INSERT TABLE 5 HERE
Out of some relevant approaches, on-site field observation and biophysical measurement are labour intensive, costly and time-consuming. Therefore, these tactics are likely applicable over small spatial areas (e.g. field scale). However, they are able to provide first-hand measurements and observation records on the dynamics and conditions on desertification, which can be used for validation of modelling results and remote sensing interpretation.
On-site field observations are also capable of detecting changes (e.g. soil properties, soil-water condition, etc.) induced by desertification that cannot easily be investigated via other methods (Zhang and Huisingh, 2018). Long-term on-site field observations are useful for investigations of desertification mechanisms and the response of desertification to different treatment/mitigation measures (e.g. Li et al., 2000, Su et al., 2010). It is important for desertification evaluation and control to establish permanent field experimental stations. In China, the “Chinese Desert Ecosystem Research Network ” that comprised with 43 National Research monitoring stations located across the N and NW China (Guo et al., 2014).
Desertification is often assessed via indirect indicators such as its driving factors (e.g. drought, anthropogenic activities), and impacts (e.g. vegetation cover change etc.) (Ajai et al., 2009, Šarapatka et al., 2010, Kust et al., 2011, Jiang et al., 2019a, Jiang et al., 2019b, Gul and Ersahin, 2019). Monitoring of drought/aridification is a common way to evaluate the risk of desertification in regions that have been suffering from drought like countries in Western Asia (Erian et al., 2011, Elagib and Abdu, 1997), in S Asia (Shahid and Behrawan, 2008), E Asia (Wang et al., 2017) and E Europe (Lupu et al., 2010, Mateescu et al., 2013). Many indicators of drought have been developed, such as “Aridity Index ” (Park et al., 2018), “Standardized Precipitation Index ” (SPI); “Standardized Precipitation-Evapotranspiration Index ” (SPEI) and “Palmer Drought Severity Index ” (PDSI) (Wang et al., 2017). Among these indicators, the SPI has been implemented in the Baribo Basin, Cambodia and showed that the extremely drought events occurred in 1987, 1993, 2001 and 2004 respectively (Sok et al., 2017).
Besides, the indicators of SPI, SPEI and PDSI were applied to investigate the development of drought in the dryness region of NW China between 1960 and 2010 (Wang et al., 2017). Furthermore, the composite index has been applied to assess the desertification risk in the BRI region, which considers multipole natural and anthropogenic factors and thus provides a more comprehensive evaluation of desertification (Hien et al., 2019). For example, the “Environmentally Sensitive Area Index” (ESAI), which is estimated as the geometric mean of four quality indexes on soil, climate, vegetation and management, has been used by (Jiang et al., 2019a) to assess the desertification risk of the C Asia.
Remote sensing ” is an effective tool for spotting and distinguishing indicators of various stages of desertification processes through capturing the changes caused by desertification (e.g. changes in vegetation coverage) (Zhang and Huisingh, 2018, Albalawi and Kumar, 2013, Wei et al., 2020). Remote sensing data, often freely accessible, have been applied in the BRI countries for the desertification evaluation (e.g. Li et al., 2013, Liu et al., 2004). Indices like the “Normalized Difference Vegetation Index (NDVI) ” (NDVI), “Enhanced Vegetation Index (EVI)”, “Fractional Vegetation Coverage ” (FVC), “Lead Area Index ” (LAI) derived from remote sensing datasets are commonly used indicators (Shalaby et al., 2004, Zhang and Huisingh, 2018). For example, the “Desertification Risk Index ” (DRI), derived from climate factors and NDVI, was developed by Feoli et al. (2000) for desertification evaluation in Turkey and has been successfully applied in the Middle East Lebanon (Dragan et al., 2005).
Remote Sensing technologies have been widely adopted to investigate and understand the changes of soil properties, for example, soil organic matter content (Kumar et al., 2017, Mirzaee et al., 2016) and soil texture (da Silva Chagas et al., 2016, Wu et al., 2015), further improving the capability of remote sensing in desertification detection. However, Remote Sensing technology is subject to inaccuracy. For example, the NDVI and other vegetation information-based methods often overestimate desertification in sparsely vegetated areas mainly owing to the unsteadiness of seasonal vegetation changes and the intensity of precipitation, and the rainstorm-caused severe effects (Dawelbait and Morari, 2008, Wessels et al., 2012, Wei et al., 2020).
Some models have been employed for evaluating the desertification in the BRI countries that are the soil erosion / desertification models such as the “Revised Universal Soil Loss Equation ” (RUSLE), “Mediterranean Desertification and Land Use” (MEDALUS) . TheRUSLE, developed by Renard (1997), estimates soil erosion rates driven by “Rill and Interill Erosion ” based on factors including, “Rainfall Erosivity ”, “Soil Erodibility ”, “Topography ”, “Vegetation Cover ” and “Erosion Control ”. That model has been popularly applied worldwide to assess soil erosion rates through the cascade of hillslopes study area (Borelli et al., 2017, Li et al., 2020). The model has also been adopted in the BRI countries, Myanmar (Tun et al., 2015), Thailand (Wijitkosum et al., 2013) and China (e.g. Sun et al., 2014, Fu et al., 2011, Li et al., 2020). Whilst, the MEDALUS model was developed during the “MEDALUS EU-funded project” (Kirkby et al., 1998) and can be effectively used to identify areas of desertification, in relating to various parameters (including “landforms, soil, geology, vegetation, climate and human actions ”). The MEDALUS model was developed for the Mediterranean climatic conditions, such as in Greece (Karamesouti et al., 2018). It has now been widely applied in the BRI countries outside the Mediterranean region such as Iran (Zehtabian et al., 2005, Sepehr et al., 2007) and Thailand (Wijitkosum, 2016).
Another type of model approach is via climate circulation models, which were applied in the BRI area. For example, the “National Centre for Ocean-Land-Atmosphere Studies ” developed the Atmospheric General Circulation Model (Xue, 1996), namely the “Community Atmospheric Model Version 3 ” (CAM 3) to assess climatic change and its possible impacts on desertification (Jin et al., 2012). The “NationalCentre for Ocean-Land-Atmosphere Studies ” also further adopted some other atmospheric general circulation models, including a biosphere model by Xue (1996), which tested the response of regional climate to the desertification in the Mongolian Plateau. Other modelling results found that the surface water and energy balance influenced by desertification. The CAM3 modelling results also demonstrated that dryness level in C Asia was closely related to declined wind velocity and decreased evaporation upstream from the Mediterranean Sea, Black Sea, and Caspian Sea in Boreal Winter (Jin et al., 2012). Whilst, the “Intergovernmental Panel on Climate Change ” (IPCC) models (Doutreloup et al., 2011, Miao et al., 2015) also reported that mean temperature in the Province of Binh Thuan, SE Vietnam was projected to rise by about 1.6 °C (over 2046-2065) and 2.5 °C (over 2081-2100) and the area is likely to suffer from extreme temperatures and extreme rainfall events (Doutreloup et al., 2011).
Most of contemporary models either simulate only one form of desertification (i.e. soil erosion) or evaluate desertification through prediction of climate change. Few quantitative models have been developed to directly assess the desertification status. To the authors’ knowledge, the MEDULAS is the only model that can be used to quantitatively assess the desertification condition. The model is applied under the empirical research perspectives, which constraints its capability of reproducing different desertification processes and the response of these processes to natural and socio-economic factors such as climate condition (precipitation, temperature), vegetation growth, land use, human population etc. (Khachatryan, 2013, Li et al., 2017, de Vente et al., 2013). Unfortunately, we have to understand that these tools and models are still having factual limitations. For example, they have an intensive input data that limits their application because of lacking sufficient data support (Li et al., 2017). Field investigations are appropriate for an exhaustive understanding on the processes and mechanisms of desertification over a small area that is beneficial for the development of desertification models. Indirect detection, unlike field investigations, assesses desertification situation based on relevant indicators, which, however, are sometimes not available. Remote sensing is suitable for detection of desertification indicators over a large area, given that remote sensing images have large spatial coverage. Their applicability is often restricted by the technical issues, such as the accuracy and resolution of the images (Albalawi and Kumar, 2013). Quantitative and mathematical models can be adopted to assess and/or predict desertification situation at various spatial scales. The comprehensive desertification model is still rare. Some models that incorporate the reciprocal relationships among natural and socio-economic factors and their influences on desertification are still desirable. Besides, the models often require massive input data, which limit their application in places without sufficient data support. Overall, the desertification evaluation methods are of their own characteristics and applicability (Table 5), and the selection of methods should be based on an evaluation of the nature of research and data availability.

6. Conclusion

In this article, we studied the desertification and its control in the countries associated with the BRI initiative proposed by China in 2013, via an extensive review of literature. We argued that the desertification has been widely spread in the BRI countries, particularly in the Central Asian countries, despite quantitative assessment and prediction of desertification is still limited. We also found that the most important factors of desertification by natural and anthropogenic processes are dry climate condition and mis-land-use/management respectively. Anthropogenic (human-induced) activities are a crucial driving force for land desertification in the BRI countries, while natural factors dominated in some dryland regions. However, the relative importance of natural and anthropogenic factors has yet been (quantitatively) assessed for most of the BRI countries.
Desertification in the BRI region has a series of negative impacts on socio-economy and eco-environment, mainly including loss of GDP, loss of productive land and regional climate. Most of the BRI countries have recognized the desertification issue and have ratified the UNCCD and worked out the NAPs. The comprehensive desertification mitigation/conservation programmes are still lacking. BRI countries are therefore encouraged to work together under trans-national or cross-boundary cooperation framework to combat desertification. Research methods employed for desertification evaluation include quantitative approaches (e.g. via mathematical models), indirect detection, and direct observation and biophysical measurement (e.g. measuring the tree and vegetation growth). They have their own characteristics and applicability, and the selection of methods should be based on an evaluation of the nature of research and data availability.
Overall, the desertification situation and their driving mechanisms should be thoroughly evaluated for the BRI region. The all-inclusive and rational conservation measures should be recognised and implemented, in prior to lessen the influence of climatic change and possible human interventions under the BRI framework.