Study On The Spatial And Temporal Dynamic Changes Of Wetland And Its NPP In The West Of Jilin Province Based On Remote Sensing Techniques

Wetland ecosystems distribute in every climate and on all the continents except for Antarctica.They are playing an important role in the terrestrial ecosystem carbon cycle.Net primary productivity?NPP?of the wetland vegetation refers to the net amount of dry matter accumulated by the green wetland plants per unit ground area and time.It determines the ability of vegetation to provide the available material and energy to the other heterotrophic organisms within wetland ecosystems and is recognized an important parameter to assess the function of wetland.The west of Jilin Province is located at the junction of Songnen Plain and Horqin Grassland,which belongs to the typical ecotone.The region has diverse biological species and vulnerable habitats.The environment of western Jilin is relatively poor with serious saline and sandy desertification.The conflicts between human and land in this region are also serious.All these conditions make the wetland here occupy an even more important position and exert stronger wetland functionality than wetland at other places.In order to protect and reasonably utilize wetland or to maintain the function of its ecosystem,it is also of great significance to study the distribution as well as the dynamic trends of the area and NPP of different types of wetland there.Currently,the studies taking the west of Jilin as a complete unit mainly focus on the NPP of grassland and farmland ecosystem.Few studies were about the NPP of different kinds of wetland as well as their dynamics and influencing factors.In this study,western Jilin province was taken as the study area.The remote sensing imagery,meteorological data and field survey data of western Jilin were fused to form a comprehensive data source.By adopting the model simulation and GIS spatial analysis methods,the dynamics of wetland and NPP as well as its response to the environmental factors in this area were investigated.The key technologies and results of this study are drawn as follows:?1?A new hybrid filtering method was built to reconstruct the time series of normalized difference vegetation index?NDVI?,the land surface water index?LSWI?and the leaf area index?LAI?.This hybrid method was based on one global filtering method called Guass-Vanicek?G-V?,one semi-global fitting algorithm Asymmetric Gaussian?AG?,one local filtering algorithm Savitzky-Golay?S-G?,Grubbs test and iteration method.The phenological features were extracted from the reconstructed time series.All these data laid the foundation for the automatic extraction of the wetland region and the model simulation of its NPP.The results showed that the hybrid filtering method took the advantages of each algorithm and have an improving noise detecting and data processing ability.?2?A new hybrid feature selecting method integrating Relief F,Random Forest and NSGAII?Non-dominated Sorting Genetic Algorithm II?multi-objective genetic algorithm was constructed.It was used to pick the most efficient features from the spectral,index,phenological,textural and shape features.By adopting the Random Forest classifier,the land use maps of 2000,2009 and 2016 were obtained.The results showed that compared with the methods using the single type of features,the comprehensive multi-source and multi-type features improved the classification accuracy of wetland.The sepctural,index,phenological features were important features in this study.?3?The dynamic changes of wetland during the whole study period was analyzed.Wetland occupied 14%¹5%of the total area of western Jilin province.The total area reduced from 2000 to 2009 and increased from 2009 to 2016.The natural wetland and constructed wetland?mainly was paddy field?showed different dynamic trends.The total area of the natural wetland kept decreasing during the whole study period,among them,the swamp meadow had the largest extent of decrease,followed by marsh.The loss area mainly distributed in the area near the rivers and lakes in Zhenlai,Qianguo and Tongyu counties.The area of paddy field kept increasing during the whole study period.The increased area mainly distributed within the region of the major water conservancy projects in Zhenlai,Taonan and Qianguo.?4?A two-leaf light use efficiency model often used to estimate the gross primary productivity was fixed with the module of autotrophic respiration in MODIS BIOME-BGC model to simulate the NPP of vegetation.Besides,the soil moisture scaler,phenology scaler and photosynthetic photon flux density scaler were also integrated to the hybrid light use efficiency-process model.Then the daily NPP in western Jilin in 2000,2009 and 2016 were simulated.The daily NPP data were integrated on ten-day,month,season and year scales.The the total amount of NPP were divided into two parts.One part was NPP_su accumulated by the sunlit leaves and the other part was NPP_sh accumulated by the shaded leaves.The spatial and temporal dynamic changes of all the three kinds of NPP at different time scales were analyzed.The annual average NPP of the whole area showed an increased trend from152.9gC/m²/yr in 2000 to 198.6 gC/m²/yr in 2016 and made the total NPP of the region increase from 14.081TgC/yr to 18.287TgC/yr.The annual average NPP was larger in the east part of the region than in the west part.Wetland and the cropland had higher annual average NPP.Among all the types of wetland,the paddy field often had the largest NPP,followed by marsh.The value of annual average NPP_sh accumulated by the shaded leaves of these two wetland types were higher than NPP_su accumulated by the sunlit leaves.However,the annual average NPP_sh of flood plain and swamp meadow were lower than their annual average NPP_su.The changes of NPP,NPP_su and NPP_sh of the wetland during the year showed obvious seasonal pattern.All types of wetland had the max NPP in the summer followed by the NPP in autumn and spring.The proportion of NPP_su to NPP decreased from spring to summer and then increased from summer to winter.In the summer,for several types of wetland,the NPP_sh were larger than the NPP_su,and in other seasons,NPP_su accounted for the main part of NPP.?5?The partial correlation coefficient and the multiple correlation coefficient were adopted to reveal the response of NPP to the influencing factors such as mean temperature,precipitation,solar radiation,relative humidity,the concentration of CO₂in atmosphere,leaf area index and the land use types.For the majority of the wetland stable region,the annual average NPP was in significant positive relationship with LAI,annual mean temperature and the concentration of CO₂ in atmosphere.The increase of these factors promoted the carbon fixation by the wetland vegetation.In the contrast,the annual average NPP was in significant negative relationship with annual average relative humidity,annual precipitation,and the annual average solar radiation.The increase of these three factors hampered the carbon fixation process.For a certain region,the correlation between NPP and the energy-related factors were usually opposite to the correlation with the water-related factors.There was a synergistic action between the energy-related factors and water-related factors.The result of multiple correlation analysis showed that for the majority of the stable region,the annual average NPP was in significant related to the comprehensive factor set contains both energy-related and water-related factors.The influence from water-related factors on NPP is stronger than that from heat-related factors.It indicated that water-related factors were the main influencing factors in this region,especially in the wetland area.?6?The paired adjacent pixels for two different land use types in the same land use map were extracted.These paired samples were used to isolate the land use type controls on NPP.Besides,the pixels with the same land use type and similar LAI values during the whole study perid were extracted to isolate the climate controls on NPP.By conducting the single factor experiments,the effect of each influencing factors on NPP were quantified.In the test isolating the land use control,when the marsh converted into other land use types except for paddy field,all the annual average NPP,NPP_su and NPP_shh decreased.When it converted into paddy field,annual average NPP and NPP_sh were still generally reduced,but the annual average NPP_su increased.When paddy field converted into other types except for marsh,all the annual average NPP,NPP_su and NPP_sh decreased.In the test isolating the climate control,the climate change first decreased the Absorbed Photosynthetically Active Radiation?APAR?and LAI of the swamp meadow and then incrased them.It also decreased APAR of flood plain.For other land used types,the climate changes continuously increased their APAR and LAI.The climate changes first increased the annual average NPP of marsh but decreased it then.It also continuously decreased the annual NPP of flood plain.However,for other land use types,it continuously made quite a slight promotion on the annual average NPP.The result of the single factor experiments showed that,during the whole study period,the change of mean air temperature,the solar radiation and the precipitation made the largest contribution to the total annual NPP of the whole region.Among them,the solar radiation makes the most important influence on the actual value of annual average NPP.