Data for: How Much More Carbon Can Be Realistically Captured from Grassland Vegetation? Quantitative Assessment Using Focal Analysis on Soil-Topography-Vegetation Unit in the Inner Mongolia

Published: 18-10-2019| Version 1 | DOI: 10.17632/x8pdwgrykj.1
Contributors:
zongyao sha,
Y Bai,
Hai Lan,
Xuefeng Liu,
Yichun Xie

Description

The data provided in this work were processed from five original datasets, including 1) soil map from the Harmonized World Soil Database (ver. 1.2) (http://www.fao.org/soils-portal), 2) topography data (DEM from http://earthexplorer.usgs.gov), 3) Climate dataset (http://cdc.cma.gov.cn), 4) vegetation type map (from http://www.nsii.org.cn/chinavegetaion), and 5) MODIS net primary productivity (NPP) (MODIS 17A3, http://e4ftl01.cr.usgs.gov/MOLT), which were downloaded for the Inner Mongolia Autonomous Region (IMAR) of China. The time window for most of the datasets was during 2000-2014. Topography, after classifying the DEM into three groups (<500m, 500m~1500m, and >1500m), and monthly precipitation and temperature data collected from about 680 weather stations from the climate dataset were applied to make climate grid maps at 1 km×1 km using an ANUSPLIN approach (Price et al., 2000). The climate grid maps were used to model potential NPP (PNPP) using the Miami NPP model (Adams et al., 2004; Gang et al., 2014; Lieth, 1973). Tiles from MODIS 17A3 were mosaicked and taken to represent actual NPP (ANPP) of the grassland vegetation. A differential analysis between PNPP and ANPP at pixel level, was presented to represent a theoretical potential space in carbon capture. The maps of soil (S), topography (T), and vegetation (V) were overlaid to segment the area into spatially homogeneous S-T-V patches. Three types of focal statistics, including mean (Mean), maximum (Max), and 95% percentile threshold (95%PCT), of ANPP for each S-T-V unit were computed as the target level for ANPP. The gap from ANPP to each target level for each S-T-V patch was computed as being the practically realistic potential space. The gaps for the entire IMAR area were aggregated. The temporally averaged maps of the gaps derived from the pixel-based and focal analysis approaches along with ANPP and PNPP were provided. The temporal trajectories of spatially averaged gaps as well as ANPP and PNPP were illustrated.

Files

Steps to reproduce

All the raw files were downloaded from the websites, including 1) soil map from the Harmonized World Soil Database (ver. 1.2) (http://www.fao.org/soils-portal), 2) topography data (DEM from http://earthexplorer.usgs.gov), 3) Climate dataset (http://cdc.cma.gov.cn), 4) vegetation type map (from http://www.nsii.org.cn/chinavegetaion), and 5) MODIS net primary productivity (NPP) (MODIS 17A3, http://e4ftl01.cr.usgs.gov/MOLT), which were downloaded for the Inner Mongolia Autonomous Region (IMAR) of China. The time window for most of the datasets was during 2000-2014. DEM was classified into three groups, <500m, 500m~1500m, and >1500m. Fig. 1 outlines the location of the study area. The data comes from China’s administration boundary. Fig. 3d is the Soil-Topography-Vegetation patches from the three layer overlay, including soil map, topography map, and vegetation cover map. The download locations for the three maps are detailed in the abstract part. The three layers are overlaid to produce spatially homogeneous S-T-V patches using ArcMap. Figure 5a and 5b show ANPP that were processed from MODIS 17A3 (Tiles: h25v4, h25v3, h26v3, h26v4, h26v5, h25v5). The downloaded MOD17A3 tiles covering the study area were mosaicked and re-projected with the projection coordinate of Lambert Conformal Conic (LCC) on a yearly basis during 2000-2014 using MODIS MRT tool (Dwyer and Schmidt, 2006). Annual NPP data was extracted from the re-projected images, producing a 16-bit signed integer image dataset with the range maximum of 65,535. To convert the images to the same unit as that of PNPP (gC/m2‧y), a scale factor of 0.1 was applied. The 15 images were clipped using IMAR boundary as given in Fig. 1 Figure 5c and 5d are modelled result by the Miami NPP model (Adams et al., 2004; Gang et al., 2014). The model requires input of annual precipitation and annually averaged temperature. The climate input data was acquired from the China’s national weather data stations (http://cdc.cma.gov.cn). We obtained the two variables at monthly scale during the years 2000-2014. Those two climate variables were spatially interpolated as the grid maps at 1 km×1 km using an ANUSPLIN approach (Price et al., 2000). After that, annual data were aggregated from the monthly datasets. Figure 5e, 5f, and 5g demonstrate the comparison and dynamics of the ANPP and PNPP based on the annual ANPP and PNPP. The difference between PNPP and ANPP as well as the standard deviation of the difference for each year during 2000-2014 were computed using ArcMap map algebra function. Figure 6a was computed based on the annual ANPP and PNPP datasets. Figure 6b, 6c, and 6d are the carbon gaps processed by focal analysis, using each Soil-Topography-Vegetation patch. The dynamics in Figure 6e was processed from the gaps for each year during 2000-2014. MODIS MRT tool is used to masaic and reproject MODIS dataset. ArcGIS (ArcMap, ver. 10.3) was applied to do all the other analysis.