Measurement and retrieval of leaf area index using remote sensing data in Kanas National Nature Reserve,Xinjiang

Leaf area index(LAI) is one of the most important vegetation structural parameters in terrestrial ecosystems.It significantly regulates the exchanges of matter and energy between the atmosphere and terrestrial ecosystems due to its effects on many biophysical and physiological processes,including photosynthesis,respiration,transpiration,precipitation interception,energy transfer,and so on.The study on LAI is of scientific importance for assessing the response of vegetation to climate change and human activities in ecologically fragile areas in northwest China.In this paper,effective LAI(LAIe) and actual LAI of forests and grasslands were measured with the LAI-2200 and TRAC instruments in Kanas National Nature Reserve,Xinjiang Autonomous Region.Based on measurements of LAIe and LAI at 50 sampling plots,models for estimating LAIe and LAI from Landsat 5-TM remote sensing data were developed.Then,the changes of LAI with topographic factors(including elevation,slope,and aspect) were analyzed.The possibility of estimating vegetation biomass density based on LAI estimated from TM remote sensing data(TM LAI) was explored.Finally,the quality of the MODIS LAI product in the study area was assessed using TM LAI as the benchmark.The results show that both LAIe and LAI significantly change with land cover types.Both LAIe and LAI exponentially change with vegetation indices.The best fitted models for estimating LAIe and LAI are LAIe=0.4861e2.6801ARVI and LAI=0.4162e3.2706ARVI for needle-leaved forests,LAIe =2.3405e0.0611ARVI and LAI=1.5325e2.174MAVI for broad-leaved forests,LAIe =0.5575e2.499ARVI and LAI=0.5675e2.7732ARVI for mixed forests,LAIe=0.3627e4.3037SR and LAI=0.5125e4.1258ARVI for grasslands,respectively(AVRI is the atmospherically resistant vegetation index,SR is the simple ratio vegetation index,and MAVI is the moisture adjusted vegetation index).The averages of remotely sensed LAIe of broad-leaved forests,mixed forests,needle-leaved forests,and grasslands are 4.40,3.18,2.57,and 1.76,respectively.The corresponding values of LAI are 4.76,3.93,3.27,and 2.30,respectively.High values of LAIe and LAI mainly appears in locations nearby the lakes and rivers.The changes of LAI with altitude,slope and aspect exhibit obviously vertical patterns.LAI tends to increase first and then decrease with the increases in altitude and slope.Aspect has significant influences on the LAI of needle-leaved forests and grasslands,but less influences on LAI of broad-leaved forests and mixed forests.LAI can act as a valuable predictor of forest biomass density(BD)(BD=44.396LAI-25.946,R2=0.83,BD was derived from the forest resources inventory data).The average BD of forests estimated from remotely sensed LAI is 120.3 t/hm2 and the estimated total forest biomass is 5.0×106 t in the study area.The re-sampling technique was adopted here for getting 1 km resolution TM LAI data from 30 m TM LAI data of the study area.1 km MODIS LAI show similar patterns with LAI re-sampled from the 30 m LAI map generated from TM data and measured LAI,with R2=0.42 for forests and R2=0.53 for grasslands.However,MODIS LAI is 16.5% and 24.4% lower than LAI estimated using TM remote sensing data for forests and grasslands.