Remote estimation of crown size and tree density in snowy areas

Precise estimation of tree density in the forest leads us to understand the amount of carbon dioxide fixed by plants. Aerial photographs have been used to measure the number of trees. Campaign using aircraft, however, is expensive ( $50,000/1 campaign flight) and the research area is limited in drone. In addition, previous studies estimating the density of trees from aerial photographs have been performed in the summer, so there was a gap of 15% in the estimation due to the overlapping of the leaves. Here, we have proposed a method to accurately estimate the number of forest trees from the satellite images of snow-covered deciduous forest area, using the ratio of branches to snow. The advantages of our method are as follows; 1) snow area could be excluded easily due to the high reflectance, 2) tree branches are small overlapping compared to leaves. Although our method can use only in the snowfall region, the area covered with snow in the world becomes more than 12,800,000 km2. Our proposition should play an important role in discussing global warming. As a test area, we have chosen the forest near Mt. Amano in Iwate prefecture in Japan. First, we made a new index of (Band1-Band5)/(Band1+Band5), which will be suitable to distinguish between the snow and the tree trunk using the corresponding spectral reflection data. Next, the index values of changing the ratio in 1% increments were listed. From the satellite image analysis at 4 points, the ratio of snow to tree trunk showed the following values, I:61%, II:65%, III:66% and IV:65%. To confirm the estimation, we used the aerial photograph from Google earth; the rate was I:42.05%, II:48.89%, III:50.64%, IV:49.05%, respectively. There is a correlation between the numerical values of both, but there are differences. We will discuss in detail at this point, focusing on the effect of shadows.