Main content area

Identifying long-term variations in vegetation and climatic variables and their scale-dependent relationships: A case study in Southwest Germany

Liu, Zhiyong, Menzel, Lucas
Global and planetary change 2016 v.147 pp. 54-66
case studies, climatic factors, development planning, normalized difference vegetation index, seasonal variation, temperature, time series analysis, variance, vegetation types, water management, wavelet, Germany
Geographic time series are usually non-stationary and contain different frequency components (e.g., seasonal variations, long-term and short-term fluctuations) which may significantly affect the overall variance structure in the original data. Based upon the monthly normalized difference vegetation index (NDVI), precipitation and temperature data for six different vegetation types in two precipitation regimes (low and high precipitation regimes) of Rhineland-Palatinate (Southwest Germany), this study aims to examine the temporal trends in the original time series of these variables and their relationships. In addition, the further objectives are to evaluate which time-scale is dominantly responsible for the trend production found in the original data and find out the certain time-scales that represent the strongest correlation between NDVI and climatic variables (i.e., precipitation and temperature). A combined approach using the discrete wavelet transform (DWT), Mann-Kendall (MK) trend test and correlation analysis was implemented to achieve these goals. The trend assessment for the original data shows that the monthly NDVI time series for all vegetation types in both precipitation regimes have upward trends, most of which are significant. The precipitation and temperature data for six vegetation types in two precipitation regimes present weak downward trends and significant increasing trends, respectively. The most important time-scales contributing the trend production in the original NDVI data are the 2-month and 8-month events. For precipitation, the most influential ones are 2-month and 4-month scales. The 4-month periodic mode predominantly affects the trends in the original temperature time series. Based on the original time series, the change in temperature is found to be the primary driver influencing the variability in vegetation greenness over this study area, while there is a negative correlation between NDVI and precipitation for all vegetation types and precipitation regimes. For the scale-dependent relationships between NDVI and precipitation, the 2-month and 8-month scales generally present the strongest negative correlation. The most significant positive correlations between NDVI and temperature are obtained at the 8- to 16-month scales for most vegetation types. The results of the present study might be beneficial for water resources management as well as agricultural and ecological development planning in Rhineland-Palatinate, and also provide a helpful reference for other regions with similar climate condition.