In recognition of the important role of forests in the global carbon cycle, particularly with respects to mitigating carbon dioxide emissions, the ability of accurately and precisely measure the carbon sequestration in forests is increasingly gaining global attention. As being a major part of the carbon cycle, accurate quantification of the forest above ground biomass (FAGB) and net primary productivity (NPP) at local to global scales has become one of a central topic for carbon cycle researchers, foresters, land and resource managers, and politicians. In order to estimate, FAGB and NPP adequately, methodologies, such as forest inventory, remote sensing, and vegetation and carbon modeling have been successfully utilized. However, insufficiency of direct field biomass and NPP observations has severely limited the parameterization, validation, and their estimation. If satellite-derived estimations become precise, efficient and reliable it will help to estimate and monitor forest carbon information in the global forest ecosystem and play a very important role in global climate change mitigation efforts. Despite so many efforts going on, there is still a lack of proper regional and national spatiotemporal FAGB and NPP information in developing countries like Nepal. In addition, there are not many, proper area/nation specific convenient method, for their estimation, that gives a quality estimate with low cost and good replicability, especially for the developing countries. Based on these research problems and research themes, to overcome them our objectives can be broadly classified as i) develop a two-scale FAGB estimation method with the use of multi-resolution optical imageries and Google Earth Very High Resolution (GEVHR) imageries as virtual sample plots, ii.) estimation and spatiotemporal change analysis of forest cover and FAGB in different physiographic regions and forest types of Nepal and iii.) estimation and spatiotemporal trend analysis of NPP in the forest of Nepal with Boreal Ecosystem Productivity Simulator (BEPS) model over years 2000-2015.
For the development of the two-scale method of FAGB estimation, the study was conducted in Chitwan district of Nepal using GeoEye-1 (0.46 m), Landsat (30m) and GEVHR Quick Bird (0.65m) imageries. For the local scale (Kayerkhola watershed), tree crowns of the entire area were delineated by object-based image analysis (OBIA) technique on GeoEye imageries. The overall accuracy of 83% was obtained in the delineation of tree canopy cover (TCC) plot-1. A TCC vs. FAGB model was developed based on TCC from GeoEye and FAGB from field sample plots. The coefficient of determination (R2) of 0.76 was obtained in themodeling and 0.83 in the validation of the model. To upscale FAGB to the whole district, open source GEVHR imageries were used as virtual field plots, delineated their TCC and then calculated it’s, FAGB (based on TCC vs. FAGB model). Using Multivariate Adaptive Regression Splines (MARS) machine learning algorithm, model was developed from Landsat 8 bands and vegetation indices. It was then used to extrapolate FAGB in the entire district. This approach considerably reduces field data and commercial very high resolution imageries requirements to achieve two scale forest information and FAGB estimate at high resolution (30m) and accuracy (R2=0.76 & 0.7) with minimal error (RMSE=64 & 38 tons ha-1) at both local and regional scales. The proposed methodology can be one of the promising techniques for the FAGB and carbon estimation in a very cost-efficient way and can be replicated with limited resources and time. It is especially applicable for developing countries with a low budget for carbon estimation and it is very much applicable to the “reducing emissions from deforestation and forest degradation” (REDD+) and “monitoring reporting and verification” (MRV) processes.
In the case of estimation and spatiotemporal analysis of forest cover and FAGB, here, we present so far first national scale forest cover type and FAGB study along with TCC of Nepal at 30m resolution for the year of 2000, 2010 and 2015. With the integrated used of Landsat imageries, field sample plots and Google earth imageries the forest cover type and FAGB of Nepal was estimated. A good overall accuracy of 87% with Kappa statistics of 0.89 was obtained for the forest cover type, classification with OBIA. Similarly, the estimation of FAGB with multiple linear regression was significant enough with aggregate R2 of 0.7 and RMSE 98 tons ha-1 at P<0.001for the year 2010. For the FAGB estimation for the year of 2000 and 2015, the FAGB vs. TCC model with aggregate R2 of 0.8 was used over the TCC estimated for them. The overall forest area of Nepal is found to be gradually increasing from 37.9% of the total area in 2000 to 40.2% in 2010 and 42.8% in 2015. Also, the FAGB was increased from 911million tons in 2000 to 1102 million tons in 2010 and 1109 million tons in 2015. The Broad-leaved closed forest (BLCF), was found to play major role, in terms of total FAGB contribution (47%, 47% and 50% of total contribution for years 2000, 2010 and 2015), forest area occupancy (37%, 36% and 38% of total forest for years 2000, 2010 and 2015), and FAGB productivity (214, 242 and 240 tons ha-1 for 2000, 2010 and 2015). In terms of physiographic region, the plain area was found to produce more FAGB (36%, 42% and 39% of the total FAGB for 2000, 2010 and 2015) although the forest area coverage by it was the least (29%, 30% and 27% of total forest for 2000, 2010 and2015) among the threephysiographic regions. The resulting nationwide wall-to-wall FAGB maps will help to improve the accuracy of carbon dynamic prediction in Nepal. It has huge importance to support diverse issues of environmental conservation. The data has big potential use for national and regional level sustainable land use planning strategies and meeting several global commitments.
In our study, we also aimed to understand the temporal and spatial variations of NPP in the forest of Nepal. The daily, monthly and annual NPP of the forest was estimated using the BEPS model for the years 2000-2015. The Leaf area index (LAI), meteorological datasets and other parameters as soil data, tree cover, biomass, field capacity, and wilting points were the main input for the BEPS model. We found that the NPP value varied spatially and temporally across the whole forest, which is increasing in general, though there were fluctuations in some years. The average daily NPP over the entire study period ranged from 1.3 to 1.7 gm m-2day-1 with highest in years 2014 and lowest in the year 2000 and overall average NPP trend of 1.65 gm m-2day-1. Within the overall forest, the average NPP productivity is generally highest in the plain followed by the hill and least in the mountain physiographical region. Looking at the intra-annual variability, the average monthly NPP ranged from 4.1 to 7.1 kg m-2month-1 with an average of 6.2 kg m-2month-1. The highest NPP rates were generally in the months of October and then May and the lowest in December and January. Mean seasonal NPP is largest during post-monsoon and lowest during the winter period, thereby indicating the importance of soil moisture and solar radiation for vegetation productivity. The average annual NPP is 1.2 kg m-2year-1 and the total average of 19 kg m-2. NPP was found to be highly influenced by LAI, rainfall, solar radiation and temperature mostly positively correlated in overall. The NPP was found to be highly correlated with the LAI especially in the plain region over the years from 2000-2015. In addition, while looking at the variation of NPP on different forest types, the broad-leaved forest was found to have almost 1.7 times more NPP (1.97 gm m-2day-1) than that of needle leaved forest (NPP 1.18 gm m-2day-1). We also found that the slope percent of <15% is more favorable (NPP 2.15 gm m-2day-1) for higher NPP among different slope percent. The result from this study gives us important information on intra and inter annual spatiotemporal trend and variability of NPP in the forest of Nepal overall and in different physiographic regions. It also gives us information on the relation between NPP and various climatic and vegetation parameters. All these information are very important for the proper forest ecosystem monitoring, management, and planning operations.
Keywords: Forest above ground biomass, Forest cover change, Net primary productivity, Spatio-temporal dynamics, Change analysis