Abstract Rainfall is a key driver of terrestrial vegetation. Clarifying the response mechanism of vegetation to rainfall can advance the understanding of expected changes in ecosystems under projected rainfall scenarios. Besides the rainfall amount over a period of time, the frequency, duration and intensity are of importance in driving ecosystem processes. The pulsating nature of rainfall forcing and of vegetation response applies particularly to arid and semi – arid regions and led to conceptualize the “pulse-reserve” paradigm, which we used to explain the approach we propose in this study. We introduce the Transfer Function Analysis (TFA) method in the frequency domain to capture the response of vegetation to rainfall at multiple temporal scales. The TFA method determines coherence, gain, and phase to characterize the existence, strength, and time-lag, respectively, of the vegetation response to rainfall at different temporal scales. Specifically, the coherence measures the existence of the response and only with a significant response are the gain and phase values significant and valuable for further analysis. The gain measures the strength of the relationship between fluctuations in vegetation growth and fluctuations in rainfall, while the phase value (i.e. the time-lag) measures how fluctuations in vegetation growth lag (or lead) fluctuations in rainfall. The TFA method was applied to the 34-years (1982–2015) NDVI3g and CHIRPS precipitation dataset in the Sahel-Sudano-Guinean region (20°W ~ 60°E, 0–25°E). The Sahelian zone was characterized by a significant vegetation response to rainfall across all inter- and intra- annual time-scales, while the Sudano-Guinean zone was dominated by significant response at annual or 6-month scales. The negative phase lag indicated that rainfall variation normally led NDVI change for most areas and across timescales. However, a positive phase observed in part of the tropical rainforest area indicated that NDVI changes led rainfall variations, which may be caused by the strong vegetation-rainfall feedback through recycling of precipitation by evapotranspiration. In summary, these results suggested that the TFA method is a powerful tool to quantify the vegetation-rainfall response regime across a range of timescales, as conceptualized by the “pulse-reserve” paradigm. Unraveling the response of fluctuations in vegetation growth to separate components of the forcing by precipitation might improve our understanding of environmental change in the past decades in the Sahel - Sudan - Guinean region.

中文翻译：

---

使用传递函数分析表征萨赫勒-苏丹-几内亚地区多个时间尺度的植被对降雨的响应

摘要 降雨是陆地植被的主要驱动力。阐明植被对降雨的响应机制可以促进对预测降雨情景下生态系统预期变化的理解。除了一段时间内的降雨量外，频率、持续时间和强度对于驱动生态系统过程也很重要。降雨强迫和植被响应的脉动性质特别适用于干旱和半干旱地区，并导致了“脉冲储备”范式的概念化，我们用它来解释我们在本研究中提出的方法。我们在频域中引入了传递函数分析 (TFA) 方法，以捕捉植被在多个时间尺度上对降雨的响应。TFA 方法确定相干性、增益和相位以表征存在性、强度、和时滞，分别表示不同时间尺度下植被对降雨的响应。具体而言，相干性衡量响应的存在，并且只有响应显着时，增益和相位值才显着且对进一步分析有价值。增益衡量植被生长波动与降雨波动之间关系的强度，而相位值（即时滞）衡量植被生长波动如何滞后（或领先）降雨波动。TFA方法应用于萨赫勒-苏丹-几内亚地区（20°W~60°E，0-25°E）的34年（1982-2015）NDVI3g和CHIRPS降水数据集。萨赫勒地区的特点是植被对所有年际和年内时间尺度的降雨都有显着的响应，而苏丹-几内亚地区则以每年或 6 个月的显着响应为主。负相位滞后表明降雨变化通常导致大多数地区和跨时间尺度的 NDVI 变化。然而，在部分热带雨林地区观察到的正相位表明 NDVI 变化导致降雨量变化，这可能是由于蒸发蒸腾作用对降水再循环而产生的强烈植被-降雨反馈造成的。总之，这些结果表明 TFA 方法是一种强大的工具，可以量化跨时间尺度范围内的植被 - 降雨响应机制，正如“脉冲储备”范式所概念化的那样。