Введение.
Introduction. In the Republic of Kazakhstan, the agricultural sector is significantly determined by climatic conditions and their interannual variability. Especially important is the study of changes in heat and moisture availability regimes, which have a direct impact on the yield of grain crops, which are the main element of food security in a number of regions, including Northern Kazakhstan [1; 2]. In recent decades, research in the field of agro-climatology has demonstrated that even minor changes in temperature and precipitation conditions can significantly modify biological processes in agricultural crops [3; 4; 5]. However, despite the accumulated experience, questions remain in terms of a reliable assessment of the contribution of climatic parameters to crop yields, as well as disagreements in assessing the contribution of individual climate factors. Some authors point to the dominant value of temperature during the warm period, while others emphasize the importance of water balance for optimizing crop growth [6; 7].
The purpose of this study is to comprehensively analyze the impact of changes in heat and moisture availability regimes on grain crop yields in Northern Kazakhstan. The importance of the work is determined by the need to create scientifically sound recommendations for adapting the agricultural sector to changing climatic conditions and increasing the sustainability of agricultural production. The main emphasis is placed on comparing the actual data obtained as a result of instrumental measurements with the calculated yield values, which makes it possible to identify key dependencies and assess the share of influence of each climatic factor [8; 9].
Thus, the presented study not only expands the understanding of climate processes affecting agriculture but also contributes to the development of strategies aimed at mitigating the negative effects of climate change on the agricultural sector, which is especially important for regions with high risks of adverse climatic events.
Методы.
Materials and methods. Methods of hydrological and climatic calculations, methods of statistical analysis: correlation dependence, regression and variance analysis, as well as mapping were used to assess changes in the modes of heat and moisture supply of the territory.
The largest concentration of sown areas of grain crops in Kazakhstan is observed in the North Kazakhstan, Kostanay, Akmola, Pavlodar, and Karaganda regions. According to statistics from the Bureau of National Statistics of the Agency for Strategic Planning and Reforms of the Republic of Kazakhstan [10] for 2024, these regions account for 77%, or 17.8 million hectares, of the total amount of cultivated land, which is 23.1 million hectares (Figure 1,2).
Figure 1- The share of the regions of Northern and Central Kazakhstan from the total sown area of the Republic of Kazakhstan, (%)
Figure 2-The sown area of the main crops of the northern regions of Kazakhstan,
million hectares
To analyze the impact of climatic fluctuations on grain yields in Northern Kazakhstan, meteorological data from representative weather stations located in grain-growing areas for the period 1999-2024 were used. To determine the degree of correlation, the following parameters were used: measured values of the average monthly air temperature during the warm period (from May to August) as indicators of thermal and energy resources of the climate, as well as the average monthly precipitation over similar time intervals as an indicator of the humidity of the region [11]. Additionally, the effect of the moisture coefficient on the productivity of grain crops was investigated.
Результаты.
Results. Northern Kazakhstan, which is located in an arid climate and has the largest area of agricultural crops in the country, demonstrates high productivity of grain crops (Figures 3,4). Limited water resources slow down biological processes and reduce the ability of geosystems to self-repair, which negatively affects the condition of agricultural land. As a result, the natural complexes of the region have a low potential for resistance to external influences.
The region under study is classified as a zone of risky agriculture, characterized by significant interannual variability of water and heat resources. The territory is mainly represented by dry-steppe, steppe, and forest-steppe landscapes, dominated by low-humus chernozems and chestnut soils. The climate of the region is determined by harsh winters, hot, dry summers, and pronounced variability in temperature and precipitation, which is especially noticeable in the spring and summer period, when dry and arid conditions are often observed. The limited moisture resources significantly affect the fertility of the region's soils [12; 13; 14].
Figure 3- Average grain yield (c/ha) for 1999-2024 in Northern Kazakhstan by region
Figure 4- Grain yield (c/ha) on average for 1999-2024
in Northern Kazakhstan
In 2011, when there was enough moisture, grain yields reached 20.4 c/ha in the North Kazakhstan region and 18.4 c/ha in the Kostanay region. During dry periods, there was a decrease in the yield of cultivated grain, which was especially evident in 2010 in the Akmola region (5.2 c/ha), as well as in 2008 and 2012 in the Pavlodar region, where this indicator was 3.7 c/ha and 3.8 c/ha, respectively (Figure 5).
Figure 5-Fluctuations in the yield of grain crops (c/ha) in years with sufficient moisture
and dry years in 1999-2024
Based on a database containing information on grain yields and meteorological characteristics for Northern Kazakhstan, regression models were developed demonstrating the influence of key factors—air temperature and total precipitation during the warm period (May–August) (Table 1). The obtained regression equations can be used to estimate yields in a retrospective analysis in the absence of actual data.
Table 1-Regression statistical analysis used to quantify crop yields Y (c/ha), taking into account the influence of meteorological factors
Weather station | R | R2 | The normalized R-square | The standard error | The regression equation |
Ruzaevka (North Kazakhstan region) | 0,51 | 0,26 | 0,17 | 2,52 | Y =30,52-1,08Twp1+0,01Хwp2 |
Kostanay (Kostanay region) | 0,66 | 0,43 | 0,37 | 2,24 | Y =45,58-1,82Twp-0,005Хwp |
Yesil (Akmola region) | 0,80 | 0,64 | 0,60 | 1,50 | Y =27,11-1,09Twp+ 0,02Хwp |
Irtyshsk (Pavlodar region) | 0,71 | 0,51 | 0,46 | 2,08 | Y =33,94-1,60Twp+ 0,02Хwp |
Karaganda (Karaganda region) | 0,36 | 0,13 | 0,06 | 2,23 | Y =18,62-0,62Twp+ 0,008 Хwp |
Remarks: 1Twp - is the average monthly temperature of the warm period (V–VIII months).
2 Хwp - is the total amount of precipitation of the warm period (V−VIII months).
An analysis of the cumulative impact of meteorological factors on yields showed that the degree of their influence, expressed by the multiple correlation coefficient (R), is maximum in the Akmola region (using the example of Yesil, R=0.80) and in the Pavlodar region (using the example of Irtyshsk, R=0.71). In the regions of Kostanay, North Kazakhstan, and Karaganda, lower coefficient values are observed (Kostanay, R=0.66; Ruzaevka, R=0.51; Karaganda, R=0.36). The revealed dependencies make it possible to explain from 13% to 64% of the yield variance depending on the impact of the considered climatic indicators.
Given that the most pronounced relationship between meteorological factors and crop yields was established at the Yesil weather station in the Akmola region, the data from this point is used to assess the quality of the model. The value of the coefficient of determination (R² = 0.64) indicates that the model explains approximately 64% of the variation in the dependent variable, which is air temperature and the total amount of precipitation during the warm period. The high coefficient of multiple correlation (R=0.80) confirms a significant relationship between yield and the parameters included in the model: air temperature and the amount of precipitation during the warm period.
The statistical significance of the regression equation was estimated based on the calculation of the Fisher F-test (Table 2), the value of which was 15.7. At the same time, the tabular value of the F-criterion for a confidence probability of 0.95 and degrees of freedom v₁ = k – 1 = 1 and v₂ = n – k – 1 = 18 is 4.41. Since the calculated value significantly exceeds the tabular value (15.7 > 4.41), the regression equation is recognized as statistically significant, which confirms the reliability of the estimate obtained [15].
Table 2-Regression statistics values
Analysis of variance |
| df - number of degrees of freedom | SS - sum of squares | MS - the average value | F- the Fisher criterion |
Regression | 2 | 71.070 | 35.535 | 15.716 |
Remains | 18 | 40.699 | 2.261 | |
Total | 20 | 111.769 | | |
| Coefficients | The standard error | t- statistics | P- value |
Y- intersection | 27.11 | 7.68 | 3.53 | 0.002 |
Variable X 1 | -1.09 | 0.39 | -2.82 | 0.011 |
Variable X 2 | 0.02 | 0.01 | 3.01 | 0.007 |
The calculation of the average approximation error makes it possible to estimate the accuracy of the regression model (Table 3). Given that the standard value of the average approximation error does not exceed 8-10%, the achieved accuracy of the regression model at the level of 9% indicates its high reliability.
Table 3-Results of checking the significance of the regression equation based on the average approximation error
Years | Actual yield, (y) | Estimated yield, y(x) | The discrepancy,y-y(x) | (y-y(x))/y*100 |
1999 | 13.3 | 11.9 | 1.4 | 10.5 |
2000 | 7.9 | 10.9 | -3 | -38.6 |
2001 | 11.2 | 11.4 | -0.2 | -1.6 |
2002 | 9.1 | 11.02 | -1.9 | -21.1 |
2003 | 9.1 | 11.4 | -2.3 | -25 |
2004 | 7.1 | 7.8 | -0.7 | -9.2 |
2005 | 8.5 | 10 | -1.5 | -17.9 |
2006 | 9.6 | 10.2 | -0.6 | -5.8 |
2007 | 11.6 | 10.5 | 1.1 | 9.5 |
2008 | 7.5 | 7.9 | -0.5 | -6 |
2009 | 11.2 | 10 | 1.2 | 10.4 |
2010 | 5.2 | 6.5 | -1.3 | -25.1 |
2011 | 15.6 | 14.4 | 1.2 | 7.7 |
2012 | 7 | 6.8 | 0.2 | 3.4 |
2013 | 10.4 | 10.7 | -0.3 | -3.2 |
2014 | 11 | 8.9 | 2.1 | 19.2 |
2015 | 10.8 | 9.4 | 1.4 | 13 |
2016 | 11.6 | 12 | -0.4 | -3.8 |
2017 | 11.2 | 9.1 | 2.1 | 18.3 |
2018 | 11.7 | 11.3 | 0.4 | 3.1 |
2019 | 9.5 | 10.3 | -0.8 | -8.4 |
2020 | 11.6 | 9.0 | 2.6 | 22.4 |
2021 | 8.7 | 6.2 | 2.5 | 28.7 |
2022 | 11.6 | 8.6 | 3 | 25.9 |
2023 | 6.9 | 8.4 | -1.5 | -21.7 |
2024 | 13.1 | 11.4 | 1.7 | 13 |
А=1/n*∑ ((y-y(x))/y)*100% =9% |
The statistical significance of the regression coefficients was estimated using the Student's t-statistics. The initial hypothesis H₀ assumed that the coefficients of the model do not differ significantly from zero (a = bᵢ = 0). The obtained t-statistic values were ta = 3.5, tb₁ = -2.8, and tb₂ = 3.01, while the critical value of the t-criterion at a significance level of 5% and 18 degrees of freedom (n – k) is 2.1. Since the absolute values of the calculated indicators exceed the critical value (|3.5| > 2.1; |-2.8| > 2.1; |.01| > 2.1), the hypothesis H₀ is rejected, which indicates the statistical significance of the regression coefficients.
In climate change studies, many experts analyze the annual surface air temperature and, less often, seasonal temperature indicators (for example, winter and summer values), which is especially important in factor analysis of yields. The use of regression analysis and variance analysis for Northern Kazakhstan revealed that the greatest contribution to the yield variance is made by the average air temperature during the warm period. In this regard, the calculated yield values were compared with the data on the average temperature of the warm period, determined by the results of actual instrumental measurements at the Yesil weather station (Figure 6).
Figure 6-Comparison of grain yields in Akmola region with the average temperature of the growing season (May-August) for 1969-2024
The analysis of the data presented in Figure 6 and in the regression equations (Table 2) indicates a pronounced inverse relationship between yield and temperature conditions of the warm period, which is confirmed by the correlation coefficient R = -0.85. In particular, years with low yields are characterized by high temperatures during the warm period, accompanied by arid conditions, while years with higher yields are marked by lower temperatures, providing an optimal ratio of heat and moisture. This result is expected for the steppe regions of Northern Kazakhstan and is additionally confirmed by the analysis shown in Table 4, where there is a coincidence of periods of low yields with the highest temperature indicators of the warm period.
Table 4- Years with minimum yield and highest average temperature of warm period in Akmola region
Years | Y, c/ha | Тwp,°С |
1984 | 8.3 | 19.0 |
1987 | 7.8 | 19.0 |
1989 | 8.3 | 19.4 |
1991 | 8.1 | 19.5 |
1998 | 4.9 | 21.2 |
2004 | 7.1 | 19.6 |
2008 | 7.5 | 19.7 |
2010 | 5.2 | 20.1 |
2012 | 6.7 | 20.4 |
2021 | 6.2 | 20.3 |
Figure 7 shows the relationship between yield and average temperature of the warm period, characterized by the correlation coefficient R equal to 0.85. Analysis of the graph demonstrates that an increase in average temperature is associated with a decrease in yield, while lower temperature regimes contribute to an increase in yield.
In addition to heat and energy resources, the level of crop yield is also affected by the degree of moisture in the territory [16; 17].
Figure 7- Yield dependence on average air temperature warm period
Wetness of the territory is determined by the water cycle between the earth's surface and the atmosphere due to continuous heat and moisture exchange between the active layer of soil and the surface layer of air. According to V.S. Mezentsev's method of hydrological-climatic calculations, in case of inequality of atmospheric moisture content to the optimal value obtained by means of the water equivalent of heat and energy resources of evaporation, their difference expresses the value of moisture deficits (or excesses) [18].
To determine the water equivalent of heat energy resources (Zm), we used the formulas (1-4) proposed by I.V. Karnatsevich [19], according to which evapotranspiration is calculated as a function of the sum of average monthly positive air temperatures:
Zm = TZ/L (1)
TZ = 17.6 ∑t + 400 (2)
where: TZ—heat and energy resources of total evaporation in MJ/(m² year);
L - 2.512 MJ/(m²×mm) - specific heat of vaporization;
∑t—the sum of positive average monthly temperatures of the growing season.
The wetting deficit indicator (∆KX) is calculated as the difference between precipitation (KX) and the water equivalent of heat and energy resources (Zm).
∆KX=KX/Zm (3)
The structure of relations between heat and moisture resources determines the level of wetting (natural or anthropogenic); therefore, the indicator of territory wetting for any intra-annual interval of an average year is the ratio:
βkx=KX/Zm (4)
The boundary of the optimal ratio of heat and moisture is spatially expressed by the isoline of the unit value of the wetting coefficient, zero deficit of wetting, and soil moisture (in fractions of the smallest moisture capacity) equal to one. This boundary is the upper limit of optimal moistening for most agricultural crops [18,20].
In the framework of the study, the impact of moisture supply on the productivity of grain crops is determined through the analysis of the moisture coefficient. According to the studies of Professor V.S. Mezentsev [21], the relationship between thermal and moisture resources is determined by the level of natural and anthropogenic moistening, which determines the use of the ratio given in formula (4) as an indicator of moistening for any intra-annual period.
In this paper, the influence of natural heat and moisture availability parameters, namely moisture coefficient (βkx) and moisture deficit (∆KX), on yield by applying multivariate correlation and regression analysis is investigated (Table 6).
The multiple correlation coefficient (R = 0.77) indicates a significant relationship between yield and the two factors selected in the model, i.e., moisture coefficient and growing season moisture deficit. The statistical significance of the regression model was evaluated using Fisher's F-criterion, where the calculated F value was 12.8. At the same time, the critical value of F at the confidence level of 0.95, with degrees of freedom v₁ = k - 1 = 1 and v₂ = n - k - 1 = 18, is 4.41. The excess of the calculated value over the critical value (12.8 > 4.41) allows the regression equation to be considered statistically significant and reliable. In addition, at a 5% significance level, the observed values of Student's t-statistics for the model parameters were higher than the corresponding critical values, which confirms the significance of the regression coefficients.
Using regression models describing the relationship between grain crop yields in Akmola oblast and parameters of natural heat and moisture availability, predicted yield values were obtained (Table 6). The value of the correlation coefficient (R = 0.71) between the predicted and observed yields for the period 1999-2024 indicates high reliability of the model.
Table 6- Comparison of actual and calculated yields in the Akmola region
Years | Actual yield, Y (y), (c/ha) | Estimated yield, y(x), (c/ha) | Divergence, y-y(x), (c/ha) |
1999 | 13.3 | 11.8 | 1.5 |
2000 | 7.9 | 11.1 | -3.2 |
2001 | 11.2 | 11.4 | -0.2 |
2002 | 9.1 | 11.2 | -2.1 |
2003 | 9.1 | 11.5 | -2.4 |
2004 | 7.1 | 8.2 | -1.1 |
2005 | 8.5 | 10.3 | -1.8 |
2006 | 9.6 | 10.4 | -0.8 |
2007 | 11.6 | 10.7 | 0.9 |
2008 | 7.5 | 8.4 | -0.9 |
2009 | 11.2 | 10.3 | 0.9 |
2010 | 5.2 | 7 | -1.8 |
2011 | 15.6 | 13.9 | 1.7 |
2012 | 7 | 7.3 | -0.3 |
2013 | 10.4 | 10.9 | -0.5 |
2014 | 11 | 9.2 | 1.8 |
2015 | 10.8 | 9.7 | 1.1 |
2016 | 11.6 | 12 | -0.4 |
2017 | 11.2 | 9.5 | 1.7 |
2018 | 11.7 | 11.4 | 0.3 |
2019 | 9.5 | 10.5 | -1 |
2020 | 11.6 | 9.3 | 2.3 |
2021 | 8.7 | 6.8 | 1.9 |
2022 | 11.6 | 9 | 2.6 |
2023 | 6.9 | 8.9 | -2 |
2024 | 13.1 | 11.4 | 1.7 |
Correlation coefficient R=0.71 |
The analysis of Figure 8 shows that in periods with a low coefficient of moisture content of the territory, in the warm period, the yield tends to decrease, while in years with sufficient moisture content, high yields are observed.
Figure 8-Synchrony of grain crop yields in the Akmola region with the moisture coefficient for the growing season (May-August), calculated by the method of V.S. Mezentsev for 1999-2024.
Обсуждение.
At the same time, there are certain discrepancies in establishing the relationship between these indicators, which may be due to errors in the data of the observation series and insufficient volume of meteorological data. In the aggregate, the correlation coefficient was R = 0.71, and the coefficient of determination was- 0.51, indicating a statistically significant relationship: moisture explains 51% of the variability of yields in the area of the weather station Yesil, while the remaining 49% are due to the influence of other factors.
Заключение.
Conclusions and suggestion. The main determinants of crop productivity are the amount of precipitation and air temperature during the warm period, which for grain crops covers May-August. Aridity, as a rule, manifests itself in this time interval with a deficit of precipitation and increased temperatures, and the level of crop productivity can serve as an indicator of the climatic aridity of the territory.
In the steppe zone of Northern Kazakhstan, for the formation of optimal conditions of heat and moisture availability and increasing the productivity of agricultural production, it is necessary to increase soil moisture. For this purpose, irrigation systems and other agrotechnical reclamation measures are used, aimed at changing the soil structure and improving its physical properties in order to artificially increase the level of moisture.
