| As a key ecological barrier in the arid and semi-arid regions,the agro-pastoral ecotone of Northwestern China(APENC)plays an important role in preventing desertification from moving southward and eastward.The ecological environment in this region is extremely fragile and has a poor environmental carrying capacity,low self-restoration ability and high sensitivity to external disturbance.During the past few decades,improper land use has presented a serious ecoenvironmental problem in APENC,including grass degradation,groundwater decline,soil erosion,and sandy desertification.These problems have threatened the sustainability of local economic growth and social development.In order to control the degradation of the ecological system,a number of large-scale ecological restoration(ER)projects have been implemented in this area since 1970s,resulting in a great impact on land use/cover changes(LUCC),water use efficiency(WUE),soil moisture and regional water balance However,how the ER projects affect the WUE and soil moisture at the regional scale remains unclear.This study aims to answer this research question.Firstly,this study analyzed the spatio-temporal characteristics of LUCC in the APENC from 1980 to 2015 Then,the break for additive season and trend(BFAST)method was applied to analyze the regional WUE dynamic tendency and its driving forces at the APENC during 19822015 period.The optimal fitting method of dry and wet edges of temperature-vegetation feature space was used to improve the accuracy of temperature vegetation dryness index(TVDI)in monitoring soil moisture in the study area.Finally,an improved residualtrend analysis(RESTREND)method was used to distinguish and quantify the effects of climate change and human activities on vegetation and soil dryness from 2000 to 2015,and subsequently to identify soil desiccation risk areas caused by vegetation restoration.The main conclusions are as follows:(1)Grassland,desert and cultivated land were the main land use types in APENC.The dynamic change patterns of land use were mainly stable and repeated,and the dynamic change mainly occurred before 2000.After 2000,the land use types tended to be stable.The main land use change was the transformation between grassland and desert,farmland and grassland.Analysis of landscape indicators show that,the landscape fragmentation,landscape diversity,the shape of the patch tends to be regular and simple,patch spatial distribution presented a discrete situation.(2)Results show that the WUE of the ecosystem increased from 1982 to 2015,and WUE in more than 50%of the region changed significantly.On average,every year,1.5%of the APENC ecosystems experienced an abrupt change in WUE,the turning points mainly occurred before 1990 and after 2000.The mutation trend of WUE was dominated by a positive reversal trend that first decreased and then increased and an interrupted increased trend.Vegetation change,LUCC and droughts had great influence on WUE change.The correlation between ND VI(Normalized Difference in Vegetation Index)and WUE was between moderate and high in most areas.The WUE of farmland and forest was mainly increasing,while the WUE of desert and grassland was decreasing.The transfer between land use types affected the change of WUE.When the grassland was converted into farmland,the WUE showed monotonic increasing trend and a suddenly trend of increase and then decrease,when the grassland was converted into desert,most of WUE showed a monotonic decreasing trend in most of the area.Although the WUE of grassland was converted into forest mainly showed monotonic increasing trend and a suddenly trend of increase and then decrease,the proportion of the trend type of interrupted decrease and a suddenly trend of decrease and then increase was also relatively large.About 52.5%of the detected trend shifts coincided with dry years,of which the large majority corresponded to moderate dry conditions.Within a certain range,drought intensification led to the increase of WUE.(3)In the calculation of TVDI,the binomial function is better than linear and exponential function in fitting the dry and wet edges of temperaturent-vegetation feature space,which with R2 of the dry and wet edges of binomial fitting being the best among the three fitting methods.The results of 3 kinds of function fitting were compared with those from the in situ soil moisture observations in the study area.Mean absolute error(MAE)and root mean square error(RMSE)were calculated,for the 5cm,10cm,15cm and 30cm depths of the soil.In the 4 different depths,RMSE values at 5cm,10 cm,15 cm and 30cm depth were 0.44,0.42,0.44 and 0.43,respectively.MAE values at 5cm,10 cm,15 cm and 30cm depth were 0.46,0.46,0.47 and 0.48,respectively.The results indicated that the TVDI feature space inversion based on dry edges and wet edges fitting with the binomial function was the most accurate for the soil surface water content in APENC,and it was most sensitive to water content at 10 cm depth in soil.(4)The generalized additive model(GAM)was used to improve the traditional residual trend analysis method and improve the ability of the RESTREND method to separate the effects of climate factors and human activities on vegetation and soil moisture.The results showed that NDVI and TVDI changes were mainly affected by anthropogenic activities.Climatic factors contributed 43%to the vegetation variability and 22%to the TVDI dynamics,and anthropogenic activities contributed 57%to the vegetation variability and 78%to the TVDI dynamics.The positive and negative anthropogenic contribution rates to the NDVI and TVDI were compare to quantify the geographic extent and direction of anthropogenic impacts on the variation in vegetation and soil moisture.The positive contribution(37%on average for the study area)of vegetation expansion caused by anthropogenic activities was higher than the negative contribution(20%)of vegetation reduction.On the contrary,the impact on soil moisture decrease by vegetation expansion from anthropogenic activities was observed in most areas,with a total contribution rate of 47%,which was more than that of anthropogenic increase in soil moisture(covering 31%).The vegetation expansion,which was mainly related to the large number of ER programs,was found to increase soil moisture depletion,especially in the northwestern,northeastern,eastern and southern parts of the APENC.By comparing areas where anthropogenic activities had a high contribution rate to vegetation increase and areas where soil moisture consumption was severely increased,hotspot areas of soil moisture consumption caused by the ER programs were identified and mapped.The current methodological workflow and results represent a novel foundation to inform and support water resource management and ecologicalrestoration-related policy making. |
PDF Full Text Request