**4. Results**

#### *4.1. Albedo Variations and Spatial Patterns*

The average albedo of the Jing-Jin-Ji region from 2001 to 2011 was 0.12 ± 0.02 and the spatial distribution of the multiyear mean albedo is shown in Figure 2. According to this spatial pattern, the albedo increased from the Central Business District (CBD) to the suburbs, and a majority of the Core Area had the lowest mean value of albedo compared to that in other areas. Based on the shift linear regression method [42], the results of the breakpoint detection showed that 2005 was the breakpoint year for the ~10 years of albedo data. With the Chow test, the *p*-value of the breakpoint detection was 0.002, which was substantially less than 0.05 and significant. The trend in albedo from 2001–2005 (T1 period) was the highest (6.5 × 10−<sup>3</sup> year<sup>−</sup>1, *p* < 0.05), with a lower trend of 1.2 × 10−<sup>3</sup> year<sup>−</sup><sup>1</sup> (*p* > 0.05) from 2006–2011 (T2 period), whereas the lowest trend occurred from 2001–2011 (T3 period; 0.78 × 10−<sup>3</sup> year<sup>−</sup><sup>1</sup> at *p* > 0.05). These trends show big differences in the T1 and T2 periods. The albedo trend during T2 was approximately 1/5 the trend during the T1 period, which led to a reduction in albedo of approximately 0.05. This indicates that the long-term growth trend of albedo was suppressed after 2005 due to several influential factors.

**Figure 2.** Spatialtemporal variation of albedo in the Jing-Jin-Ji region from 2001 to 2011. (**a**) Spatial distribution of the multiyear average albedo, and (**b**) the yearly average albedo and its trends in 2001–2005 (red line), in 2006–2011 (blue line), and in 2001–2011 (orange line) since 2005 is the breakpoint year. *p* stands for p-value, which is gained from an F test in a simple linear regression model.

To identify the factors contributing to this difference, we explored the spatial patterns of albedo before and after the breakpoint (Figure 3).

From 2001–2005 (Figure 3a), the albedo showed an increasing trend (k > 0) in over 99.5% of the whole region, while the number of pixels with a decreasing trend (k < 0) in albedo was small and accounted for only 0.5% of the total area. The percentage of the area where the albedo had significant trends (*p* < 0.05) is 13.2%, in which the percentage of the significant increasing trend (*p* < 0.05) was 13.13% and the significant decreasing trend (*p* < 0.05) was 0.07%. Albedo showed a general increasing trend in the study area. Spatially, from southwest to northeast, the interannual variation rate of albedo gradually decreased. Pixels of decreasing trends were mainly distributed across the northern fringe areas and along east coast areas near Bohai Bay.

**Figure 3.** Spatial distribution of the annual change rate of albedo in (**a**) 2001–2005, (**b**) 2006–2011, (**c**) and their relative frequency in different slopes range of the study area.

From 2006–2011 (Figure 3b), the albedo generally had a lower increasing trend than that in 2001–2005. The percentage of the area of increasing albedo trends decreased to 94%. In addition, the area of decreasing albedo trends increased, which accounted for 6% of the total area, distributed in the northern and eastern coastal areas. The area of significant albedo trends (*p* < 0.05) was 11.11%, in which the area of increasing trends (*p* < 0.05) took up 10.44% of the whole area and the area of decreasing trends took up 0.67%.

## *4.2. Urbanization Spatial Patterns*

In this study, we use the interannual variation rate of the DN values from corrected DMSP/OLS nighttime light data to define the urbanization rate (Figure 4) and study the developmental characteristics in the Jing-Jin-Ji region during T1 and T2 periods.

**Figure 4.** The spatial pattern of urbanization rate. (**a**) Represents the urbanization rate in T1 (2001–2005), and (**b**) represents the T2 (2006–2011).

From 2001 to 2005 (Figure 4a), the urbanization rate in most parts of the Jing-Jin-Ji region was less than 2 year<sup>−</sup>1, and the urbanization rate in only a few areas of the Expanded Area surrounding Beijing, Tianjin, and Tangshan was greater than 5 year<sup>−</sup>1. After 2005 (Figure 4b), although the rate of urbanization in the urban Core Area almost decreased to 0, the rate of urbanization in other areas increased significantly, and the urbanization area increased significantly, indicating that there is rapid urbanization occurred after 3 April 2005.

#### *4.3. Sensitivity of Urbanization and Vegetation to Albedo*

Numerous studies have shown that vegetation and urbanization are two major factors that cause changes in surface albedo [21,75]. For vegetation, different types of vegetation have different levels of albedo. For example, forests usually have a lower albedo (0.05–0.2) while grasslands have a higher albedo (0.16–0.26) [76]. In addition, changes in surface roughness caused by vegetation growth are also responsible for changes in albedo. In a similar way, surface roughness also changed with the process of urbanization. Land cover changes in the process of urbanization play a decisive role in the properties of three-dimensional surfaces in urban areas, which equally has a decisive influence on albedo in urban areas. The three-dimensional surfaces formed by buildings and roads etc. create large inner spaces and cracks for lights to transfer, which result in the multiple reflection of lights, trapping lights, and leading to a decrease in albedo. Considering that the units of vegetation data and nighttime light data are not uniform, and the range of values for these data is quite different, the corrected DMSP/OLS data in this study has been normalized. The sensitivity of albedo to vegetation and urbanization intensity is calculated by multiple linear regression. The spatial distribution patterns of each factor's sensitivity term during different periods, as well as their variations, are shown in Figure 5. The sensitivity of albedo to vegetation and urbanization displays differences in period T1 and period T2.

**Figure 5.** Spatial pattern of albedo sensitivity to vegetation and urbanization. T1 means the period from 2001 to 2005, T2 means 2006 to 2011, ΔT represents difference between T2 and T1, S(V) represents sensitivity of albedo to vegetation ( *∂*A *∂*V ) in which A denotes albedo and V denotes EVI, S(U) represents the sensitivity of albedo to urbanization ( *∂*A *∂*U ) in which A denotes albedo and U denotes DMSP/OLS, and S(V), S(U) were calculated via multiple regression of albedo to EVI and DMSP/OLS.

In period T1 (2001–2005), the sensitivity of albedo to urbanization showed a significant spatial distribution difference, of which the fifth percentile was −0.006 and the 95th percentile was 1.95. Spatially, the relatively high positive S(U) is mainly concentrated in the surrounding areas of major cities, such as Beijing, Tianjin, and Tangshan, positive-correlated to albedo. In contrast, S(U) in other regions is much smaller and generally negative, which has a weak negative correlation with albedo. In these regions, the sensitivity of albedo to vegetation(S(V)) shows high positive sensitivity, especially in the southeastern plains. In summary, urbanization has stronger promoting effects on albedo around large cities and has much weaker suppressing effects on albedo in other regions, where vegetation plays a promoting role in these areas.

In period T2 (2006–2011), instead of concentrating in areas surrounding large cities, the sensitivity of albedo to urbanization shows a regional diffusion feature compared to that in T1. In total, 72% of the S(U) is positive, and the fifth percentile of S(U) is −0.24 and the 95th percentile is 0.73. As the sensitivity of urbanization increases over a large area, the sensitivity of vegetation decreases significantly and extensively. In total, 53% of the vegetation sensitivity(S(V)) shows negative effects and mainly ranges from −0.4 to 0 (i.e., vegetation tends to inhibit the increase in albedo during this period).

From T1 to T2, the sensitivity of albedo to vegetation generally decreases and changes from positive to negative. The effect that this change causes is that the strong increased effect of vegetation on albedo turns into a weak decreased effect. However, the sensitivity of albedo to urbanization has extensively increased. The increased effects of urbanization on albedo exist not only in areas surrounding large cities during T1, but also in other large areas during T2, although the sensitivity intensity is much larger in T1 than that in T2.

#### *4.4. Effects of the Influential Factors on Changes in Albedo*

The sensitivity of albedo to vegetation and urbanization indicates a correlation between albedo and various factors. Because it is dimensionless, this correlation does not quantify the effects of various factors on albedo. Therefore, our study also quantifies the effects of vegetation and urbanization on albedo based on sensitivities. The effects of each factor on the interannual variation rate of albedo are shown in Figure 6.

From 2001–2005, the area with positive effects of vegetation on the interannual variation rate of albedo comprised more than 80% of the entire region (Figure 6). The areas with negative effects of vegetation mostly existed in the surrounding areas of cities and partially in the northern area of the study region. Urbanization had extensive positive effects on albedo, which were distributed around large cities. Shijiazhuang, Handan, and their surrounding areas had greater positive effects on variations in albedo compared to those from Beijing, Tianjin, and Tangshan, with the greatest effects exceeding 0.01 year<sup>−</sup>1. The effects in other regions were almost 0. Other factors had both positive and negative effects on variations in albedo, but most of these effects were positive and located in Shijiazhuang, Handan, and their surrounding areas, with effects greater than 0.01 year<sup>−</sup>1. In the northern part of the region, the effects of other factors on albedo were almost in the range of −0.005 to 0.005 year<sup>−</sup>1. The statistics of the relative contributions of each factor (Figure 7) show that from 2001–2005, the relative contribution percentage of vegetation, urbanization, and other factors was 44%, 15%, and 41%, respectively. Urbanization had the lowest contribution to regional albedo, whereas vegetation and other factors were the two main controlling factors in variations in albedo, which were both greater than two times the amount of contribution from urbanization. One thing that must be explained is that due to the type and distribution differences of each factor, the spatial heterogeneity was relatively obvious. As a result, some pixels were under the absolute control of vegetation, and some were under the absolute control of urbanization, which led to an expected large standard deviation (STD) value for each factor's contribution.

**Figure 6.** Spatial distribution and statistics of the effects of vegetation, urbanization, and other factors on interannual variation in albedo in T1 (2001–2005) and T2 (2006–2011). C(V) represents effects of vegetation on albedo, C(U) represents the effects of urbanization on albedo, C(Δ) represents effects of other factors on albedo, and the statistics of the relative percentage of vegetation, urbanization, and other factors in T1 and T2 are calculated based on the study area.

From 2006–2011, the effect of vegetation on variations in albedo mainly ranged from −0.005 year<sup>−</sup><sup>1</sup> to 0.005 year<sup>−</sup>1, which was generally lower than that from 2001–2005 (Figure 6). Over 99% of the urbanization effects on albedo were positive. The areas effected by urbanization expanded although the value of this effect decreased compared to the urbanization effect from 2001–2005. The effect of other factors was distributed uniformly across the study area, ranging from −0.005 to 0.005. From 2006–2011, the relative contribution percentages of vegetation, urbanization, and other factors to albedo (Figure 7) were 24%, 48.5%, and 27.5%, respectively. Urbanization became the highest contribution factor, which increased by more than 200%. In contrast, the contributions from vegetation and other factors decreased by 20% and 13.5%, respectively.

**Figure 7.** Average contribution percentages of vegetation (P(V)), urbanization (P(U)), and other factors (P(Δ)) to the interannual variation of albedo in the study area in T1 (2001–2005), in T2 (2006–2011), and in T3 (2001–2011). Std represents the standard deviation.

The relative contribution percentages of vegetation, urbanization, and other factors to albedo are different not only in size, but also in spatial distribution pattern (Figure 8).

Although the relative contribution percentage of each factor for the whole region is approximately 30% from 2001 to 2011 (Figure 7), the dominant controlling factors are different in different regions (Figure 8). From 2001–2011, the variations in albedo were mainly controlled by vegetation in the southeast and parts of the eastern region. Other regions were mainly controlled by urbanization, especially in regions surrounding cities, except for the Core Area, which was largely affected by both vegetation and other factors. From 2001–2005, the locations where the urbanization contribution was greater than 60% were the Expanded Area and the Fringe Area, while other regions were mostly controlled by vegetation. Other factors playing dominate roles were distributed in the triangular region formed by Beijing, Tianjin, and Tangshan, as well as near the connecting line between Shijiazhuang and Handan. From 2006 to 2011, the contributions of vegetation and other factors showed a significant reduction. The contribution percentage of vegetation was generally lower than 20%, whereas the contribution percentage of urbanization increased substantially in other regions (generally greater than 60%), except for the Core Area, where the urbanization contribution percentage was equal to zero. Other factors mainly affected the changes in albedo in the Core Area and a partial region near the connecting line between Shijiazhuang and Handan.

#### *4.5. Urbanization in Representative Cities*

According to the above analysis, urbanization transformed from a secondary influential factor from 2001–2005 into a major influential factor from 2006–2011, indicating that the effect of urbanization on regional albedo has increased since the breakpoint year (2005). However, the urbanization intensity of each individual city is substantially different. The Core Are, Expanded Area, and Fringe Area in our study represent the initial stage, middle acceleration stage, and final stage of the urbanization process, respectively; why is albedo different during these different urbanization stages? What are the main controlling factors for these areas? Are there any regional differences among the impact factors? We still do not know much about these issues. Therefore, in this paper, we also calculate the relative

contribution percentages of vegetation, urbanization, and other factors in different functional areas (Figure 9).

**Figure 8.** Spatial patterns of relative contribution percentages for vegetation (P(V)), urbanization (P(U)), and other factors (P(Δ)) in the T1, T2, and T3 period. T1 means the period from 2001 to 2005, T2 means 2006 to 2011, and T3 means 2001 to 2011.

For each individual city, the variations in albedo in Core Areas are mostly affected by other factors (Δ), followed by vegetation, and the contribution of urbanization is minimal. The average contribution percentages of vegetation, urbanization, and other factors are 34.5%, 8.4%, and 57.1%, respectively, in the Core Area. This result indicates that because the Core Area is generally composed of old cities that have generally completed urbanization, the influence of human activities on the variations in albedo is basically at a stable level; therefore, the urbanization contribution to albedo (8.4%) is much smaller than the contributions from other factors and vegetation. In the Core Area of the six major cities, other factors contribute more during T2 than those during T1 in 84% of our cities. Vegetation contributes more during T2 than that during T1 in 67% of our cities. The urbanization contribution decreases during T2 compared to that during T1 in 100% of our cities. It is safe to say that the contributions from vegetation and other factors in the Core Area will increase with time, while the contribution from urbanization will decrease.

**Figure 9.** Relative contribution percentages of vegetation (P(V)), urbanization (P(U)), and other factors (P(Δ)) in different functional areas (Core Area, Expanded Area, and Fringe Area) of cities in T1 (2001–2005), and T2 (2006–2011). BJ, TJ, SJZ, HD, TS, and BD represent Beijing, Tianjin, Shijiazhuang, Handan, Tangshan, and Baoding, respectively.

In the Expanded Area, the average contribution percentages of vegetation, urbanization, and other factors were 26.5%, 46.7%, and 26.8%, respectively, and urbanization was the most important contribution factor in the Expanded Area. Spatially, compared to the Core Area, urbanization had the greatest change in the Expanded Area, with an increase of 456%; the contributions of other factors and vegetation decreased by 53% and 23%, respectively. Temporally, the urbanization contribution percentage decreased to various degrees during T2 compared with that during the T1 period. Meanwhile, the vegetation contribution percentage increased in 67% of the cities, while the contribution percentage of other factors decreased. However, Beijing and Tianjin, whose populations are greater than five million, showed the opposite effect; that is, the vegetation contribution percentage decreased slightly in the Expanded Area, while the contribution of other factors increased significantly. For small cities, urbanization accompanying land cover change might have been finished in the Expanded Area and there would be no land cover changes for several years at least in the future, due to the limited population and limited population increasing ability. So, the number of the residents living in this area is relatively stable, which could lead to the stable growth of the vegetation and green space for a comfortable living environment. Thus, the contribution of vegetation will increase. On the contrary, for Beijing and Tianjin, which are the two biggest cities in the Jing-Jin-Ji area, they have been maintaining a fast-speed of urbanization for a long time. With reference to the Globeland30 land cover maps, we know that land cover changes are still happening, and they happened in the Expanded Area. That is to say, the urbanization process is going on in Expanded Areas in Beijing and Tianjin. So, due to the limited urban space, the rate of green space to urban space in Beijing and Tianjin would be smaller than small cities. Thus, the contribution of vegetation in Expanded Area in these big cities is relatively small compare to those in small cities.

In the Fringe Area, the average contribution percentages of vegetation, urbanization, and other factors were 24.9%, 45.8%, and 29.3%, respectively. Compared to the Expanded Area, the urbanization contribution percentage in the Fringe Area decreased, but it was still the dominant contribution factor. However, the vegetation contribution percentage decreased, and the contribution rate of other factors increased. Temporally, the contribution percentage of urbanization increased significantly from period T1 to period T2, and the contribution of other factors decreased significantly. The vegetation contribution percentage also slightly decreased in most of the cities.

In summary, from the Core Area to the Fringe Area, the spatial urbanization contribution percentage rapidly increases at first, followed by a slow increase. The vegetation contribution percentage rapidly declines during period T1, while it slowly declines during period T2. The contribution of other factors decreases quickly during period T1, but slowly increases during period T2. Temporally, the urbanization contribution in the Core Area and the Expanded Area decreases over time, whereas the contribution of vegetation increases. In the Fringe Area, the urbanization contribution increases rapidly with a rapid decrease in vegetation contribution. The other factors only have a significant increase in the Core Area, with decreasing trends in the Expanded Area and Fringe Area.
