**1. Introduction**

Snow is one of the main forms of water in the cryosphere and is involved in most land surface energy and moisture transport [1–3]. It influences local and regional land–atmospheric processes and circulation [4,5] and is considered an important indicator of environmental changes at multiple scales [6,7]. Variations in snow and its phenology directly affect the formation of mountain discharge and the evolution of water resources in river source areas [8,9], influencing water utilization and supporting local society–economy–ecological sustainability in the middle and lower reaches [10,11]. Quantitative analysis can help better understand the responses of land surface hydrological systems to environmental changes [12–14].

Snow is very sensitive to climate change. According to the 6th Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), the current global air temperature

**Citation:** Wu, L.; Li, C.; Xie, X.; Lv, J.; Zou, S.; Zhou, X.; Shen, N. Land Surface Snow Phenology Based on an Improved Downscaling Method in the Southern Gansu Plateau, China. *Remote Sens.* **2022**, *14*, 2848. https:// doi.org/10.3390/rs14122848

Academic Editors: Massimo Menenti, Yaoming Ma, Li Jia and Lei Zhong

Received: 17 May 2022 Accepted: 11 June 2022 Published: 14 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

is approximately 1 ◦C higher than before industrialization. In terms of the predicted average temperature change in the next 20 years, the global temperature rise is expected to reach or exceed 1.5 ◦C [15,16]. Future warming may lead to abnormal precipitation and accelerated and earlier glacier and snow melts, which, in turn, will affect the distribution and dynamics of snow in time and space [17,18]. Studies have revealed that the spatiotemporal distribution of snow cover shows strong differentiation in China, and relatively stable snow areas are found mainly in northwestern and northeastern China, the Tibetan Plateau, and Inner Mongolia [19–21]. Land surface snow phenology (LSSP), such as snow cover start date, snow melt end date, and snow depth, is better correlated with temperature than other meteorological factors.

Underlying conditions, such as land surface topography and vegetation types, also affect the distribution and dynamics of snow [22,23]. For example, as the temperature gradually decreases with increasing altitude, snow melt slows, making it easier for snow to accumulate [24,25]. The absorption of solar radiation changes with different terrain conditions (i.e., slope and aspect), leading to diverse environmental temperatures and heating, consequentially influencing snowmelt processes [26,27]. In addition, relatively open areas, such as forest edges and sparse woodlands, are prone to snow accumulation, and the opposite is the case in well-covered woodlands due to canopy interception [28–30].

In recent decades, an increasing number of programs have been initiated internationally to facilitate snow research, such as the Climate and Cryosphere Project of the World Climate Research Program (WCRP), the Cold Region Land Surface Processes Experiment carried out by NASA, and the Western Environmental and Ecological Science Research Project, effectively advancing not only the study of snow and its dynamics as the key objects [31–33], but also techniques for snow monitoring and data derivation. In particular, the idea of using optical remote sensing to obtain snow information has made grea<sup>t</sup> progress [34–36], a large number of derivations as data products have been released, such as microwave radiometer-based (i.e., AMSR-E) and MODIS-based (i.e., MOD10A1 and MOD10A2) [37–39]. Among them, MODIS-based snow products have high spatial resolutions, can better reflect the distribution of snow cover, and are widely used in regional snow variation-related studies [40–42]. In contrast, passive microwave monitoring-based snow depth data are useful for equivalent evaluation, but their spatial resolution is generally low [43–45]. To obtain high-resolution snow depth information, downscaling of the data is needed. There are two common methods for this purpose: one is based on statistics, and the other is based on deep learning such as machine training [41,46]. The development of downscaling methods is important for snow studies, especially when conducted at smaller scales [47–49].

The southern Gansu Plateau (SGP), located on the northeast edge of the Tibetan Plateau, is an important water source area in the upper reaches of China's Yellow River and Yangtze River. Snow dynamics effectively influence runoff formation and evolution, and mechanistic exploration is beneficial to the scientific planning and utilization of basin water resources [5,50]. Over the past 20 years, river discharge on the SGP have sharply decreased, although few studies on snow phenology and its hydro effectiveness have been published. In view of the above, the objectives of this study are (1) to improve a downscaling method to obtain high-resolution snow depth (SD) data for the analysis of spatiotemporal variations in LSSP on the SGP during the time period from 2003–2018 and (2) to use a geostatistical method to analyze the effects of topographic and climatic factors on the LSSP. Related results may help improve our knowledge of alpine-cold region snow and can provide basic data and methodological support for comprehensive hydrological simulations and predictions in the water source area of large river basins.

### **2. Study Area**

As an important part of the water source area in the upper reaches of the Yellow River, the SGP administratively includes the whole of the Gannan Tibetan Autonomous Prefecture in Gansu Province of China, geographically located between 33◦06N–35◦34N, 100◦45E–104◦45E (Figure 1a). The elevation ranges from 1159~4866 m and averages approximately 3000 m, topographically featuring higher elevations in the northwest and lower elevations in the southeast. The regionally averaged annual air temperature is 1.7 ◦C, featuring a short frost-free period and plentiful sunshine throughout the year. The annual total precipitation is 620 mm, concentrated in the rainy season from June to September. The relatively lower air temperature and abundant precipitation, corresponding to a typically continental plateau climate, make the SGP naturally develop many tributary systems of the Yellow River (i.e., the Tao River, the Daxia River, etc.) and Yangtze River (i.e., the Bailong River), becoming remarkable in terms of water conservation on the Tibetan Plateau (Figure 1b,c). Along with the increasing intensity of human activities such as cultivation and overgrazing, ecosystems such as grasslands and wetlands become ecologically fragile, and water yield recharge to rivers are reduced, both seriously affecting the protection of regional water resources and ecological security. Due to the significance of snowmelt to SGP hydro-processes, the analysis of snow distribution and dynamics is important for the formation, evolution, rational development, and utilization of water on the SGP and across all the related basins.

**Figure 1.** Overview of the study area, including the geographic location on the northeast edge of the Tibetan Plateau (**a**), the water source area in the upper reaches of the Yellow River and Yangtze River (**b**), and the distribution of elevation, stream networks, and snow and hydrological observation stations (**c**).

### **3. Data and Methods**
