**1. Introduction**

The Tibetan Plateau (TP), due to its unique high altitude, large topography and hollow heating effect, plays an important role in the modulation of Asian and even global atmospheric circulation [1,2]. In recent decades, the TP has been experiencing rapid warming and humidification characteristics, with the warming rate almost 1.5 times the global average value [3]. Thus, the TP is known as the "initiator" and "amplifier" of climate change. The source region of Three Rivers (SRTR) is located on the eastern TP and includes the source region of the Yangtze River, Yellow River and Lantsang River and is honored as the "Asia's Water Tower" [4]. Precipitation is one of the most important climatic factors affecting the ecological system and water resources over the SRTR [5]. Because it is located in the intersection area of the Indian monsoon, East Asian monsoon and westerly belt, the SRTR has a complex variety of climate types and inter-annual variation of precipitation [6]. Generally, the precipitation in the eastern and southern parts of the SRTR is significantly more than that in the northwest [7], and there is a complex coupling relationship between land surface processes and precipitation in different regions. Therefore, it is of grea<sup>t</sup> value for climate change, water resources research and ecological protection to study the characteristics and mechanisms of precipitation change in the SRTR.

The precipitation in the SRTR has been widely analyzed by using the in situ observation data. Besides the dominant pattern with high and low-value centers located in the

**Citation:** Meng, X.; Deng, M.; Liu, Y.; Li, Z.; Zhao, L. Remote Sensing-Detected Changes in Precipitation over the Source Region of Three Rivers in the Recent Two Decades. *Remote Sens.* **2022**, *14*, 2216. https://doi.org/10.3390/rs14092216

Academic Editor: Simone Lolli

Received: 12 April 2022 Accepted: 4 May 2022 Published: 5 May 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/).

southeast and northwest SRTR, respectively, a dipole pattern with southwest–northeast reverse distribution also exists in the SRTR [8]. From 1961 to 2019, the SRTR average annual precipitation was 470.7 mm and increased by 10.31 mm·10 a<sup>−</sup><sup>1</sup> [9], and the frequency of extreme events has increased [10]. The period 1971–1980 was the driest period since the year 1961, and 2001–2015 was the wettest period [11]. The trends in precipitation variation during spring, summer and autumn decreased from northwest to southeast, but the opposite trend was observed in winter [9]. Even in cold seasons, the precipitation has discordant trends in different months, with an increasing trend in November and February and a decreasing trend in other winter months [12].

In addition to meteorological station data, remote sensing products and reanalysis data have also been used to analyze precipitation in the SRTR. The Global Precipitation Climatology Project (GPCP) data are in agreemen<sup>t</sup> with the in situ measured precipitation [13]. The Integrated Multisatellite Retrievals for Global Precipitation Measurement (IMERG) products are affected by the temporal scale, precipitation intensity and phase, and the performance in the wet season is superior to that in the dry season [14]. Compared to the in situ observation data, the Climatic Research Unit (CRU) dataset underrated the annual precipitation but gave a similar variation characteristic in the SRTR [15]. On the same time scale, the consistency of NOAA Climate Prediction Center (CPC) products and Tropical Rainfall Measuring Mission (TRMM) products is better than that of the NOAA PERSIANN Precipitation Climate Data Record (PERSIANN-CDR) products [16]. Overall, remote sensing products have a higher ability to detect precipitation in high-altitude areas (>3000 m) than in low-altitude areas (<3000 m), and they have a better detection performance for light rain than moderate and heavy rain events [16].

Many studies have focused on the sources of water vapor and mechanisms of precipitation variation over the SRTR under different climates, but the results remain inconclusive. A study using the GPCP data suggests that the abnormal wind convergence and the lowpressure system, combined with the effects of the western Pacific subtropical high and the Mongolian high, provide conditions for the transport of water vapor and precipitation over the SRTR [13]. Another study suggests that Niño3.4, North Atlantic oscillation and Arctic oscillation play more important roles in the variation of dryness/wetness patterns in the SRTR [12]. In the cold season, the mechanisms for the interannual variation in precipitation are significantly different in different months. The main factors modulating the interannual variability of precipitation are the anomalous westerly water vapor transport (WVT) branch in November and southwesterly WVT anomalies in January and February [13].

In general, previous studies mostly focused on the analysis of long-term interannual or seasonal variations of SRTR precipitation and rarely discusses the changes in specific precipitation types (such as afternoon convective precipitation or nocturnal precipitation). A few studies have found that precipitation in the TP occurred mostly in the afternoon and night due to the thermal processes and the longwave radiation cooling [17,18]. In this study, we used the remote sensing precipitation product to diagnose precipitation changes in the SRTR in the last two decades. As previous studies emphasized an important influence of the hydrological cycle on local precipitation [19–21], we also present variation of afternoon precipitation as it is a dominant part of local triggered precipitation and is strongly related to the local thermal and hydrological processes. The paper is organized as follows. Section 2 introduces the study area and data used in this study. Section 3 presents the results. Section 4 is the discussion. Section 5 presents the conclusions.

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

### *2.1. Study Area*

The SRTR is located in the northeastern of the TP, with an average elevation of 3500 m; we mainly focused on the area of 30–37◦N and 88–104◦E in this study (Figure 1). Previous studies show that 38% of runoff in the source region of the Yellow River, 15% of runoff in the source region of the Lantsang River and a considerate amount of runoff in the Yangtze River originate from the SRTR [4,22]. The GSMaP\_Gauge is densely covered with rivers,

lakes, wetlands, snow-capped mountains and glaciers and thus is an important ecological shelter zone in China [23].

**Figure 1.** The overview of the Source Region of Three Rivers.

### *2.2. The GSMaP Precipitation Product*

GSMaP (Global Satellite Mapping of Precipitation) and IMERG (Integrated Multisatellite Retrievals for GPM) are two widely used satellite precipitation products in the GPM era, with high spatial and temporal resolutions. GSMaP (Global Satellite Mapping of Precipitation), developed by the Japan Aerospace Exploration Agency (JAXA) (https://sharaku.eorc.jaxa.jp/GSMaP\_CLM/index.htm, accessed on 10 May 2020), is one of the most popular algorithms in the era of GPM [24,25]. The GSMaP\_Guage product we used in this study is a gauge-calibrated product that adjusts the GSMaP\_MVK estimation with CPC (Climate Precipitation Center) gauge-based analysis of global daily precipitation, whose spatial and temporal resolutions are 0.1◦ × 0.1◦ and 1 h, respectively. Kentaro et al. (2015) compared GSMaP\_Gauge and GSMaP\_MVK products in Japan and found that GSMaP\_Gauge products have a better detection performance under different time scales and precipitation intensities [26]. Previous studies show that GSMaP gets some improvements in inversion accuracy and hydrological simulation utility compared to TRMM (Tropical Rainfall Measuring Mission) products over the Tibetan Plateau [27]. In the Yellow River basins. The latest GSMaP data is evaluated as having a relatively higher accuracy than IMERG [28].

### *2.3. ERA5 Reanalysis Data*

ERA5 is the fifth generation of global climate atmospheric reanalysis information from the Copernicus Climate Change Service (C3S) at the European Centre for Medium-range Weather Forecasts (ECMWF) (https://www.ecmwf.int/en/forecasts/dataset/ecmwf-reanalysis-v5, accessed on 30 August 2021), which uses an advanced modeling and data assimilation system to combine model data with observations from around the world to form a globally complete and consistent dataset. Compared to its predecessor, ERA5 has a finer horizontal grid of about 30 km while also improving vertical resolution and providing hourly estimates of a large number of atmospheric, terrestrial and oceanic climate variables [29–31]. Moreover, ERA5 effectively corrects for overestimating some physical quantities of thermodynamics and can be used for general analysis of the Tibetan Plateau [32,33].
