Increasing carbon sequestration of terrestrial ecosystem can effectively reduce CO2 concentration in the air (
Journal of Geographical Sciences, Volume. 30, Issue 9, 1507(2020)
Ecosystem carbon storage under different scenarios of land use change in Qihe catchment, China
Regional land use change is the main cause of the ecosystem carbon storage changes by affecting emission and sink process. However, there has been little research on the influence of land use changes for ecosystem carbon storage at both temporal and spatial scales. For this study, the Qihe catchment in the southern part of the Taihang Mountains was taken as an example; its land use change from 2005 to 2015 was analyzed, the Markov-CLUE-S composite model was used to predict land use patterns in 2025 under natural growth, cultivated land protection and ecological conservation scenario, and the land use data were used to evaluate ecosystem carbon storage under different scenarios for the recent 10-year interval and the future based on the carbon storage module of the InVEST model. The results show the following: (1) the ecosystem carbon storage and average carbon density of Qihe catchment were 3.16×107 t and 141.9 t/ha, respectively, and decreased by 0.07×107 t and 2.89 t/ha in the decade evaluated. (2) During 2005-2015, carbon density mainly decreased in low altitude areas. For high altitude area, regions with increased carbon density comprised a similar percentage to regions with decreased carbon density. The significant increase of the construction areas in the middle and lower reaches of Qihe and the degradation of upper reach woodland were core reasons for carbon density decrease. (3) For 2015-2025, under natural growth scenario, carbon storage and carbon density also significantly decrease, mainly due to the decrease of carbon sequestration capacity in low altitude areas; under cultivated land protection scenario, the decrease of carbon storage and carbon density will slow down, mainly due to the increase of carbon sequestration capacity in low altitude areas; under ecological conservation scenario, carbon storage and carbon density significantly increase and reach 3.19×107 t and 143.26 t/ha, respectively, mainly in regions above 1100 m in altitude. Ecological conservation scenario can enhance carbon sequestration capacity but cannot effectively control the reduction of cultivated land areas. Thus, land use planning of research areas should consider both ecological conservation and cultivated land protection scenarios to increase carbon sink and ensure the cultivated land quality and food safety.
1 Introduction
Increasing carbon sequestration of terrestrial ecosystem can effectively reduce CO2 concentration in the air (
Previous research has shown that deforestation in tropical areas can cause terrestrial ecosystem carbon storage decrease globally (
CLUE-S (Conversion of Land Use and its Effects at Small region extent) model is a typical empirical statistical model that treats the competition between different types of land use based on systems theory, simulates different land use types simultaneously, and produces a spatially explicit display of the simulation results. This model has been recognized as an excellent tool for simulating land use changes (
2 Data source and methodology
2.1 Overview of the study area
Qihe catchment falls on the eastern slope of the southern section of Taihang Mountains, between 13°17°-114°23°E and 35°32°-36°04°N, at east edge of the second step of Chinese terrain and connects with the North China Plain (
Figure 1.
The vegetation of this area has significant vertical variation, according to the records of vertical zonation of vegetation in the southern section of Taihang Mountains (
2.2 Data sources and process
Land use data includes two terms, the first, from 2005, comes from National Earth System Science Data Sharing Infrastructure-Yellow River Downstream Scientific Data Center (
The driving factors selected in this study for land use changes include topography, soil and accessible factors. Topographic factors include elevation, slope and aspect, which were extracted from DEM (from ASTER GDEM); soil factors include soil type, soil organic matter and total nitrogen concentration (from Yellow River Downstream Scientific Data Center) and were transferred into raster data; accessible factors include the distance to urban, rural residential settlements, rivers, provincial roads and county roads, which were calculated by Euclidean distance method. Urban and rural settlements were extracted from 2005 secondary land use class graph, data for rivers, provincial roads and county roads are from Yellow River Downstream Scientific Data Center. All vector data in this study are obtained from maps at a scale of 1:100,000, raster data are all in grid format, spatial resolution is 150 m and geographic coordinate system is WGS_1984_Albers.
2.3 Prediction of land use change based on Markov-CLUE-S composite model
(1) Markov-CLUE-S Composite Model
The operation of Markov-CLUE-S composite model includes the following five aspects:
1) Setting of restricted regions. According to the actual situations of Qihe catchment, for this study, restricted regions were not set. All land use types are allowed to be transferred.
2) Converting elastic coefficient and transfer matrix. Converting elastic coefficient represents the difficulty level of transferring certain land type into others. It can be expressed by parameter ELAS (0-1), a larger ELAS value indicates higher stability, and a smaller transformation probability. This study referred to 2005-2015 land use transformation probability, combined with 2015 land use simulation precision and Kappa coefficient, and multiple debugging sessions were conducted to achieve the best simulation result. Finally, we set converting elastic coefficients of cultivated land, woodland, grassland, water body, construction land and unused land as 0.7, 0.7, 0.7, 0.8, 0.9 and 0.6, respectively. Transfer matrix represents the transformation rules of all types of land, 1 represents “can be transferred”, and 0 means “cannot be transferred”. In this study, the value was set 1 for all the situations.
3) Calculation of land use demand. This study is based on land use data in 2005 and 2015, using the Markov model and linear interpolation to calculate land use demand for both simulated and selected years.
4) Spatial analysis. According to land use pattern and driving factor trait data, we employed Binary Logistic stepwise regression to diagnose the probability of certain land types occurring in each raster grid.
5) Model test. The ROC curve was used to verify logistic regression results. If the result is larger than 0.7, then selected driving factor has relatively good explanation capacity (
(2) Land use scenarios
Since Qihe catchment is not a compact administrative district, it is hard to predict its future land use required. For this paper, scenarios analysis was performed, according to features of each scenario and
1) Natural growth scenario (Q1). According to 2015 land use data and 2005-2015 matrix of land use transformation probability, taking 10 years as the step size, we predicted areas of all land use types in 2025 within the study area under natural growth scenario.
2) Cultivated land protection scenario (Q2). This scenario aims to strengthen the protection of cultivated land through curbing the expansion of construction land and slowing down the transformation probability of cultivated land into other land types. Under this scenario, the transformation probability of cultivated land into construction land decreased by 80%, and the transformation probability of cultivated land into woodland, grassland, and water body decreased by 30%, and the probability of cultivated land being transferred into unused land decreased by 100%.
3) Ecological conservation scenario (Q3). This scenario strengthens the protection for woodland, grassland and water body, but also strengthens the transformation of other land use types into ecological land. Under this scenario, the transformation probability of wooland, grassland and water body into construction land decreased by 90%, and the probability for transforming woodland, grassland and water body into unused land decreased by 100%, and the probability for cultivated land being transformed into woodland decreased by 20%, and the probability for transforming cultivated land into construction land decreased by 60%.
Based on the 2005-2015 land use transformation probability, combined with land use demand under different scenarios, we formulated ELAS value for all land use types in different scenarios (
The ELAS of all land use types under different scenarios
The ELAS of all land use types under different scenarios
|
2.4 Ecosystem carbon storage evaluation based on InVEST model
(1) Estimation of carbon storage. Usually, the estimation of ecosystem carbon storage by InVEST model includes four basic carbon sinks (aboveground, underground, soil and dead organic matter). However, due to the difficulty of obtaining carbon sink data for dead organic matter, for this study, only the other three carbon sinks were considered. The principle of calculating carbon storage is:
where
(2) Determination of carbon density data. At first, combining the results of related studies (
Carbon density of different land use types in the Qihe catchment, China (t/ha)
Carbon density of different land use types in the Qihe catchment, China (t/ha)
|
Based on land use and carbon density data in carbon module of InVEST model, we calculated the carbon storage in 2005, 2015 and 2025 in Qihe catchment and conducted an in-depth analysis of the spatio-temporal changes.
3 Results analysis
3.1 Land use simulation test and change features
(1) Diagnose of land use change driving factors. The logistic regression model was used to conduct regression analysis on 2005 land use data and driving factors (
Results of logistic regression for different land use types in 2005 in the Qihe catchment, China
Results of logistic regression for different land use types in 2005 in the Qihe catchment, China
|
The distribution probability of cultivated land is mainly related to topography, soil type and the distance to rural residential settlements and major roads. Specifically, it is highly related to slope, namely, when slope increases by 1º, the distribution probability of cultivated land will decrease by 12.48%. The distribution probability of woodland has significant positive correlation with total nitrogen concentration in soil; when total nitrogen concentration increases by 1 g/kg, the distribution probability of woodland will increase by 10.88%. The factor for woodland distribution is slope; when slope increases by 1º, the distribution probability of woodland will increase by 10.29%. The distribution probability of grassland has significant negative correlation with soil organic matter concentration; when organic matter concentration increases by 1 g/kg, the distribution probability of grassland will decrease by 7.91%. The next most influential factor for grassland is slope; when slope increases by 1º, the distribution probability of grassland will increase by 3.95%. Compared with the above land use types, there are fewer factors that can explain the distribution of water body, construction land and unused land, moreover, the correlation between land use types and driving factors is not significant. The distribution probability of water body is mainly related to soil type and organic matter concentrations; the distribution probability of construction land has positive correlation with organic matter and negative correlation with slope; the distribution probability of unused land is significantly correlated with both altitude and slope.
(2) The features of land use change of Qihe catchment in 2005-2015.
Land use transfer matrix for 2005-2015 (ha) in the Qihe catchment, China
Land use transfer matrix for 2005-2015 (ha) in the Qihe catchment, China
|
(3) Simulation results test. For the simulated 2015 land use distribution based on land use data of 2005, a precision test based on Kappa coefficient was conducted.
Figure 2.
3.2 Scenarios analysis of land use changes
Figure 3.
Figure 4.
3.3 Spatio-temporal change of carbon storage and density in ecosystem
According to
Figure 5.
In addition, carbon density decreased mostly in the areas with altitudes lower than 600 m or between 1300 m and 1500 m in 2005-2015; carbon density mostly increased in the areas with altitudes of 1000-1100 m and 1700-1800 m. Under natural growth scenario, carbon density in low altitude area continued decreasing; under cultivated land conservation scenario, carbon density in low altitude area increased slightly; under ecological conservation scenario, the increase of carbon density mainly occurred in the area with an altitude higher than 1100 m.
Figure 6.
According to
Statistics on carbon density changes during 2005-2015 and 2015-2025 under different scenarios in the Qihe catchment, China
Statistics on carbon density changes during 2005-2015 and 2015-2025 under different scenarios in the Qihe catchment, China
|
4 Conclusions and discussion
4.1 Conclusions
(1) Cultivated land and woodland areas in Qihe catchment in 2005-2015 declined by 16.99% and 2.57%, respectively. Water body area and construction land greatly expanded, with increases of 50.27% and 37.8%, respectively. In 2015-2025, under natural growth scenario, cultivated land will continue decreasing and construction land will continue expanding; under cultivated land protection scenario, cultivated land will increase by 5.53%, whereas construction land will only increase by 0.11%; under ecological conservation scenario, cultivated land will decrease by 5.41%, and ecological land all shows trend of increasing to different degrees.
(2) Carbon storage and density in 2015 were 3.16×107 t and 141.9 t/ha, respectively, which represent decreases of 0.07×107 t and 2.89 t/ha over the preceding 10 years. In 2005-2015, carbon density mainly decreased in low altitude area, areas with increased carbon density were equal to areas with decreased carbon density in high altitude areas. The large expansion of construction land in the middle and lower reaches of Qihe and the woodland degradation in the upper reach are the main reasons for carbon density decrease.
(3) In 2015-2025, under natural growth scenario, carbon storage and density will also significantly decrease by 0.03×107 t and 1.38 t/ha, respectively. Areas with decreased carbon are far more prevalent than areas with increased carbon, which is attributed to the decrease of carbon sequestration ability in low altitude region; under cultivated land protection scenario, the reduction of carbon storage and carbon density is slowed down (0.01×107 t and 0.44 t/ha), which is due to the strengthening of carbon sequestration ability in low altitude region; under ecological conservation scenario, areas with increased carbon are more prevalent than areas with decreased carbon, both carbon storage and density will significantly increase and can reach 3.19×107 t and 143.26 t/ha, respectively, mostly occur in the area where the altitude is higher than 1100 m.
In conclusion, carbon sequestration ability in Qihe catchment shows a decreasing trend in 2005-2015. This decrease continues in natural growth scenario, whereas the cultivated land protection scenario can effectively control the decrease, and the ecological conservation scenario can strengthen carbon sequestration ability but cannot effectively control the area decrease of cultivated land. Therefore, for land use structure optimization of Qihe catchment in the future, it will be helpful to consider both cultivated land protection and ecological conservation scenario, properly control the expansion of construction land in low altitude areas, strengthen the protection for high quality cultivated land, conduct ecological restoration project in high altitude areas, and concede low quality cultivated land on steep hill to forest. At the same time, increasing carbon storage, food safety and cultivated land quality will be ensured.
4.2 Discussion
(1) Carbon density data in this study were obtained through searches and model modifications, compared with previous regional studies that directly used carbon density value at national level, method used in this paper is a novel effort and will be beneficial for improving accuracy of regional ecosystem carbon storage estimations. Due to the limitation of data availability, we collected several kinds of measured or model simulated carbon density data around study area and compared them with the results in this study.
(2) The application of Markov-CLUE-S composite model has overcome the disadvantage of single model, displayed the advantages on quantity prediction and spatial allocation of two models, achieved double optimizations quantitatively and spatially for land use change simulations, and improved the accuracy of the prediction for future land use patterns. In the meantime, there are also some shortages. At first, when setting future land use scenarios, it only estimated required areas of all land use types for different scenarios through revising Markov model transfer probabilities but did not consider related local policies. However, with the social and economic development, future land use changes will be influenced by local policies more and more. Thus, how to set more reasonable land use demand combined with policy will be a focus for future research on land use change simulations. Second, selection of driving factors has considerable influence on simulation precision. The driving factors selected in this study are all easy to be spatialized, such as topography, soil feature and distance. However, regional land use changes are usually affected by socio-economy and policy. Therefore, selected driving factors are not comprehensive and have weakened the explaining effects of regression functions to some degree. In the future, how to quantitatively and spatially express socio-economic factors and related policies, and bring them into driving factors system, will be crucial to improve simulation precision of CLUE-S model.
(3) There are some uncertainties regarding the estimated carbon storage. At first, input data for InVEST model is uncertain. When the Markov-CLUE-S composite model is used to simulate land use patterns in different scenarios, due to less consideration of effects by socio-economy and related local policies, simulated results have some uncertainties. In addition, although the modified carbon density value drawn from previous researches and meteorological data is close to carbon density data around the study area, carbon density value may undergo dynamic changes under human activities and environmental changes, thus the modified carbon density value can be uncertain for carbon storage estimation. Second, the uncertainty of estimated results comes from the model itself. Carbon storage module of InVEST model is more focused on carbon density difference among various land use types but ignores carbon sequestration difference associated with land use types and vegetation age organizations, which brought some obstacles to carbon storage service spatial pattern simulations and make the estimation uncertain. In future research, measured data through field survey should be obtained to verify the reasonability of carbon density value, or enough field monitoring data should be collected to reveal effects on carbon density under inner space heterogeneity of land use types and vegetation age structures, to improve the accuracy for regional ecosystem carbon storage evaluation.
[4] et alTotal below ground carbon allocation in China’s forests. Acta Ecologica Sinica, 27, 5148-5157(2007).
[5] Researches on carbon sequestration functions of main forest types in northern China(2003).
[8] et alMethods, tools and research framework of ecosystem service trade-offs. Geographical Research, 35, 1005-1016(2016).
[10] Biomass and net production of forest vegetation in China. Acta Ecologica Sinica, 16, 497-508(1996).
[11] et alCarbon sequestration in China’s terrestrial ecosystems under climate change: Progress on ecosystem carbon sequestration from the CAS Strategic Priority Research Program. Bulletin of the Chinese Academy of Sciences, 30, 848-857(2015).
[19] Modeling study of vegetation shoot and root biomass in China. Acta Ecologica Sinica, 26, 4156-4163(2006).
[20] Prediction of carbon exchanges between China terrestrial ecosystem and atmosphere in 21st century. Science in China Series D: Earth Sciences, 38, 211-223(2008).
[23] The vertical zonation of mountain vegetation in Henan. Journal of Henan Normal University (Natural Science Edition), 19, 91-95(1991).
[26] Accurate estimate of soil organic carbon storage in Henan Province based on high-density profile(2016).
[27] Vegetation and soil carbon storage in China. Scientia Sinica (Terrae), 33, 72-80(2003).
[31] Land use scenario design and simulation based on Dyna-CLUE model in Dianchi Lake Catchment. Geographical Research, 34, 1619-1629(2015).
[38] A research on the vegetation in the southern part of Taihang Mountain. Journal of Shanxi Normal University (Natural Science Edition), 7, 54-58(1993).
[42] Study on the ecosystem carbon storage of different land use types in the Taihang Mountain(2007).
[43] Land use optimization simulation based on low-carbon emissions in eastern part of Qinghai Plateau. Geographical Research, 34, 1270-1284(2015).
[45] et alSpatial variations of terrain and their impacts on landscape patterns in the transition zone from mountains to plains: A case study of catchment in the Taihang Mountains. Scientia Sinica (Terrae), 48, 476-486(2018).
[46] et alSpatial distribution of forest vegetation carbon stock in China based on HASM. Acta Geographica Sinica, 24, 1212-1224(2013).
Get Citation
Copy Citation Text
Wenbo ZHU, Jingjing ZHANG, Yaoping CUI, Lianqi ZHU. Ecosystem carbon storage under different scenarios of land use change in Qihe catchment, China[J]. Journal of Geographical Sciences, 2020, 30(9): 1507
Category: Research Articles
Received: Nov. 29, 2019
Accepted: Feb. 10, 2020
Published Online: Apr. 21, 2021
The Author Email: ZHU Wenbo (zhuwb517@163.com)