CHINA FOREST COVER EXTRACTION BASED ON GOOGLE EARTH ENGINE

Forest cover rate is the principal indice to reflect the forest acount of a nation and region. In view of the difficulty of accurately calculating large-scale forest area by traditional statistical survey methods, it is proposed to extract China forest area based on Google Earth Engine platform. Trained by the enough samples selected through the Google Earth software, there are nine different random forest classifiers applicable to their corresponding zones. Using Landsat 8 surface reflectance data of 2018 year and the modified forest partition map, China forest cover is generated on the Google Earth Engine platform. The accuracy of China's forest coverage achieves 89.08%, while the accuracy of Global Forest Change datasets of Maryland university and Japan’s ALOS Forest/Non-Forest forest product reach 87.78% and 84.57%. Besides, the precision of tropical/subtropical forest , temperate coniferous forest as well as nonforest region are 83.25% , 87.94% and 97.83%, higher than those of other’s accuracy. Our results show that by means of the random forest algorithm and enough samples, tropical and subtropical broadleaf forest, temperate coniferous forest and nonforest partition can be extracted more accurately. Through the computation of forest cover, our result shows that China has a area of 220.42 million hectare in 2018.

Among many earth systematical processings, vagetation land cover is the indispensible element.
Vegetation land cover is required by a number of general to be the boundary layer of execution model (Sellers et al. 1997). As a significant component of land cover research topics, forest cover detection is now more than ever becoming the focus of scientific research and resource management projects,such as investigating climate change, food security, habitat loss (Foley et al. 2005). The purpose of mapping large area forest is producing globally consistent characters possessing local relevance and practicability, in other words, cross-scale reliable information (Hansen et al. 2013). Due to the significance of forest cover data, countries in the world and international research institutes conduct a series of investigations on the topic of different scale land cover mapping.
Forest detection already raises wide concern of international society and achives s series of results.
Recently, remote sensing satellite data reveal a greening pattern that is strikingly prominent in China and India and overlaps with croplands world-wide and China alone accounts for 25% of the global net increase in leaf area with only 6.6% of global vegetated area . Meanwhile, mangrove forests along the coastal zones in China were mapped by integration of the GEE platform, time series Landsat and Sentine-1A SAR images (Chen et al. 2017).
Besides, PALSAR-based forest map in China demonstrate the potential of integrating PALSAR and MODIS images to map forests in large areas (Qin et al. 2015). On the other hand, some novel approachs were proposed to produce more accurate 25m forest maps by Comparing above products, there are some problem existing on the data processing and reprocessing, or the precision to be improved. Therefore, how to produce big scale forest maps efficiently and precisely is a puzzle to be solved.
The Google Earth Engine based on cloud compute platform combines the high-performance abilities with large=scale geographic data processing missions. This solution settles a train of major information technology challenges,such as data acquisition and storage, file pattern analysis, database management and equipment distribution (Gorelick et al. 2017).
In this study, we produce China forest cover maps of different partitions in 2018 using Google Earth Engine for data acquisition and operation platform. This forest distribution product is made from Landsat image data and random forest classification method. To guarantee the accuracy of this map, this study compares the forest map with Global Forest Change data and Forest/Non-Forest data.

2.CHINA FOREST PARTITION
Global land covers are usually divided into fourteen biocoenosis and eight geographic zones and China has eight biocoenosis (Olson and Dinerstein 2002;Olson et al. 2001

Google Earth Engine (GEE) contains a a range of
Landsat image collections, among which is Landsat-8 Surface Reflectance Tier and it comes from Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS).
In this paper, the USGS Landsat-8 Surface Reflectance Tier datasets are used for import data, which is provied by GEE. These data have been atmospherically corrected using Landsat Surface Reflectance Code

Index computation
In order to avoid the cloud and shadow influence, quality assessment band is used for mask band to reject the cloud pixels. After that operation, six spectral indexes are computed to act as character indexes for different land cover species. These indexes include To ensure only clearly forest pixels were selected, the  Classification results are hard voting of three labels. This is the overall consequences of all decision trees in the classifier. In the other hand, random forest classifier can also export the probability of each category. This results act as the confidence level output of every category,which contain the confidence index ranging from 0 to 1.

Classification results
In view of area forest partion, different number of validation points were produced for every study region.
After that, extract the real categories of points based on (2)