*3.2. Methods*

#### 3.2.1. Water Extraction Methods

The NDWI (normalized difference water index) proposed by McFeeters was initially designed to enhance the difference between water and nonwater bodies, in order to generate initial water body maps. However, it cannot efficiently suppress the signal noise coming from the land cover features of built-up areas [28,29]. Hence, Xu [30] proposed a modified NDWI (MNDWI), which is defined as:

$$\text{MNDWI} = \frac{(\text{Green} - \text{MIR})}{(\text{Green} + \text{MIR})} \tag{1}$$

where Green is the green band, which corresponds to the TM/ETM + image band 2 (0.52–0.60 μm) and the OLI image band 3 (0.525–0.600 μm). MIR represents the mid-infrared band, which corresponds to the TM/ETM + image band 5 (1.55–1.75 μm) and OLI image band 6 (1.560–1.660 μm) [30,31].

Water bodies have positive values in the MNDWI, while soil, vegetation, and built-up classes tend to have negative values. Hence, the normal empirical threshold is zero [29].

#### 3.2.2. Method for FVC Estimation

FVC (fractional vegetation cover) is defined as the fraction of green vegetation seen from nadir, which can characterize the growth conditions and horizontal density of land surface live vegetation. There are three major algorithms for FVC estimation, which include empirical methods, pixel unmixing models, and machine learning methods [32]. The dimidiate pixel model, which was used in this study, is a method of pixel unmixing and widely used for FVC estimation [33]. It is assumes the NDVI (normalized difference vegetation index) in a pixel consists of soil and vegetation, and the NDVI can be derived by [34,35]:

$$\text{NDVI} = \text{FVC} \times \text{NDVI}\_{\text{Vve\%}} + (1 - \text{FVC}) \times \text{NDVI}\_{\text{soil}} \tag{2}$$

where NDVIveg represents the NDVI value of a pure vegetation pixel and NDVIsoil represents the NDVI value of a pure soil pixel. The NDVIsoil selected an NDVI value with a cumulative frequency of 0.5%, and the NDVIveg selected an NDVI value with a cumulative frequency of 99.5% [36,37].

Hence, the FVC can be derived by modifying Equation (2) as:

$$\text{FVC} = \frac{(\text{NDVI} - \text{NDVI}\_{\text{soil}})}{(\text{NDVI}\_{\text{veg}} - \text{NDVI}\_{\text{soil}})} \tag{3}$$

#### 3.2.3. Classification Method of FVC

For the entire Minqin Basin, there are few vegetation areas and vegetation types. The vegetation coverage types of the Minqin Basin can be divided into five main vegetation coverage categories [38]: extremely low, low, medium, medium and high, and high (Table 1).


**Table 1.** Vegetation Coverage Classification Standards [38].
