Accuracy assessment of the global forest watch tree cover 2000 in China

Remote-sensing products have emerged as key tools in forest cover monitoring. Their quality vary spatially, local validations are recommended before using the data for inventory and management tasks. We conducted a validation based on a visual interpretation procedure using high-resolution optical imagery on Google Earth to map the uncertainties and inaccuracies of Global Forest Watch (GFW) Tree Cover 2000 in China. The article provides the reference dataset applied in Zhang et al. (2020). The reference data has a total amount of 96 364 sample pixels collected using spatially stratified random sampling method. The samples were labelled with land use classifications and can provide further usage for remote sensing products.

Assessing change in forest cover is of critical importance in studying natural and anthropogenic impacts on natural ecosystems. In particular, forest degradation accounts for almost 20% of anthropogenic greenhouse gas emissions (GHG) and thus is a significant driver of climate change, which, in turn can impact the health of the global ecosystem. Hence there has been a significant increase in international efforts such as the United Nations Program on Reducing Emissions from Deforestation and Forest Degradation (UN-REDD). In addition to politically negotiated treaties, a market based approach has been proposed in which corporations or countries that are significant emitters of atmospheric carbon offer monetary payments for forest preservation in exchange for carbon credits to be used in a carbon trading system. A key ingredient for effective forest management, whether for carbon trading or other purposes, is quantifiable knowledge about changes in forest cover. Rich amounts of data from remotely sensed images are now becoming available for detecting changes in forests or more generally, land cover. However, in spite of the importance of this problem and the considerable advances made over the last few years in high-resolution satellite data acquisition, data mining, and online mapping tools and services, end users still lack practical tools to help them manage and transform this data into actionable knowledge of changes in forest ecosystems that can be used for decision making and policy planning purposes. Providing this actionable knowledge requires innovations in a number of technical areas: (i) identification of changes in global forest cover, (ii) characterization of those changes, and (iii) discovery of relationships between the number, magnitude, and type of these changes with natural and anthropogenic variables. To realize progress in the above areas, a number of computational challenges in spatio-temporal data mining need to be addressed. Specifically, analysis and discovery approaches need to be cognizant of climate and ecosystem data characteristics such as seasonality, inter-region variability, multi-scale nature, spatio-temporal autocorrelation, high dimensionality and massive data size. This talk describes our initial efforts, achievements and challenges in addressing some of the above areas. 1 Research funded by Planetary Skin Institute, NSF, NASA, CISCO, and MN Futures Program. 7 © Atlantis Press, 2010

Quantification of spatial and temporal changes in forest cover is an essential component of forest monitoring programs. Due to its cloud free capability, Synthetic Aperture Radar (SAR) is an ideal source of information on forest dynamics in countries with near-constant cloud-cover. However, few studies have investigated the use of SAR for forest cover estimation in landscapes with highly sparse and fragmented forest cover. In this study, the potential use of L-band SAR for forest cover estimation in two regions (Longford and Sligo) in Ireland is investigated and compared to forest cover estimates derived from three national (Forestry2010, Prime2, National Forest Inventory), one pan-European (Forest Map 2006) and one global forest cover (Global Forest Change) product. Two machine-learning approaches (Random Forests and Extremely Randomised Trees) are evaluated. Both Random Forests and Extremely Randomised Trees classification accuracies were high (98.1–98.5%), with differences between the two classifiers being minimal (<0.5%). Increasing levels of post classification filtering led to a decrease in estimated forest area and an increase in overall accuracy of SAR-derived forest cover maps. All forest cover products were evaluated using an independent validation dataset. For the Longford region, the highest overall accuracy was recorded with the Forestry2010 dataset (97.42%) whereas in Sligo, highest overall accuracy was obtained for the Prime2 dataset (97.43%), although accuracies of SAR-derived forest maps were comparable. Our findings indicate that spaceborne radar could aid inventories in regions with low levels of forest cover in fragmented landscapes. The reduced accuracies observed for the global and pan-continental forest cover maps in comparison to national and SAR-derived forest maps indicate that caution should be exercised when applying these datasets for national reporting.

Forest cover dynamics (defined as tree canopy cover change without regard to forest land use) within the Russian European North have been analyzed from 1990 to 2005 using a combination of results from two Landsat-based forest cover monitoring projects: 1990–2000 and 2000–2005. Results of the forest cover dynamics analysis highlighted several trends in forest cover change since the breakdown of the Soviet planned economy. While total logging area decreased from the 1990–2000 to the 2000–2005 interval, logging and other forms of anthropogenically-induced clearing increased within the Central and Western parts of the region. The most populated regions of European Russia featured the highest rates of net forest cover loss. Our results also revealed intensive gross forest cover loss due to forest felling close to the Russian-Finland border. The annual burned forest area almost doubled between the two time intervals. The 2000–2005 gross forest cover gain results suggest that tree encroachment on abandoned agriculture land is a wide-spread process over the region. The analysis demonstrates the value of regional-scale Landsat-based forest cover and change quantification. Our results supplemented official data by providing independently derived spatial information that could be used for assessing on-going trends and serve as a baseline for future forest cover monitoring.

Detecting vulnerability of humid tropical forests to multiple stressors

Pollen map data at 2000-year intervals are used to reconstruct changes in Holocene forest cover in China north of the Yangtze River. In almost all regions, the early Holocene from 10 ka b.p. to 6 ka b.p. is characterised by an increase in forest cover although there was a reversal to lower forest cover at 8 ka b.p. in the eastern monsoon regions. A maximum of forest cover was reached at 6 ka b.p. in all regions except for Northeast China where forest cover peaked in the late Holocene. All regions except for Northeast China experienced a marked decline in forest cover after 6 ka b.p. Since 6 ka b.p., forest cover has decreased by about 92% in the middle and lower reaches of the Yellow River, 64% in the easternmost part of the Qinghai-Tibet Plateau and 37% between the Yangtze River and the Huaihe River. In contrast, forest cover in Northeast China increased significantly from 6 ka b.p., but has declined between 2 ka b.p. and the present. Changes in forest cover prior to 6 ka b.p. were probably caused by climate, but the evident drop in forest cover since that time in most regions may have been induced predominantly by human activities.

Long-term forest cover and height changes on abandoned agricultural land: An assessment based on historical stereometric images and airborne laser scanning data

Comparing annual MODIS and PRODES forest cover change data for advancing monitoring of Brazilian forest cover

In Europe, the change in forest cover caused by felling activities, especially clear-cutting, is the most significant driver of forest ecosystem change. Monitoring the magnitude of harvest activities and their spatial and temporal distribution can provide essential change indicators of the pressure on forest ecosystems.Satellite remote sensing offers the means to assess and map indicators of forest change dynamics related to a high spatial- and temporal resolution over large areas cost-effectively and objectively. Large-scale maps of forest cover change over time produced by LandTrendr (LT), a temporal segmentation algorithm, and Global Forest Watch (GFW) are evaluated in a Norwegian boreal environment. Their accuracy to detect change at stand-level and their potential to produce landscape-level spatial and temporal change indicators was assessed against a 20-year historical record of harvest activities in Southern Norway. Data from LT were found to be spatially and temporally more coherent with the reference data than GFW. LT detected 85.4% of the clear-cuts and a decrease in harvest activities between 2001 and 2017, a trend confirmed by the reference data, while GFW detected 63.1% of the clear-cuts and an increase in harvest activities for the same period. It was concluded that LT map of changes (LT-map) provides efficient spatial and temporal indicators of forest change dynamics and performed better than GFW tree cover loss map (GFW-map) to identify and monitor clear-cuts in a Norwegian boreal forest.

Deforestation and changes in forest cover are subject to the effects of many factors: economic, social, cultural, political, and environmental and their interactions. To assess these factors, one should necessarily know their nature and then model them. Nowadays many models have been introduced to explore deforestation, however most of them are based on past events and do not pay enough attention to socioeconomic factors. Moreover, considering the future is sometimes absent in those models. The present study develops an agent-based model for studying deforestation and changes in forest land cover using a case study. The model assesses deforestation in the past and then concentrates on the future of forest land cover. The study area is located in the north of Iran. The socioeconomic factors considered in this research as agents are farmers, ranchers and lumberjacks. The information and statistics of the agents were gathered over 10 years and so their manner in the forest was simulated. To validate the model, satellite imagery was used. The comparison between the simulated forest and the real one shows the power of the model as being 76% correct considering the Kappa coefficient. The results of forecasting reveal that the forest land cover will be reduced by 189 ha in 10 years time. Also the results show that in the study area the ranchers have more severe effects on deforestation than the beneficiaries.

Many studies have recently been devoted to the study of landscape change, and some have even focused on an analysis of the dynamics of forest cover change. However, few of the studies have worked on a methodology for making a detailed investigation of long-term forest change dynamics based on historic cartographic sources. The goal of this study is to further develop a method for analyzing long-term changes in forest cover on the basis of old maps and orthophoto maps in the GIS environment. The study area is located in Central Bohemia, to the east of Kutna Hora, a UNESCO World Heritage Site. The area consists of 21 cadastral units with a total area of 113 km2. The maps of the First (1780), Second (1851) and Third Military Surveys (1877) and the present-day orthophotomap (2007) of the Czech Republic were used as data resources. Source data have been processed in GIS. Forest cover is the subject of our study. However, the term is perceived from a broader perspective. What we call forest cover in our study refers to forest wood elements and other wood species in the landscape. In this study, forest cover has been structurally considered as a whole, without dividing it into the two categories mentioned. We counted the extent of the forest cover in each particular time horizon in hectares and as a percentage of the area under study, also the absolute changes in forest cover between the individual time horizons in hectares as well as the intensity of the changes in forest cover in hectares per year. The spatial changes in forest cover were evaluated in a GIS environment using specialized features to analyze spatial variation. The forest cover occupied 16.60% (1,880 ha) of the total area in the First Military Survey (1780). In 2007, the proportion was slightly higher at 16.64% (1,884 ha). More than half of all forest land (53%) from the time of the Second Military Survey (1851) survived until 2007. Not only the information on absolute changes but also the information on the rate of change is of great importance. The old Military Survey maps and the orthophotomap enable us to carry out studies of long-term changes in forest cover. However, the geodetic inaccuracy of the First Military Survey maps precludes reliable and exact quantification of the landscape changes between the First Military Survey and the Second Military Survey, and also between the First Military Survey and present-day (orthophoto map). These maps cannot be used for evaluating forest cover changes on the level of individual plots. The method presented in our paper may contribute to a better understanding of the long-term dynamics of forest land, covering a period of more than 250 years. This knowledge can be applied in forest management planning procedures. Apart from their application in forestry, the methods presented in this study may be of interest for historians and biologists.

Forests play an important role in the world ecosystem and their value in the environmental, economic and social aspects are uncountable. Human, as well as some natural factors are responsible for the decline in the forest cover and is adversely affecting the ecosystems. The monitoring of forest cover is necessary for the preventive measure for its conservation. Recently Remote Sensing (RS) and Geographic Information System (GIS) have emerged as an effective tool for the monitoring of forest cover changes. The present study focuses on the monitoring of forest deforestation in Mahabaleshwar, Maharashtra using remote sensing and GIS. The multispectral satellite data of 1977, 1995, 2014 and 2021 were used to assess spatial changes in forest cover. The results show the considerable loss of forest cover in the study area during the study period, accounting for 42.76 km2 between 1995 to 2014 and a substantial increase of 31.42 km2 between 2014 to 2021. The net change of forest cover over the 46 years was 17.64 km2. The area closer to agriculture and human settlement shows a high rate of forest degradation. Therefore, the study suggests that the afforestation initiatives by civilians, the Forest department and NGOs resulted in the increase in the forest cover in some parts of the study area.

Sub-Saharan Africa is well endowed with both renewable and non-renewable natural resources critical in supporting several forms of development on the continent. Key among these is natural forest resources. However, the population explosion in sub-Saharan Africa in general and Uganda, in particular, is threatening the survival of these forests due to the associated increasing demand for food, fodder, energy, and land for settlement. The study was conducted in Yumbe district where the forests considered included woodland and bushland since tropical high forests have been depleted or degraded by human activities. We used a predictive model to map future forest cover loss amidst the rapidly increasing population in Yumbe district in Uganda. Specifically, the study analyzed the relationship between population dynamics and forest cover change to predict future forest cover changes. To analyze changes in forest cover, the study utilized Landsat satellite imagery for 1990, 2000, 2010, and 2021; while the population data for the respective years was obtained from the Uganda Bureau of Statistics (UBOS). To explain the role of anthropogenic forces on forest cover change, the study considered different land use types as explanatory variables: planted forests, subsistence farmland, built-up areas, and other land use types. It then explored the interactions between these variables and forest cover change in the study area. Population-forest cover change model was developed to evaluate three decades of population and trends of forest cover to predict forest cover for 2032. The results indicate that in the three decades, the population increased by more than sixfold, and land area under subsistence agriculture, a proxy of population increased by 195.2%, but the forest cover declined by 80.3%. It is predicted that the forest cover will be lost completely by 2032 when the population reaches an estimated 838,078 from the current 657,430 people. This study, therefore, recommends that off-land employment opportunities such as tourism, apiary, transport, and manufacturing industries should be expanded in order to save forest resources from spatially extensive agricultural land uses. Key words: Forest, Forest cover loss, Predictive modeling, Population dynamics, Land use

Для оцінювання втрат лісового покриву Українських Карпат на прикладі території Сколівських Бескидів використано дистанційні методи. Для території досліджень на основі аналізу цифрових моделей рельєфу виокремлено ділянки, де, відповідно до чинних інструкцій та нормативів, заборонені суцільні рубки головного користування. На таких ділянках були виявлено та проаналізовано зміни лісового покриву. Для аналізу довгострокових змін лісового покриву використано Карту глобальних змін лісу (Global Forest Change – GFC). За даними аналізу такої інформації встановлено, що у 2010 р. частка природних лісів становила 19 % від загальної площі країни, або від 60,1 млн га. За період з 2001 по 2018 рр. в Україні втрачено 958 тис. га, що відповідає 8,6 % відносно площі лісового покриву за 2000 р. Для порівняння карт змін використано знімки із супутників Sentinel2 з роздільною здатністю 10 м×pix-1 для аналізу втрат лісу за 2015-2018 рр. Розмежування вододілу проведено для досліджуваної території за допомогою інструменту SAGA "Басейни вододілу" з використанням цифрової моделі рельєфу ASTER GDEM. За допомогою інструменту QGIS розраховано стрімкість схилів на основі цифрової моделі рельєфу ASTER GDEM2. Окрім цього, обчислено середнє значення, мінімум та максимум стрімкості схилу для порівняння її із наведеними даними стрімкості в базах лісовпорядкування для кожного виділу. Для визначення площі для екорегіону Українські Карпати на території Сколівських Бескидів спочатку вирізано растрову карту змін за даними Глобальної лісової варти (Global Forest Watch – GFW) за контурами екорегіону, векторизовано растр за картою змін, а потім обчислено площі за кожною категорією змін. Розраховано площі втрат лісового покриву. Встановлено, що вища частка втрат лісового покриву припадає на 2014-2018 рр. Він істотно вищий за середній щорічна частка втрат. Також виявлено, що останніми роками втрати лісового покриву зумовлені рубками, значна частка, котрих припадає на висоту понад 1100 м н.р.м. Аналіз змін лісового покриву для території Сколівських Бескид дав змогу порівняти такі зміни в лісах різної відомчої приналежності: Національного природного парку "Сколівські Бескиди", державного підприємства "Сколівське лісове господарство" та деяких лісництв, котрі належать до юрисдикції Сколівського війського лісгоспу ДП "Івано-Франківський військовий ліспромкомбінат". Порівняння даних втрати лісового покриву показав значні обсяги втрат на території військових лісництв, які були набагато вищими, ніж на інших територіях, що свідчить про їх антропогенне походження, тобто значні обсяги рубок.

Global, Landsat-based forest-cover change from 1990 to 2000