| As an important parameter of vegetation canopy structure, leaf area index(LAI) provides scientific basis for the study of vegetation canopy characteristics, crop growth monitoring and pest evaluation. Using remote sensing technology to obtain regional LAI quickly and dynamically is of great significance to study the growth of vegetation.We made field measurement of winter wheat in Luancheng County of Hebei Province in late April, 2014. Using the measured parameters as inputs to PROSPECT+SAIL model, the sensitivity of parameters was analyzed. Based on model simulation, the relationships between LAI and vegetation indices(VI) under different observation zenith angles were analyzed. Using Landsat-8 OLI image on May 6, 2014 and statistical models, LAI of winter wheat in Luancheng County was estimated. In combination with MODIS MOD15A2 and MCD12Q1 products and meteorological data, spatio-temporal change of LAI in growing seasons in Hebei Province from 2002 to 2011 was analyzed. The relationships between LAI and air temperature and precipitation were also analyzed. We made LAI field measurement of corn in Xingtai City on August 9-21, 2013. Based on Landsat-8 OLI image on August 23, 2013, using correlation analysis and relative error analysis methods, the accuracy of MODIS LAI and SPOT/VGT LAI was compared. Spatio-temporal distribution difference of these two LAI products in Hebei Province in 2013 was analyzed. The main conclusions were as follows:(1) In combination with physiological and biochemical parameters of winter wheat measured in Luancheng County, the sensitivity of parameters was analyzed using PROSPECT+SAIL model. At visible light wavelength(400-700nm), the sensitivity of chlorophyll content to canopy reflectance of wheat was obvious. The sensitivity of LAI was high at 750-1400 nm. The sensitivity of water content was high at 1300-2400 nm.(2) Based on PROSPECT+SAIL model, the relationships between LAI and NDVI, EVI, RVI, DVI, SAVI and OSAVI under different view zenith angles were analyzed. When view zenith angle was 0°, the logarithmic relationship between LAI and each vegetation index had maximum correlation coefficient, the order was RVI > DVI > NDVI > OSAVI > SAVI > EVI.(3) In the experimental area of Luancheng Country and Xingtai City, based on measured LAI and OLI images, regression models between measured LAI and six vegetation indices were established and the accuracy was verified. In Luancheng County, the correlation coefficient of exponential model between OSAVI and measured LAI was the maximum(R2=0.8634), but root mean square error and mean relative error of power model between EVI and measured LAI were the minimum(RMSE = 0.374, MRE = 5.7%). Considering two kinds of verified results(12 sampling points and 42 sampling points), LAI-OSAVI exponential model was used to estimate LAI of winter wheat in Luancheng County. In Xingtai experimental area, the correlation coefficient of logarithmic model between NDVI and measured LAI was the highest(R2=0.6448), NDVI was selected to estimate LAI of corn.(4) In Luancheng experimental area, OSAVI calculated from OLI image on May 6, 2014 was used to estimate LAI and to make LAI map. LAI of winter wheat in the south and southeast of Luancheng County was higher; the main reason was that vegetation coverage of farmland was high; while LAI was very low in residential areas in northern and central regions of Luancheng County.(5) The annual average LAI values in growing seasons in Hebei Province were 1.0-1.3 from 2002 to 2011. There was no correlation between temperature and annual average LAI, precipitation was the main factor and it had positive correlation with annual average LAI(R2=0.6428). The spatial distribution of annual average LAI presented the trend of first increase and then decrease from east to west, but slightly increase in Taihang Mountainous area in the west.(6) The scale effect of LAI estimation using different sensor data was analyzed. In Xingtai experimental area, 30 m OLI estimated LAI on August 23, 2013 was resampled to 1km OLI LAI, and LAI values changed little. The distribution pattern of 1km MODIS LAI on August 21, 2013 and 1km SPOT/VGT LAI on August 24, 2013 was similar, but both of them underestimated vegetation LAI. The underestimated areas were mainly located in the edge of villages with sparse vegetation. MODIS LAI product underestimated by 31.4%, SPOT/VGT LAI product underestimated by 13.3%.(7) The spatial and temporal distribution patterns of MODIS and SPOT/VGT LAI were basically consistent in Hebei Province in 2013. The area where MODIS LAI was lower than SPOT/VGT LAI accounted for about 76.7% of Hebei Province, and the possible reason was that with the increase of surface heterogeneity, the underestimation of mixed pixel LAI using MODIS was more serious than using SPOT/VGT. The accuracy of SPOT/VGT LAI products was higher than that of MODIS LAI products, and SPOT/VGT LAI products were more suitable to monitor LAI dynamic changes of farmland in Hebei province. |
PDF Full Text Request