As stated in the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (
Journal of Geographical Sciences, Volume. 30, Issue 3, 439(2020)
Effect of farmland expansion on drought over the past century in Songnen Plain, Northeast China
The effects of human activities on climate change are a significant area of research in the field of global environmental change. Land use and land cover change (LUCC) has a greater effect on climate than greenhouse gases, and the effect of farmland expansion on regional drought is particularly important. From the 1910s to the 2010s, cultivated land in Songnen Plain increased by 2.67 times, the area of cultivated land increased from 4.92×10 4km 2 to 13.14×10 4km 2, and its percentage of all land increased from 25% to 70%. This provides an opportunity to study the effects of the conversion of natural grassland to farmland on climate. In this study, the drought indices in Songnen Plain were evaluated from the 1910s to the 2010s, and the effect of farmland expansion on drought was investigated using statistical methods and the Weather Research and Forecasting Model based on UK’s Climatic Research Unit data. The resulting dryness index, Palmer drought severity index, and standardized precipitation index values indicated a significant drying trend in the study area from 1981 to 2010. This trend can be attributed to increases in maximum temperature and diurnal temperature range, which increased the degree of drought. Based on statistical analysis and simulation, the maximum temperature, diurnal temperature range, and sensible heat flux increased during the growing season in Songnen Plain over the past 100 years, while the minimum temperature and latent heat flux decreased. The findings indicate that farmland expansion caused a drying trend in Songnen Plain during the study period.
1 Introduction
As stated in the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (
A historic, large-scale immigration event called “Chuang Guandong” (meaning go and settle down in Northeast China) occurred in the late Qing Dynasty in China. During this event, land reclamation occurred in northeast China, resulting in considerable LUCC. This LUCC was particularly significant in Songnen Plain, where the area of cultivated land increased almost exponentially. Thus, the effect of human activities on LUCC in Songnen Plain has great significance to coupled human-environment systems (
2 Study area
Songnen Plain is located in the central and western parts of Northeast China (
Figure 1.Figure 1
The diamond-shaped Songnen Plain is bounded to the south by the Songliao watershed, which separates the Songnen Plain from the Liaohe Plain. The northern boundary of Songnen Plain is connected to the Lesser Khingan Mountains, the western boundary is the eastern part of the Greater Khingan Mountains, and the eastern boundary is the eastern part of Changbai Mountains. The geologic structure of Songnen Plain is a sag area that is part of the Songliao fault zone; as a result, Songnen Plain is known as the wavy plain. The middle of Songnen Plain has a large number of wetlands and lakes. The soil in Songnen Plain is very fertile, with black soil and chernozem accounting for more than 60% of the soil. Songnen Plain is a major grain-producing area, with major crops including soybeans, wheat, corn, sugar beets, flax, and potato. Animal husbandry is also developed in Songnen Plain due to the high concentration of grasslands. The climate of Songnen Plain, which contains semi-humid and semi-arid zones, is temperate and monsoonal with four distinct seasons. The annual average temperature increases gradually from north to south, presenting a zonal distribution. In winter, the climate is cold and dry. The coldest month is January, when the average temperature is between -16℃ and -26℃, and total precipitation is generally 10-24 mm in January. The temperature rises quickly in spring, which is characterized by frequent southwest winds and relatively little precipitation. In summer, Songnen Plain is affected by southeast winds along with heavy rainfall and even rainstorms, with total precipitation of approximately 270-417 mm. In autumn, the intensity of solar radiation along with the sunshine time decrease, the temperature decreases sharply, and precipitation is slightly higher than in the spring (
3 Data
3.1 Land use data
The time series data of land use and land cover in Songnen Plain used in this study (1910s, 1930s, 1950s, 1970s, 1980s, 1990s, 2000s, and 2010s) are detailed by
Farmland area and fraction of farmland area in Songnen Plain from the 1910s to the 2010s
Farmland area and fraction of farmland area in Songnen Plain from the 1910s to the 2010s
|
Figure 2.Figure 2
Figure 3.Figure 3
3.2 Meteorological data
CRU data: The Climatic Research Unit (CRU) at the University of East Anglia reconstructed a complete, high-resolution, non-disruptive monthly mean surface climate factor dataset (CRU TS 3.1) by integrating several well-known databases (spatial resolution is 0.5°×0.5°) starting in 1901. CRU data have many advantages compared to existing climate data in China. First, relatively few observational data are available for the early 20th century. According to the conclusion of relevant literature research, the annual mean temperature time series which is reconstructed from CRU data are in good agreement with the observations in China, with the correlation coefficient of 0.84. At the regional scale, the variable rates of temperature of the CRU data are also in good agreement with the observations in 10 typical zones, and the overall correlation coefficient remains at about 0.8. The annual total precipitation revealed by CRU data from 1951 to 2000 coincides with the observations which come from 160 meteorological stations in China, with the correlation coefficient of 0.93. The seasonal precipitation data reconstructed by CRU are in good agreement with the actual precipitation data in eastern China, and the best consistency in autumn with the correlation coefficient of 0.93. Above all, the CRU data can always represent the main characteristics of the interdecadal variations of temperature and precipitation in China. So it is reliable to study the large-scale climate over the past 100 years. (
NNR data: The NNR data is the NCEP/DOE global atmospheric reanalysis data jointly proposed by the National Center for Environmental Prediction (NCEP) and the US Department of Energy (DOE). The starting date of NNR is 1979. This article has registered and downloaded the global average 2 m monthly average temperature data from 1982 to 2010 from http://www.esrl.noaa.gov/psd/data/gridded/tables/temperature.html. For Gaussian points, there are 192 × 94 grid points in the world.
4 Methods
4.1 Drought indices
Three drought indices were calculated in this study: drought index (K), Palmer drought severity index (PDSI), and standardized precipitation index (SPI). K was calculated using the Penman formula. PDSI is an index of relative dryness based on monthly temperature and precipitation data along with a physical model of soil water balance. PDSI is normalized and generally ranges from -6 (dry) to +6 (wet). SPI, another index used to characterize drought, has a skew normal distribution. Therefore, it is difficult to compare precipitation at different scales based on SPI. In this study, SPI was first used to calculate the probability distribution (Γ distribution function) of precipitation in a certain period of time. SPI was then converted to a normal distribution probability, and finally the cumulative precipitation frequency to divide the drought level. The calculation process of the three indices can be seen in the literature (GB, National Standard of the People’s Republic of China, GB/T, 20481-2006).
4.2 Observation minus reanalysis
It is difficult to isolate the effect of LUCC on local climate from the background of global climate change using practical observation methods. Reanalysis data only reflect the effects of large-scale circulation and greenhouse gases on climate; thus, reanalysis data cannot reveal the effects of LUCC. Since observed data contain all the radiative forcing factors, the reanalysis surface temperature data can be subtracted from the surface observations to obtain the effect of LUCC on local temperature. That is, the effect of LUCC on local temperature is given by the surface observation data minus the reanalysis surface temperature data. This observation minus reanalysis (OMR) method (
where ΔTOR represents the OMR temperature, T0 represents the measured temperature and TR represents the reanalysis of temperature.
Calculation process: The temperature OMR is the difference temperature anomaly between the CRU and the NNR data. The OMR trend value is calculated by the average variation of 10a on each grid point in order to reduce the random error, and then the hyperbolic interpolation method is used to obtain the temperature OMR. The spatial resolution of grid data is 1 km × 1 km, which is consistent with the spatial resolution of land use/cover data. The specific calculation process is referred to
4.3 Trends in meteorological elements
Trends in the time series of different meteorological elements were analyzed by linear trend estimation. For the data sequence y(x), x = 1, 2, ..., n, the original sequence was fitted by a linear function:
where the sign of “a” represents the direction of change in the data series with time, the absolute value of a represents the rate of change in the data series, and b is the y intercept.
4.4 Weather research and forecasting model
The Weather Research and Forecasting (WRF) model, a new-generation mesoscale model jointly developed by the National Center for Atmospheric Research (NCAR), was applied in the study area from the initial mesoscale weather forecast to the regional climate simulation. The WRF model is appropriate for long-term regional climate modeling. In recent years, the WRF model has been widely used to study land-atmosphere interactions in China.
The specific simulation process: The land use type parameter is set by the “land cover type-physical parameter” checklist preset by the model; the simulated area is the Songnen Plain. The center point of the simulated area was located at 125°E, 46°N. There were 94 grid points in the east-west direction and 89 in the north-south direction. The mesh size of the cell grid was 30 km × 30 km. The integration time was from January 1, 2008, to December 31, 2012. FNL reanalysis data (daily grid point data from US National Center for Environment Prediction, NCEP) reanalysis data at six-hour intervals were used as the side boundary, as shown in
Experimental design
Experimental design
|
Based on the land use and cover data for Songnen Plain, a set of control experiments and seven sets of experiments for comparison with the control were designed. The experimental program with the same experimental setup and the same physical parameters in the eight experiments excluded the land cover data. The control experiment, which was based on the land cover data from the 1910s, represented farmland before agricultural reclamation. The land cover data for the 1930s, 1950s, 1970s, 1980s, 1990s, 2000s, and 2010s represented different periods after reclamation.
5 Results and analysis
5.1 Trends in drought in Songnen Plain over the past 100 years
Three drought indices (K, PDSI, and SPI) were calculated from 1901 to 1930 and from 1981 to 2010 (
Comparison of drought indices for 1901-1930 and 1981-2010 in Songnen Plain
Comparison of drought indices for 1901-1930 and 1981-2010 in Songnen Plain
|
Figure 4.Figure 4
Figure 5.Figure 5
5.2 Factors affecting the drought indices
All the meteorological factors for 1901-1930 and 1981-2010 were subjected to variance analysis in this study. Average temperature, maximum temperature, minimum temperature, diurnal temperature range, evapotranspiration, water vapor pressure, and frost day all showed highly significant differences (0.01 probability level) between the two time periods, while cloud cover showed a significant difference (0.05 probability level). The average temperature, maximum temperature, minimum temperature, diurnal temperature range, evapotranspiration, water vapor pressure, precipitation, and precipitation days were higher from 1981 to 2010 than from 1901 to 1930, whereas the cloud cover and frost days were lower; however, the precipitation and the number of precipitation days were not significantly different between the two time periods. Thus, the change in dry rate in 1981-2010 was larger than that in 1901-1930, which may be attributed to the significant changes in meteorological factors (
Comparison of meteorological factors for 1901-1930 and 1981-2010 in Songnen Plain
Comparison of meteorological factors for 1901-1930 and 1981-2010 in Songnen Plain
|
Correlation coefficients between drought indices and all temperature elements in Songnen Plain
Correlation coefficients between drought indices and all temperature elements in Songnen Plain
|
5.3 Effects of farmland expansion on temperature in Songnen Plain over the past 100 years
5.3.1 Model simulation results
Based on the model simulation results, the average temperature in Songnen Plain increased by 0.02℃ from the 1910s to the 2010s (linear rate = 0.00℃/10a), but the increase was not significant (
Figure 6.Figure 6
Figure 7.Figure 7
5.3.2 Results of statistical analysis
The OMR values of Songnen Plain were calculated from 1980 to 2010, and the spatial distribution of OMR was obtained by spatial interpolation. Land use maps from 1980, 1990, and 2010 were overlaid with OMR spatial maps. The OMR values of farmland and grassland were then extracted, and the mean values were calculated (
OMR values for farmland and grassland (℃)
OMR values for farmland and grassland (℃)
|
6 Discussion
Up to now, there are few studies on the impact of farmland expansion on drought index, some of which involve the impact of farmland expansion on precipitation. For example, the expansion of farmland in India during 1990-2000 weakened the average precipitation intensity and reduced the total precipitation by 12.8% (
While many studies have focused on the conversion of forest to farmland (
The reflectance value of farmland and grassland in each season
The reflectance value of farmland and grassland in each season
|
Figure 8.Figure 8
Many studies have suggested that a significant increase in the minimum temperature is the main cause for climate warming.
Figure 9.Figure 9
7 Conclusions
Based on the calculated values of K, PDSI, and SPI, the entire Songnen Plain showed a significant drying trend from 1981 to 2010, and the tendency rates were 0.08, -0.82 and -0.21. In contrast, during the first 30 years of the 20th century, no significant drying trend was observed, and arid areas were only found in the northwestern part of Songnen Plain, K, PDSI and SPI tendencies were -0.003, 0.09 and 0.06, respectively. The maximum temperature, average temperature, and diurnal temperature range were significantly correlated with the three drought indices. In the last 30 years of the 20th century, the maximum temperature, average temperature, and diurnal temperature range showed significant increasing trends. WRF model simulation and statistical analysis indicated that farmland expansion in Songnen Plain over the past 100 years (1910 to 2010) caused the increases in maximum temperature and diurnal temperature range. The simulation results showed that the average temperature, the maximum temperature and the diurnal temperature range increased by 0.03, 0.13 and 0.20℃, respectively, leading to the development of drought. The findings indicate that farmland expansion has accelerated the development of aridity in Songnen Plain, resulting in the expansion of drought.
[5] D F Chen, Z Y Sun. American agriculture production and the protection of resource and environment. Ecological Economy, 9, 60-65(2001).
[8] J L Eastman, M B Coughenour, R A Pielke et al. Does grazing affect regional climate?. Journal of Hydrometeorology, 2, 243-235(2001).
[17] Climate change 2013: The physical science basis. In: Stocker T F, Qin D H, Plattner G K
[18] L Q Jiang, L J Zhang, S Y Zang et al. Comparison of approaches of spatially explicit reconstruction of cropland in the late Qing Dynasty. Journal of Geographical Sciences, 70, 625-635(2015).
[22] Q P Li, Y H Ding, W J Dong. A numerical simulation study of impacts of historical land-use change on the regional climate in China since 1700. Acta Meteorologica Sinica, 21, 9-23(2007).
[23] X X Li, X Z Zhang, L J Zhang. The varied impacts of land use/cover change on summer precipitation over eastern China. Geographical Research, 36, 1233-1244(2017).
[26] R Mahmood, R A S Pielke, K G Hubbard et al. Impacts of land use/land cover change on climate and future research priorities. Bulletin of the American Meteorological, 91, 37-46(2010).
[28] H Q Mao, Z Xiong, X D Yan et al. Modeled impact of cropland expansion on regional climate in India. Chinese Journal of Ecology, 35, 1627-1634(2016).
[37] G L Tang, Y H Ding. Impacts of average air temperature derived from maximum and minimum temperatures on annual mean air temperatures series of China. Journal of Applied Meteorological Science, 18, 187-192(2007).
[38] X Y Wen, S W Wang, J H Zhu et al. An overview of China climate change over the 20th century using UK UEA/CRU high resolution grid data. Chinese Journal of Atmospheric Sciences, 30, 894-903(2006).
[39] Y Xu, G L Tang, Q Zhang. Analysis of the variation of the air temperature over China during the global warming hiatus period. Climate Change Research, 13, 569-577(2017).
[40] J M Yan. Lessons and enlightenment from land exploration and utilization in the west of the USA. Journal of Peking University (Humanities and Social Sciences), 38, 119-126(2001).
[42] D E Zhang. Historical records of environmental changes and agricultural development in Northwest China. Advances in Climate Change Research, 1, 58-64(2005).
Get Citation
Copy Citation Text
Wanhui YU, Lijuan ZHANG, Hongwen ZHANG, Lanqi JIANG, Ankang ZHANG, Tao Pan. Effect of farmland expansion on drought over the past century in Songnen Plain, Northeast China[J]. Journal of Geographical Sciences, 2020, 30(3): 439
Received: May. 30, 2019
Accepted: Sep. 9, 2019
Published Online: Sep. 29, 2020
The Author Email: ZHANG Lijuan (zlj19650205@163.com)