*2.1. Study Area*

The Jing-Jin-Ji region is located in the North China Plain, with the Bohai Sea to the east and the Taihang Mountains to the west. The altitude is higher to the northwest and lower to the southeast, with a total area of 185,000 km2. The area has a typical temperate monsoon climate that is characterized by rainy summers with high temperatures and cold and dry winters. The fastest population growth period in the Jing-Jin-Ji region is from 2000 to 2010 [46]. The Jing-Jin-Ji urban agglomeration is one of the three major urban agglomerations along the eastern coast of China due to active economic activities. The fast-growing population and rapid urbanization make it a hotspot for urbanization-related scientific research in China [47,48]. To select representative cities, Beijing (BJ), Tianjin (TJ), Shijiazhuang (SJZ), Handan (HD), Tangshan (TS), and Baoding (BD) are chosen, as they have populations greater than one million based on the 2012 China City Statistical Yearbook, in order to identify the differences in albedo for each city during urbanization. The study area is shown in Figure 1, and the GlobeLand30 data for 2000 and 2010 are used for the statistics in this area (Table 1). The results (Table 1) show that the largest decreasing areas were characterized by cultivated land ( −6.33%, from 2000 to 2010), while the largest growth areas were characterized by shrub lands (5.42%), followed by artificial land surfaces (1.31%).

**Figure 1.** Land cover distribution pattern extracted by GlobeLand30 in 2000 in the Jing-Jin-Ji region.

Besides, we made a statistic about the typical albedo value for each type of land cover based on the MODIS albedo products from its multiyear average value (Table 1).

Although the albedo mean values exhibit little difference between 2010 and 2000 due to the coarse resolution of albedo (compared to GlobeLand30), we can still see differences among land cover types. So, we just list the typical albedo values in 2010 in the table above, where shrublands have the largest albedo mean value (0.137), and artificial surfaces have a relatively small albedo value which is just bigger than wetland and water bodies.


**Table 1.** Typical albedo values of individual land covers in 2010, and the statistic percentage of each land cover type in 2000 and 2010. Mean is the spatial mean value of the multiyear average albedo in study areas, std is the standard deviation, and variation is the percentage difference between 2010 and 2000.

#### *2.2. Surface Albedo Data Set*

The MODIS 16-day 1 km albedo products (MCD43B3, collection5) from 2001 to 2011 were used in this study. The product contains the black-sky albedo (BSA) and the white-sky albedo (WSA), which can be used to calculate the actual (blue-sky) albedo based on the fraction of diffuse skylight [49,50]. Considering the small difference and high correlation between BSA and WSA [44,51–53], WSA was used as the index of albedo in this study. MODIS surface albedo products have been validated on a global scale, and the accuracy has been demonstrated to be suitable for studies on climate change [54,55]. We synthesized the 8-d intervals for albedo into yearly scales for the contribution analysis. The corresponding albedo quality data (MCD43B2, collection5) were also used to avoid the effects of snow cover.

## *2.3. Vegetation Index Data*

The monthly MODIS Enhanced Vegetation Index (EVI) product (MOD13A3) with a spatial resolution of 1 km was used. This product was generated based on atmosphere-corrected bidirectional surface reflectance, where the atmospheric effects of water, clouds, and aerosols were removed [56,57]. This product has been widely used in studies regarding global vegetation monitoring, land cover changes, and climate researches [58,59].

## *2.4. Nighttime Light Data*

Nighttime light signals detected by remote sensing satellites derive from the Defense Meteorological Satellite Program (DMSP), specifically from its visible and near infrared sensors named Operational Linescan System (OLS). DMSP/OLS nighttime light data are widely used in research regarding urban areas, such as estimating urban population [60], extracting urban extent [61], measuring urban expansion [62], and exploring human activities and its impacts on the environment in urban areas [43] etc. DMSP/OLS nighttime light data (Version 4) were used in this study, whose spatial resolution is 1 km. We excluded pixels without light (DN = 0) to ensure that there were human activities in every part of our study areas. In addition, an invariant target area method [63–65] for image correction was used to perform continuous and saturation corrections on the data. Using this method, we gained a nighttime light data time series with comparable DN values, which have been used to identify the urbanization [44].

### *2.5. GlobeLand30 Landcover Data*

GlobeLand30 is one of the global land cover map products at a 30-m resolution, which was produced with a pixel-object-knowledge (POK)-based operational mapping approach [66]. The classification system includes ten land cover types, namely cultivated lands, forests, shrublands, grasslands, wetlands, water bodies, tundra, artificial surfaces, permanent snow and ice, and barren lands for the years 2000 and 2010. The overall classification accuracy is over 80% [67], and it has been widely validated in many other researches [68,69].
