**4.3. The decrease in NDVI during 2009–2012 and its climatic explanation**

Additional to the uptrend of NDVI from 1982 to 2013, there were some years when the NDVI decreased, that is, from 2009 to 2012. The decreasing rate during this time was −0.017/year. The significant decline was mostly in Guizhou Province where a decreased rate less than −0.02/ year was observed (**Figure 10**). Correlation analysis between NDVI and climate change, revealed that the impact of temperature on the decreased NDVI was more profound than that from precipitation (**Figure 11**). Furthermore, the negative relationships between NDVI and precipitation also indicated the indirect impact of precipitation on temperature change. The increase in precipitation with more cloud could have led to the decrease in solar radiation and temperature, thus inhibiting photosynthesis.

**Figure 10.** Spatial patterns of variations in growing-season NDVI during 2009–2012.

#### **4.4. Assessing the dynamic downscaling of WRF**

**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 tempera-

ture trend; (**d**) coefficients image for precipitation trend.

42 Land Degradation and Desertification - a Global Crisis

Uncertainty on the downscaling capability of regional climate model (RCM) has in most cases led to skepticism for its use. Despite the weakness, the RCM dynamic downscaling is better than the simulations from General Circulation Model (GCM) or reanalysis datasets [13]. Furthermore, the uncertainty increases when the RCM is used to simulate the impact of land cover change on regional climate. In this section, the state-of-the-art RCM's downscaling ability was evaluated first, and was followed by analysis of the climatic effects of land degradation.

To reveal the improvement of WRF simulations over reanalysis dataset, daily rainfall, temperature, and other circulation factors from WRF and reanalysis were compared with the APHROD (Asian Precipitation-Highly-Resolved Observational Data) precipitation dataset, the GTS (Global Telecommunication System) temperature dataset, and the JRA-25 (Japanese 25-year Reanalysis) atmospheric variables dataset. The assessment was conducted from the viewpoint of correlation coefficient (R), bias and root mean square error (RMSE) over the years of 1998, 2000, and 2004 and over 18°–52°N, 86°–136°E (**Table 3**). The lower Bias and RMSE and the higher R values indicate better performance.

**Figure 11.** Multivariate regression coefficients of temperature (**a**) and precipitation (**b**) to NDVI based on pixel during 2009–2012.


**Table 3.** Descriptive statistics of ensemble mean JJA daily precipitation, temperature and water vapor flux at 700 hpa from WRF/SSiB and NCEP R-2 over 18°-52°N, 86°-136°E.

We further observed that the phenomenon of most rainfall occurring in the south of China, especially in the south of Yangtze River, can be detected from both WRF simulation and APHROD dataset. From the WRF simulation, there was also an obvious increasing trend from the northwest to southeast in the south of about 38°N with the minimum temperature in Qinghai-Tibetan Plateau. The WRF simulation of precipitation out-performed NCEP R-2, and was probably caused by the improved simulations of low level water vapor flux (**Table 3**), a key factor influencing the atmospheric convection in East Asian summer monsoon. Although the simulated surface temperature from WRF was not improved over NCEP R-2, the clearer spatial information for temperature was presented from WRF output, which suggests that, it is also an applicable tool in downscaling temperature.

### **4.5. Influence on precipitation and temperature due to KRD**

The area over 20°–34°N, 104°–124°E was chosen to investigate the impact of Karst rocky desertification on precipitation and temperature, because the significant and consistent effects were located in this region. There was spatial variation in the precipitation changes among the regions (**Figure 12a**). The reduced rainfall was mainly observed in the middle of Guizhou Karst Plateau. The areas with increased precipitation, mainly the middle and lower parts of Yangtze River and the surrounding areas, were of much larger magnitude and extent than that with decreased rainfall. It can be inferred that the consistent but nonsignificant reduction in rainfall with Guizhou Karst Plateau was due to high moisture influence from the Bay of Bengal. The land surface warming mainly occurred in the areas where the original vegetation types were replaced with bare soil type (**Figure 12b**), while the rainfall changes not only occurred within the desertification area but also beyond the area.

#### **4.6. Influence of KRD on land surface energy balance**

**Figure 11.** Multivariate regression coefficients of temperature (**a**) and precipitation (**b**) to NDVI based on pixel during

WRF/SSiB 1.57 3.16 0.78

WRF/SSiB −2.29 4.21 0.85

WRF/SSiB −1.37 7.49 0.70

**Variables Bias RMSE R** Precipitation NCEP R-2 1.95 4.22 0.60

Temperature NCEP R-2 −1.93 3.62 0.86

VQ700 NCEP R-2 2.89 11.38 0.65

**Table 3.** Descriptive statistics of ensemble mean JJA daily precipitation, temperature and water vapor flux at 700 hpa

2009–2012.

VQ700, water vapor flux at 700 hpa (g/kg/ms).

44 Land Degradation and Desertification - a Global Crisis

from WRF/SSiB and NCEP R-2 over 18°-52°N, 86°-136°E.

As shown in **Figure 13**, the substantial changes of surface energy components occurred in Guizhou Karst Plateau. In the degraded areas, the higher albedo (**Figure 13a**) led to more reflected shortwave radiation from the land surface (**Figure 13b**). Due to the higher surface skin temperature (**Figure 12b**), the outgoing longwave radiation increased significantly, which further caused the reduced net longwave radiation at the surface (**Figure 13c**). Both the reduction of the net shortwave radiation and the net longwave radiation certainly resulted in the decrease in land surface net radiation (**Figure 13d**). More sensible heat flux was also induced by the warmer surface (**Figure 13e**), however, the reduction in surface latent heat flux (**Figure 13f**) was much more than the sensible heat flux increase. The decrease in evaporation was probably contributed by changes in vegetation and soil properties, such as the lower LAI and roughness length, and the higher surface albedo. It can be concluded that evaporation decrease produced the most profound influence on the hydrological balance at land surface. Additionally, the above-mentioned higher temperature in the degraded areas was caused by the reduced evaporative cooling.

**Figure 12.** Ensemble mean differences in JJA **(a)** daily precipitation (mm/day) and **(b)** temperature (°C) between Case D and Case C. GKP is bounded by a heavy border.

Land-Atmosphere Interaction in the Southwestern Karst Region of China http://dx.doi.org/10.5772/64740 47

**Figure 13.** Ensemble mean differences in JJA **(a)** surface albedo, **(b)** net shortwave radiation, **(c)** net longwave radiation, **(d)** net radiation, **(e)** sensible heat flux, **(f)** latent heat flux, and **(g)** incoming shortwave radiation (W/m2 ) between Case D and Case C.

**Figure 12.** Ensemble mean differences in JJA **(a)** daily precipitation (mm/day) and **(b)** temperature (°C) between Case

D and Case C. GKP is bounded by a heavy border.

46 Land Degradation and Desertification - a Global Crisis

Consistent with the spatial changes in precipitation, there were areas with significantly changed energy budget extending beyond the degraded area. Outside the Guizhou Karst Plateau, the variations in sensible heat flux and latent heat flux were controlled by the precipitation differences. For example, in the areas between 30°–34°N, 112°– 120°E (i.e., the southeastern coastal area of China), the increased evaporation (**Figure 13f**) was caused by the increase in precipitation (**Figure 12a**), which further led to the lower temperature (**Figure 12b**), and the lower sensible heat flux (**Figure 13e**). The issue on the impact of atmospheric circulations on precipitation will be discussed in the next section.

**Figure 13(g)** shows the impacts of cloud albedo and land surface albedo on shortwave radiation. In the degraded areas within Guizhou Karst Plateau, the cloud fraction was reduced due to the less evaporation and moisture flux convergence after land degradation, and the reduced cloud fraction further led to more incoming shortwave radiation. However, the increase in upward shortwave radiation (**Figure 13a**) due to the higher land surface albedo was much more than the downward shortwave radiation, which resulted in the reduced net shortwave radiation (**Figure 13b**). Moreover, in the southeastern coastal areas of China, the increased cloud fractions, consistent with more rainfall, led to the decrease in incoming shortwave radiation, dominating the alteration in net shortwave radiation.
