**1. Introduction**

Drought is considered an environmental disaster, and many researchers, including environmentalists, ecologists, hydrologists, meteorologists, geologists, and agricultural scientists have investigated droughts [1]. Drought causes soil degradation, desertification, water deficit, plant death, sandstorm, fire disaster, and other disaster phenomena [2]. Moreover, drought also affects crop growth, influences global food prices, and contributes to political unrest [3–6]. Therefore, monitoring drought and studying its spatiotemporal dynamics are important for improving agricultural production, protecting the environment, and promoting sustainable social economic development [7].

Traditional drought-monitoring methods are based on ground- or station-based meteorological and hydrological observations, such as precipitation, air temperature, soil moisture, evapotranspiration, and surface runo ff. A series of meteorological drought indices, including the Standardized Precipitation Index (SPI) and Palmer Drought Severity Index (PDSI), were developed based on these observation data. However, it is difficult to ensure the reliability of such interpolation because of the limited spatial density and uneven distribution of the observation stations [7,8]. Therefore, there is increased focus on remote sensing for drought monitoring because of its comprehensive, fast, and dynamic features that can rapidly and accurately yield multiscale and multitemporal information [9–13].

Many remote sensing-based drought indices have been established to reflect drought conditions. One of the most extensively used one is the normalized difference vegetation index (NDVI). However, when conducting drought monitoring over nonhomogeneous areas, NDVI is less reliable because of the effects of geographical location, ecological systems, and soil conditions [9,10]. To overcome these problems, Kogan proposed the vegetation condition index (VCI) by normalizing NDVI values to the maximum range of a specific area [9]. The weather-related NDVI component is smaller than the one related to the ecosystem; therefore, normalization successfully minimizes the ecosystem component. The VCI has been widely applied to drought monitoring and analysis, and its reliability has been verified by many studies [14–21]. VCI can individually monitor the effect of drought on vegetation health but is insufficient because it indicates only one moisture condition [14]. Considering that temperature may also reflect drought conditions to some extent, Kogan further developed the temperature condition index (TCI) by normalizing land surface temperature (LST) values to the maximum range of a specific area as an indicator of drought [10]. The vegetation health index (VHI), which averages the sum of VCI and TCI, too was introduced by Kogan [22]. The VHI has also been frequently used for agricultural purposes, such as crop yield estimation [18,19,23]. The principle of using VHI for drought monitoring is that an assessment of temperature conditions helps identify subtle changes in vegetation health because the effect of drought is more drastic if shortage of moisture is accompanied by excessive temperatures. The feasibility of using VHI has been validated in all major agricultural countries [22]. Precipitation deficit is an important condition for drought formation; therefore, the precipitation condition index (PCI) can reflect drought conditions [24]. Since drought usually is induced by precipitation deficit and rise in temperature and poses a threat to vegetation health, the scaled drought condition index (SDCI), which combines the PCI, TCI, and VCI, was proposed [25]. The abovementioned indices can be calculated from easily available satellite remote sensing data. Other researchers proposed the temperature–vegetation dryness index (TVDI) using the spatial relationship between the LST and NDVI based on the spectral reflectance of near-infrared (NIR) and red channels to indicate drought [26] and soil moisture conditions [27]. These indices take advantage of one or more aspects of droughts to reflect drought conditions; as a result, these indices have distinct characteristics that makes them suitable for different scenarios.

Remote sensing can provide long-term series data with broad spatial coverage; hence, these data are a perfect source for Earth observation and land surface monitoring. Drought-monitoring research also benefits greatly from remote sensing techniques, which help track the long-term trend of drought condition easily. Liang et al. used the VCI to evaluate the spatiotemporal variations of drought in different regions in China based on a trend analysis of tendency rate (slope) [7]. The occurrence of drought events in Northeast China from 2001 to 2014 was also investigated using slope analysis [6]. Spatial and temporal variations of drought in Nepal were examined by trend analysis based on satellite-derived VCI [28]. The Greater Changbai Mountains (GCM) is extremely important from the ecological viewpoint for the entire Northeast Asia as well as the world owing to the well preserved and most abundant forests of different types. Therefore, understanding the drought conditions of the GCM is critical for understanding the ecosystem conditions and environment of this region.

The main objectives of this study are (1) to evaluate the six widely used drought indices (PCI, VCI, TCI, VHI, SDCI, and TVDI) for drought monitoring in the GCM, considering that they reflect three different aspects of drought, namely precipitation, temperature, and vegetation conditions, and also that their data are easily available; (2) to explore the spatiotemporal patterns and the changing trend of drought in the GCM during the period 2001–2018; and (3) to analyze the correlations of the drought indices with meteorological factors and land cover types.

#### **2. Study Area and Data**
