Vegetation Response and Streamflow Anomalies: Exploring the Modulating Effect of Watershed Storage as Estimated by a Regionalized Stream Recession Index

Water managers often make water allocation decisions based on data that is integrated at regional scales much coarser than those at which water management decisions are typically made. Important sub-regional variations in the data are subsumed in the aggregate, potentially leading to an improper handling of water resources. A combination of stream discharge characteristics and remotely sensed data can provide information that is responsive at local scales, such as watershed vulnerability to anomalous moisture conditions. We conducted an exploratory analysis of Normalized Difference Vegetation Index (NDVI) data (500 m2 resolution, 16 day) obtained from NASA's Moderate Resolution Imaging Spectroradiometer and USGS stream discharge (Q) records from over 100 unregulated streams in Tennessee for the years 2001-2012. The data sets were compiled to evaluate the vegetation response during a historical drought (Aug/Sept of 2007) within different streamflow recession index (SRI) regions. SRI can be applied as a metric for watershed storage and the ability of underlying aquifers to sustain streamflow through prolonged dry periods. The time series were filtered to remove seasonal trends, and bimonthly anomalies were calculated. Each of the three NDVI and Q time series (raw, filtered, and anomaly) were analyzed using cross-correlation analysis, cross-wavelet, and wavelet coherence analyses. Four SRI regions with similar land cover were chosen to spatially analyze NDVI anomalies during drought. The results from the cross-correlation analysis reveal strong biannual and annual correlations between raw NDVI and raw discharge values. Correlations between NDVI anomalies and discharge anomalies peak at lag periods of 1 to 1.5 months with NDVI leading. The wavelet coherence analysis suggests that drought dampens the monthly signal correlation between the raw values, and potentially removes a strong 2 year correlation between the anomalies. The spatial analysis shows regions with a higher SRI value tend to have higher NDVI values throughout a drought.