The Determination Of LAI In Major Forest Types In The Heilongjiang Liangshui National Reserve

Hemispherical photographs and litter collection were used to measure leaf area index (LAI) combined of remote sensing to improve precision of measuring LAI only using combination of optical instruments and remote sensing from five forest types (Broad-leaved Korean pine forest, Spruse-pir forest, Birch forest, Korean pine plantation and Larch plantation) in the Heilongjiang Xiaoxing'an Mountains, northeastern China. Hemispherical photographs taken in early November were corrected by the woody-to-total area ratio (α), clumping index (ΩE, calculated using the CC correction method), and needle-to-shoot area ratio(γE, measured in the field). To bypass the need to collect wood samples and thus avoid damage to the forest, we propose a virtual method to remove the tree stems and branches with Photoshop. Linear regression analysis was then carried out between true LAI (corrected LAI on November 1 of combined litter collection LAI in each period) and the effective LAI (obtained only by hemispherical photography in each period). In order to detect the understory vegetation whether limit the accuracy of forest LAI estimated from remote sensing data in this study, we combined harvest method and hemispherical photography method to determinate understory LAI, and considered the gap fraction to estimate leaf area under gaps. Regression equations were established between total LAI (which was calculated from leaf area under gaps and canopy LAI) and reflectivity of TM, and the total LAI in the whole research area were measured. Again according to the relationship between the total LAI and canopy LAI, and the canopy LAI in the whole research area were also measured.The results showed that the canopy LAIs measured by hemispherical photographs were lower than the true values. The canopy LAIs measured from hemispherical photographs taken in early November in Broad-leaved Korean pine forest, Spruse-pir forest, Birch forest, Korean pine plantation and Larch plantation were underestimated by 52.1%,37%,16.2%,50%,46.1%, respectively. In July, Linear regressions between canopy LAI and true LAI were good in various forest types, regression coefficients were all above 0.5. On the whole, the understory LAI measured by hemispherical photographs were lower than harvest method, and the linear regressions between the two were good, the coefficients were all above 0.7 in July.After considering the gap fraction, the relationship between vegetation index and total LAI were increased, especially in such canopy which LAI was smaller and gaps were larger, because the influence of understory vegetation. This suggests that the reflectivity on images were both affected by reflectivity of understory leaves under the gaps and the canopy reflectivity. After analysis, the two polynomial model between LAI and NDVI in Korean pine plantation, and the linear regression models between LAI and SR in other four forest types were the best models, which can be applied to estimate LAI by remote sensing in this region.