干旱区生态环境知识资源中心

The ecosystem state or ’land condition’ can be characterized by a set of attributes, which show variations at different temporal scales. A multi-resolution analysis (MRA) based on the wavelet transform (WT) has been implemented to examine the land condition of a dryland region in Ferlo (Senegal) over the period 2001-2009. This methodology has proven to be useful for smoothing time series while considering those temporal resolutions that incorporate information about the vegetation dynamics. For this purpose, time series of the 1-km Apparent Green Cover (AGC) from the 10-day composites SPOT Vegetation (VGT) data are analyzed. Two relevant outputs from the MRA. A(1) (de-noised) and the A(6) (inter-annual) components have served us for characterizing the annual vegetation production and assess the long-term variation, respectively. In a first stage, the vegetation seasonality (or intra-annual variation) over the area is described by using several metrics related to vegetation phenology derived from the de-noised time series (A(1)). In a second phase, the temporal variability of the inter-annual component series (A(6)) is accomplished to detect potential vegetation changes over the considered period. A Mann-Kendall test has been applied to confirm the significance of the observed inter-annual changes. A higher number of significant pixels (86%) are obtained when considering the inter-annual component in the trend analysis instead of the original time-series (47%). The results confirm a general greening up over the period 2001-2009, not fully explained by precipitation, as well as rather local negative trends. The Rain-Use Efficiency (RUE) ratio computed using the AGC as a proxy of vegetation production has been considered to further analyze the detected changes. Inter-annual changes in RUE provide a potential method of separating vegetation declines due to lack of rainfall from declines associated with degradation. Some spots of negative values in inter-annual RUE changes are found in certain areas for which several evidence of land degradation have been documented. (C) 2011 Elsevier B.V. All rights reserved.