Conference Agenda

Desertification is one of the most important environmental problems in
China's arid regions, and the damage is very serious. Soil organic matter (SOM)
content is an effective indicator for effectively reflecting desertification status. Large
area monitoring of organic matter content is of great significance for mastering the
status and dynamics of desertification and formulating scientific and effective
prevention strategies. However, due to data limitations, vegetation signal interference,
etc., large-area acquisition of SOM content has always faced great difficulties. The
purpose of this paper is to estimate the soil surface SOM content in desertified land in
northern China by using Sentitle-2 data mainly based on the Google Earth Engine
(GEE ) platform. Two machine learning methods, classification and regression tree
(CART) and support vector machine (SVM) respectively, were employed with
different input variable combinations. The results show that the proposed approach
could estimate SOM in desertified land effectively. CART shows higher accuracy than
SVM, especially when Sentinel-2 band reflectance and ancillary factors (vegetation
index, elevation, annual average temperature, etc.) were totally selected as input
variables, the model’s R²could reach up to 0.73, RMSE can reach 0.31. Also, the
approach based on GEE has the advantage of time efficiency, less than1 hour is
needed to finish the estimation of the SOM at the 10 m spatial resolution for the entire
desertified land in northern China, which has great potential for high spatial resolution
SOM mapping at the large scale.

Observations show that in recent decades, a large area of China has been affected by drought and that frequent droughts have caused damage to the ecological environment and the economy. Because of the differences in data and methods, assessing regional droughts often leads to contradictory conclusions. The self-calibrated Palmer drought severity index (scPDSI) is based on multiple parameters, such as precipitation, temperature and soil properties, and it is considered regionally applicable and is widely used. However, some divergence has been observed in the results of drought in China using different scPDSI_PM (scPDSI based on the Penman-Monteith model) datasets. We establish an integrated scPSDI dataset (scPDSI_PM _INT) by averaging three scPDSI products through the equal-weighted method and analyze the temporal change in and spatial characteristics of drought in China from 1950 to 2009. The annual and seasonal drought intensities have increased in the past 60 years. The disturbed area has broadened significantly, especially in eastern China, which has become much drier. The intensity of most drought-prone regions is abnormally dry and moderate, while severe and extreme droughts occur mostly in the agro-pastoral zone and the Beijing-Tianjin-Hebei region. The vegetation activity of response to regional drought was analyzed in this research.

The objective of Dragon 4 Project 32396_2 aims at detecting land degradation in dry lands at a regional scale. The main achievements acquired during the last years could be summarized as follows:

(1) Assessment and monitoring of land degradation in dry lands of China: T With the development of remote sensing technology, long time series remote sensing data have been available for land degradation assessment and monitoring, and the vegetation indicators, such as the NDVI, NPP, Vegetation coverage and biomass were commonly used. However, time series vegetation index will fluctuate severely due to the impact of climate change, especially the fluctuation of annual precipitation, thereby the land production capacity could not be determined accurately. Therefore, to solve the problem, Xilin Gol League, Inner Mongolia Autonomous Region, China, where the land degradation is prevailing in the first decade of the 21st century was selected as the study area. Based on the annual NPP dataset estimated by 10-Day composite NDVI from Envisat-Meris data at 1.2km resolution during 2003 to 2013 and the same period meteorological raster dataset, a new Moisture-responded Net Primary Productivity (MNPP) method, for identifying areas of land degradation based on the change of annual NPP and MNPP over time and Moisture Index (MI) was developed. It was expected that provide technical support and scientific reference data for land degradation assessment and monitoring in study area, even in the whole drylands in China.

(2) Estimating Soil Organic Carbon Density in the Otindag Sandy Land, Inner Mongolia, China. Accurate quantitative estimates of soil organic carbon density (SOCD) can effectively represent regional carbon cycle processes and regulation mechanisms, and can serve as reference data when making land management decisions. Limited research, however, has been carried out in arid or desert zones covered with sparse vegetation, despite the fact that these cover wide areas of the earth and play a significant role in global carbon cycles. In this study, the Otindag Sandy Land and its surroundings (OSLAIS) in the Inner Mongolia Autonomous Region of China was selected as the study area. The study introduces a useful technique for making high spatial coverage SOCD estimates for drylands by utilizing GF-1 WFV optical satellite images and a time series of MODIS satellite remote sensing datasets, and using these to optimize parameters for simulation models in conjunction with other technical procedures that are described. We expect this research to provide useful technical support and scientific reference data for land management and for land degradation/desertification assessments, for the study area monitored, as well as across the whole dryland area of China.

(3) Estimating Above Ground Biomass in the Otindag Sandy, Inner Mongolia, China by Using Sentinel-2 data. Above ground biomass (AGB) is an important measure of terrestrial ecosystem productivity, and it is used in quantifying the role of vegetation in the carbon cycle, the potential for energy production, and the carbon stock estimation for climate change modelling. Dryland AGB, also recommended as the indicator of land productivity by UNCCD in desertification assessment and monitoring, need to be quantitative assessed and evaluated. In this study, the Otindag Sandy Land and its surroundings (OSLAIS) was set as the study area and a useful method for sparse vegetation aboveground biomass inversion in dryland was promoted. Firstly, the Sentinel-2 remote sensing data coved the whole area in growing season (May to September) during 2015 to 2018 and synchronous field survey data was collected and processed. Then, the estimation model was constructed by linear regression model, power function model, exponential model and machine learning model by taking band information, texture information and different vegetation index into consideration. In addition, total 2/3 field survey sampled AGB data were used for modelling, and the remaining 1/3 measured AGB data was set as the testing to evaluate the AGB estimation models. Finally, the AGB distribution of the OSLAIS was mapped and analysed based on the optimal model. This research is expected to provide technical support and scientific reference data for vegetation assessment and monitoring in the study area, and even across the entire dryland of northern China.

Land condition results from land use in a managed territory. However, the reverse holds too: only a subset of land uses can be applied given a state of land condition. Such reciprocal feedbacks can be of utmost importance for assessing the land management options of a territory.

This was the main hypothesis of this study. To test it, we explored associations and dependencies between land condition states and land cover classes in the China drylands. More precisely, the study area was the Potential Extent of Desertification in China (PEDC), determined after applying the FAO-UNEP aridity index to an archive of climatic surfaces. The study period was 2002-2012. Land condition states resulted from the application of the 2dRUE method to an archived time-series of Net Primary Productivity (NPP), derived from MERIS satellite data by the CASA algorithm. Such states describe ecological maturity in terms of aboveground vegetation biomass and turnover, and lend well to an ordinal scaling. Land cover classes resulted from the aggregation of thirty-eight classes of level II built for China for the year 2010 following the Land Cover Classification System of the FAO. Land uses were excluded from this preliminary run. The spatial resolution for all the analyses was of 4 km.

We performed two statistical tests on the described data set, stratified by aridity zones. First, associations between land condition states and land cover classes were determined by chi-square tests, using the Monte Carlo method. Wherever significant associations were found between these variables, we interpreted the standardized residuals to determine the significance and sign of individual combinations of the corresponding contingency table. The second test was a non-parametric ANOVA with unequal samples, using the Kruskal-Wallis and Median test. We also determined homogeneous groups of land cover classes (in terms of land condition) through non-comprehensive search of land condition differences in pairwise combinations of them.

The associations between land condition and land cover resulted significant for all the dryland aridity zones. In general, areas of low vegetation cover such as desert or bare soil were positively associated with more degraded states, whilst higher vegetation cover was positively associated with states of higher maturity and complexity. As for the second test, it was significant too and two homogeneous groups of land cover could be formed for all the aridity levels.

The results, as reported in this abstract, are somewhat limited because of the exclusion of proper land uses. The land cover classes used here were only five (Deserts and bare soils, Grasslands, Shrubs, Open forests and Forests) and these are likely to be controlled in balanced terms by physical gradients and human intervention. Still, 2dRUE detects land condition states after a climate correction, and both mature and reference states have been found in real deserts of North Africa, for example. This suggests that significant associations in this study between class Deserts and land degradation involves human management to some extent. In other words, this class might be a mixture of proper zonal deserts and desertified areas that were possibly with a denser vegetation cover in earlier times.

The approach will be repeated using a full classification of land uses. Meanwhile, we can preliminarily conclude that it produces interpretable results that will help determining interconversion pathways between land cover classes, which in turn supports the paradigm that land degradation can be defined as loss of management options.