Study On The Model Of Northeast Forest Carbon Cycle At Daily Step And The Integrated Application Of Remote Sensing

Forests are the major terrestrial carbon pool, an accurate assessment of forest carbon balance and storage is the key to estimate future atmospheric CO2 concentrations, climate changes and its impact on terrestrial ecosystems. Currently, forest carbon cycle is one of the focuses on global climate change research, and the simulation of forest ecosystem productivity is the key to carbon cycle study. Northeast China forest is one of the three largest temperate forests in the world, accounting more than 1/3 on the national total forest area and the storage capacity, also it plays an important role in our country and global carbon cycle, forestry and ecological environment construction. However, the study on forest carbon cycle in Northeast area is not comprehensive, research results from the region in China and the global carbon cycle assessment, modeling and forecasting are also urgently needed.In this paper, we conducted a quantitative estimate of foresty productivity in Northeast China and Heilongjiang province in 2007 by using remote sensing mechanism model, and made a deep analysis on temporal and spatial pattern change and its impact factors (LAI, weather, etc.), producing a impeccable method system to quantitative estimate the carbon cycle of northeast forest by remote sensing mechanism model through studying the temporal and spatial pattern of northeast forest ecological system carbon budget and its response mechanism of climate change. Based on the purpose of study, the model input data includes MODIS land cover type of lkm resolution, LAI data of MODIS, meteorological data and soil data,then estimate the net ptimary productivity (NPP)of the Northeast forest, gross primary productivity (GPP), plant autotrophic respiration (Ra) in northeast by corrected remote sensing mechanism model, compare the obtain NPP with the flux data, using iterative method to optimize the vegetation physiological parameters until getting the parameters that fitting northeast forest. An estimate of forest NPP,GPP and Ra through vegetation physiological parameters by kinds of optimize method using 100m resolution Landsat data, meteorological data and soil data, and made an analysis of temporal and spatial pattern change and impact factor on this three types data of two resolutions in the model.Aim at the above issues address the following research and draws the main conclusions of the corresponding:1,Preprocessing MODIS and TM data. TM data preprocess contains Multi-spectral data form, geometric correction, radiometric calibration, atmospheric correction, etc. In order to reduce the workload of remote sensing data preprocessing, this study developed a remote sensing image IDL language preprocessor (radiometric calibration procedures, calibration procedures for the solar elevation angle, radiation normalization procedures, etc.). The preprocessing on three MODIS data products included image mosaic, crop, projection change and so on. Focus on the application of LACC algorithm procedures, achieved very good results on processing the 46 MODIS LAI data of Northeast in 2007.2,TM remote sensing image classification. This study used a more sophisticated maximum likelihood classifier for forest classification of TM remote sensing images in Heilongjiang Province in 2007, classification system is coniferous forest, broadleaf forest and conifer, respectively, of the three regional TM remote sensing image classification and comparative analysis of the intensive through TM band and increase using different four auxiliary information (NDVI, DEM, VDVI+DEM) methods, using the highest classification accuracy method, finally, Daxing'anling and Xiaoxing'anling region were classify by 6B, using the 6B+ DEM classification method in Changbai Mountain. The classification accuracy in the three areas:68.4%,77.7%,79.1%. And the classification results were compared with the MODIS land cover products.3,LAI quantitative inversion. In this paper, base on mid-Auguest TM remote sensing data, made use of geometric optics model 4-Scale and exponential statistical model combining mixed methods, estimated LAI in the three major forest areas in Heilongjiang Province (Daxing'anling, Xiaoxing'anling, Changbai Mountaint), compared conventional and spatial statistical analysis on LAI inversion results of the three regions.The results were:in Heilongjiang forest, broad-leaved forest LAI average was about 3.1, Mixed forest LAI average was about 2.3, coniferous forest LAI average was about 2.3, regardless of forest type in the entire forest area of Heilongjiang Province, the average LAI was about 2.56.4,Establish the soil database. The soil data (AWC) data is determine by the soil texture, we can obtain soil texture map soil type data in the Northeast region by using the conversion relationships based on soil types, also obtained lkm of Heilongjiang soil water raster data files by applying the forms models of De Jong and Loebe.5,The establishment of the Northeast regional meteorological databases. A daily meteorological data change analysis was made on 96 sites of the three northeastern provinces and 10 sites of Inner Mongolia meteorological data from National Meteorological Information Center, as the results for the year 2007 without the normal phenomenon of drought or floods. Based on the data, produced the daily grid 1km data by Kruger interpolation. The database contains five meteorological data 1825 grid file included the daily minimum temperature, maximum temperature, daily solar radiation, daily precipitation, relative humidity, the data capacity of 15 G.. The workload of Meteorological Data Processing was particularly large, so we write a batch preparation procedure for meteorological data processing in this study.6,Model parameter optimization and algorithm improvements. The first consideration of BEPS model application is the issue on model adaptation, this study optimized the vegetation physiological parameters of BEPS by four methods:measurement method, literature, remote sensing method, iterative method, and finally made the vegetation Physiological parameters of BEPS model fit the estimates of Northeast forest NPP and GPP. So the model algorithm has been optimized at the same time, not only improve the model speed, but also make the model to calculate input data of different spatial resolutions.7,The development of visual BEPS model running systems. Applications Microsoft Visual C++ 6.0 platform, developed a comprehensive, advanced visualization algorithms to run BEPS model system,then realized the intelligence read of LAI data, daily weather data and the choice output of NPP, GPP and Ra, also, input and output data format was fully compatible with the ENVI standard format. Subsystems after the non-professional staff to obtain input data, the system can be applied in estimating NPP, GPP and Ra of the study area, and it is benefit to BEPS model extension and application.8,Estimation and temporal and spatial analysis of NPP, GPP and Ra. Apply GIS and statistical methods, data estimated by MODIS net primary productivity of the Northeast (NPP), gross primary productivity (GPP), plant autotrophic respiration (Ra) for the spatial and temporal distribution of the entire Northeast Forest. Average GPP in the whole northeast forest was 897.33gC·m-2·a-1 in 2007, the maximum was 1294.89gC·m-2·a-1, and the minimum was 494.51gC·m-2·a-1. Average NPP was 369.92gC·m-2·a-1 in 2007, the maximum was 632.82gC·m-2·a-1, and the minimum was 75.17gC·m-2·a-1. Average Ra was 296.86 gC·m-2·a-1 in 2007, the maximum was 529.51gC·m-2·a-1 and the minimum was 80.35gC·m-2·a-1. In case of Xiaoxing'anling, this paper performing a variety of comprehensive comparative analysis on different NPP of different spatial resolution at a year, a season, a month and a day, respectively. From the comparison results, it had little relationship using two sets of disaggregated data (classification data of 1km resolution MODIS and 100m resolution TM) to obtain NPP of different spatial resolutions in Xiaoxing'anling forest in 2007.9,Variety of authentication methods.Compared the consistency conclusions by using permanent plots of forest resources inventory data, MODIS NPP products and studies of other comparison model for NPP with obtained NPP simulation results of this study,it was more realistic. MODIS NPP data with other researchers to obtain research results of a comparative analysis, this paper simulated the average NPP was 369.92gC·m-2·a-1 in Daxing' anling and Xiaoxing' anling, the maximum was 632.82 gC·m-2·a-1 while the minimum was 75.17gC·m-2·a-1, all simulated values were within the range, indicating that BEPS model NPP simulation is more reasonable and reliable in Daxing'anling and Xiaoxing'anling.Also made the uncertainty analysis of the estimation results to clarify the cause of the errors.10,NPP impact factors and sensitivity analysis. In this paper we apply SPSS13.0 software, using a linear regression model region of small impact factor of NPP (LAI, temperature, precipitation, solar radiation, dimensions).The regression analysis showed that, the relationship between LAI and NPP was extremely significant, The correlation coefficient was 0.515, the reason solar radiation LAI directly determine the energy absorption of vegetation. Second, the dimension correlation coefficient was 0.197. The relationship of NPP and precipitation and temperature was positively correlated with a low correlation coefficient.Selected the vegetation parameters (LAI), meteorological factors (including temperature, precipitation and solar radiation) and other major factors to simulate the impact of different factors increase or decrease the results after the NPP, the impact ty analysis, statistics, after changes in the impact factor the change in NPP, and analyzed the variation of different factors.