**4. Results and discussion**

#### **4.1. Variations in growing-season NDVI**

As shown in **Figure 3**, the rate of 0.0015/year during 1982–2013 was estimated for the growingseason NDVI trend in the Karst region of southwest China. The maximum value can be found in 2009 with significant variations between different years. It is indicated in **Figure 3(b)** that the year of 1994 was a tipping point, which means that there were two states before and after this year for the NDVI anomaly. We observed decreasing trend for some years, although the overall trend was increasing. Furthermore, the M-K trend test showed significant increasing trend, especially after the year 2004. As for the variation in NDVI of different vegetation types, the increasing rate was highest for coniferous forest, and the smallest value for meadow (**Table 2**).

**Figure 3.** Interannual variations in growing-season NDVI (a) and NDVI anomaly (b) during 1982-2013 in the entire region, using the annual average growing-season NDVI.


**4. Results and discussion**

(**Table 2**).

**4.1. Variations in growing-season NDVI**

36 Land Degradation and Desertification - a Global Crisis

gion, using the annual average growing-season NDVI.

As shown in **Figure 3**, the rate of 0.0015/year during 1982–2013 was estimated for the growingseason NDVI trend in the Karst region of southwest China. The maximum value can be found in 2009 with significant variations between different years. It is indicated in **Figure 3(b)** that the year of 1994 was a tipping point, which means that there were two states before and after this year for the NDVI anomaly. We observed decreasing trend for some years, although the overall trend was increasing. Furthermore, the M-K trend test showed significant increasing trend, especially after the year 2004. As for the variation in NDVI of different vegetation types, the increasing rate was highest for coniferous forest, and the smallest value for meadow

**Figure 3.** Interannual variations in growing-season NDVI (a) and NDVI anomaly (b) during 1982-2013 in the entire re-

**Table 2.** Statistical characteristics of growing-season NDVI for different vegetation types during 1982–2013.

**Figure 4.** Spatial patterns of average values in growing-season NDVI during 1982–2013.

**Figure 4** shows the spatial distribution of NDVI values in the study area, ranging from 0.32 to 0.85. Due to higher temperature and more precipitation in Guangxi Zhuang Autonomous Region, there were high values of NDVI in the east part of the study area. Under the background of complex climate change, there was also spatial heterogeneity for the dynamical variation of NDVI. The higher increasing rate was observed in the northwest and the smaller values in the southeast (**Figure 5**).

**Figure 5.** Spatial patterns of temporal trend in growing-season NDVI during 1982–2013.

#### **4.2. Correlations between NDVI and climate factors**

We observed warming rate of 0.018°C/year in the study area (**Figure 6a**). It fluctuated from −0.6 ∼ 0.8°C for average growing-season temperature. The year of 1995 was a tipping point for temperature and NDVI changes. Specifically, the average temperature for different months presented obvious variations with a maximum temperature (25.2°C) in July. For the changes in precipitation, **Figure 6(c)** shows a decrease of −1.21mm/year during 1982–2013. The dynamic processes for precipitation can be classified as falling under three stages: 1982–1992, 1993–2002, and 2003–2013 (**Figure 6d**). Additionally, the significant uptrend for temperature can be concluded from the Mann-Kendall test.

Land-Atmosphere Interaction in the Southwestern Karst Region of China http://dx.doi.org/10.5772/64740 39

**Figure 4** shows the spatial distribution of NDVI values in the study area, ranging from 0.32 to 0.85. Due to higher temperature and more precipitation in Guangxi Zhuang Autonomous Region, there were high values of NDVI in the east part of the study area. Under the background of complex climate change, there was also spatial heterogeneity for the dynamical variation of NDVI. The higher increasing rate was observed in the northwest and the smaller

**Figure 5.** Spatial patterns of temporal trend in growing-season NDVI during 1982–2013.

We observed warming rate of 0.018°C/year in the study area (**Figure 6a**). It fluctuated from −0.6 ∼ 0.8°C for average growing-season temperature. The year of 1995 was a tipping point for temperature and NDVI changes. Specifically, the average temperature for different months presented obvious variations with a maximum temperature (25.2°C) in July. For the changes in precipitation, **Figure 6(c)** shows a decrease of −1.21mm/year during 1982–2013. The dynamic processes for precipitation can be classified as falling under three stages: 1982–1992, 1993–2002, and 2003–2013 (**Figure 6d**). Additionally, the significant uptrend for temperature can be

**4.2. Correlations between NDVI and climate factors**

concluded from the Mann-Kendall test.

values in the southeast (**Figure 5**).

38 Land Degradation and Desertification - a Global Crisis

**Figure 6.** Interannual variations in average growing-season temperature trend and anomaly (**a**); monthly temperature (**b**); precipitation trend and anomaly (**c**); monthly precipitation (**d**); and the results (**e**, **f**) of Mann-Kendall test during 1982–2013 in the entire region.

#### *4.2.1. Traditional linear regression for NDVI and climate variables*

As shown in **Figure 7(a)**, there was obvious synergy for NDVI and temperature, but the synergy for NDVI and precipitation was relatively weak (**Figure 7b**). The lower regression coefficients of precipitation indicated the weaker impact of precipitation on vegetation cover change. The reason may be that there was rich rainfall in the study area, and the annual variation cannot play significant roles. Moreover, the correlations between NDVI and climatic variables were different for different vegetation types (shown in **Table 2**). The largest regression coefficient was in grassland.

**Figure 7.** The overall relationship between annual growing-season NDVI and temperature (**a**); precipitation (**b**) during 1982–2013.

In most areas, the relationship between NDVI and temperature (**Figure 8a**) was positive due to the strengthened photosynthesis and vegetation activity by the increase in temperature. It should be pointed out that only within an appropriate range, the temperature rise can result in beneficial effects, and if the temperature is too high, it will cause negative impact on vegetation growth. **Figure 8(b)** shows the regression coefficient for NDVI and precipitation. Although the correlation was positive in most of the areas, there were some negative values in the northern part of the study area.

#### *4.2.2. Local regression for the spatial relationships*

The later one means applying the changing rate of NDVI (**Figure 5**) as the dependent variable of GWR while the changing rate of climatic factors as independent variables. **Figure 9** lists the GWR regression coefficients, where colors ranging from blue to red represented values from low to high. Additionally, the standard errors were analyzed by the points with different sizes.

There was positive relationships between multiyear average NDVI and temperature (**Figure 9a**), however, the regression coefficients for NDVI and precipitation contained both positive and negative values (**Figure 9b**). It was found that the positive values for NDVI and precipitation were mainly located in Yunnan Province, where the climate is more arid than other areas of the study area. The GWR regression coefficients for dynamic relationships were listed in **Figure 9(c)** and **(d)**. The NDVI was lower with increasing surface temperature, which may be explained as more serious aridity due to the warming. On the other hand, the correlation between the changing rate of NDVI with precipitation were positive, meaning that the increase in NDVI during 1982–2013 could have been caused mainly by the precipitation variations.

*4.2.1. Traditional linear regression for NDVI and climate variables*

sion coefficient was in grassland.

40 Land Degradation and Desertification - a Global Crisis

in the northern part of the study area.

*4.2.2. Local regression for the spatial relationships*

1982–2013.

As shown in **Figure 7(a)**, there was obvious synergy for NDVI and temperature, but the synergy for NDVI and precipitation was relatively weak (**Figure 7b**). The lower regression coefficients of precipitation indicated the weaker impact of precipitation on vegetation cover change. The reason may be that there was rich rainfall in the study area, and the annual variation cannot play significant roles. Moreover, the correlations between NDVI and climatic variables were different for different vegetation types (shown in **Table 2**). The largest regres-

**Figure 7.** The overall relationship between annual growing-season NDVI and temperature (**a**); precipitation (**b**) during

In most areas, the relationship between NDVI and temperature (**Figure 8a**) was positive due to the strengthened photosynthesis and vegetation activity by the increase in temperature. It should be pointed out that only within an appropriate range, the temperature rise can result in beneficial effects, and if the temperature is too high, it will cause negative impact on vegetation growth. **Figure 8(b)** shows the regression coefficient for NDVI and precipitation. Although the correlation was positive in most of the areas, there were some negative values

The later one means applying the changing rate of NDVI (**Figure 5**) as the dependent variable of GWR while the changing rate of climatic factors as independent variables. **Figure 9** lists the GWR regression coefficients, where colors ranging from blue to red represented values from low to high. Additionally, the standard errors were analyzed by the points with different sizes.

There was positive relationships between multiyear average NDVI and temperature (**Figure 9a**), however, the regression coefficients for NDVI and precipitation contained both

**Figure 8.** Multivariate regression coefficients of temperature (**a**); and precipitation (**b**) to NDVI based on pixel during 1982–2013.

**Figure 9.** Geographically weighted regression analysis between NDVI and temperature and precipitation during 1982– 2013. (**a**) Coefficients image for temperature; (**b**) coefficients image for precipitation; (**c**) coefficients image for temperature trend; (**d**) coefficients image for precipitation trend.
