -0.12. Using historical data on n values in this area (figure 34.1), it can be concluded that the probability of drought (that is, when the wheat production loss will exceed the standard deviation of the n-indicator, which is 23% for Colorado) can be more than double in the decade. In the case of any other area for which detailed information is not available, we could use the generalized graphs by combining the empirical data of the climatically analogous regions for forecasting purposes. An example of such a correlation graph is shown in figure 34.6b, which generalizes the empirical data of the three selected U.S. states.

A-Indicator

We also find it convenient to introduce an additional parameter, A-indica-tor, which is the standard deviation of the relative crop yield. We determined A-indicator for wheat producing regions of North America and the former USSR (Menzhulin et al., 1987; Gleik et al., 1990; Menzhulin, 1992). It is commonly assumed that up to the middle of the 1980s global climate changes were insignificant; therefore, the estimates obtained for A-indicator using the data on crop productivity for the previous years can be assumed to correspond to the anthropogenically undisturbed climate. For this analysis the data on annual wheat yields for separate small areas of the two regions for the period 1945-82 have been used.

In analyzing the spatial distribution of the A-parameter for wheat crops over the grain zone of the former USSR, it is important to note that the areas of winter wheat cultivation are mainly located in European Russia. The climate in eastern Russian territories is too severe. For the European part of the former USSR the annual variability of winter wheat production is

444 INTERNATIONAL EFFORTS AND CLIMATE CHANGE Tl11 Wheat, Colorado, USA

T|ii Wheat, (Colorado+N.Dakota+Kansas), USA

444 INTERNATIONAL EFFORTS AND CLIMATE CHANGE Tl11 Wheat, Colorado, USA

• •

• * •

• X.

•v

• • *

• •

y

• •

* jp

• • <

••

* ••

*

T

T|ii Wheat, (Colorado+N.Dakota+Kansas), USA

• •,

. • j.

1 —'

• •

it HA/

■«

ell

>

• •••1

a

J

•V \ _

* r*

P f#

i'J.

t y ■

••

»V

• •

• •,

*

• *

• •

Figure 34.6 Graphs showing relationship between 11-year means of S-aridity index S11, and 11-year means of wheat yield anomalies. R2 = 0.32 (upper graph) and 0.21 (bottom graph). Data period is from 1891 to 2000.

Figure 34.6 Graphs showing relationship between 11-year means of S-aridity index S11, and 11-year means of wheat yield anomalies. R2 = 0.32 (upper graph) and 0.21 (bottom graph). Data period is from 1891 to 2000.

very large (with A-indicator values varying between 0.20 and 0.40). Spring wheat cultivation in the former USSR is much larger than winter wheat. Except for European Russia, spring wheat crops are widely grown in the south of western Siberia and partly in East Siberia and the Far East. The A-indicator ranged from 0.20 to 0.60.

Yield variability in Northern America and Canada was studied during 1934-82. The A-indicator ranged from 0.10 to 0.35, though this range is less than in the case of the former USSR agricultural zone. For winter wheat, annual variability was more pronounced for the southwestern part of the U.S. wheat belt. The unsteady character of the rainfall regime during the plant-growth season results in significant annual variation in crop yields. For the grain areas of New Mexico and Colorado, the A-indicator reached 0.30. The frequent anomalies of wheat yields in these states are usually caused by droughts and dust storms which also cover the adjacent states. Especially large damage to the grain production in these U.S.

areas was caused by strong dust storms in the 1930s. The A-indicator values for winter wheat calculated for a long period for the Oklahoma and Texas grain areas were about 0.23. In central and northern areas of the U.S. wheat belt the stability of winter wheat yield was a little higher, with the A-indicator ranging from 0.15 to 0.20.

The absence of the winter soil freezing and the uniform annual distribution of atmospheric precipitation allows one to obtain rather stable yields of winter wheat in the forest-steppe areas of the U.S. corn belt. For agricultural areas of Ohio, Indiana, Illinois, Missouri, and Iowa, A-indicator ranged from 0.10 to 0.15, though it is also possible to note some tendency toward an increase in the value of this indicator in the western parts of this area where the climatic normal of moistening is lower. For the southern Atlantic U.S. states, the A-indicator was 0.15. In general, the A-indicator for spring wheat in North America remained between 0.10 and 0.30.

0 0

Post a comment