Interplay of drought and tropical cyclone activity in SE U.S. gross primary productivity

Abstract

Tropical cyclones (TCs), often associated with massive flooding and landslides in the Southeast U.S. (SE U.S.), provide a significant input of freshwater to the hydrologic system, and their timing and trajectory significantly impact drought severity and persistence. This manuscript investigates the sensitivity of gross primary productivity (GPP) in the SE U.S. to TC activity using the 1-D column implementation of the Duke Coupled Hydrology Model with Vegetation (DCHM-V) including coupled water and energy cycles and a biochemical representation of photosynthesis. Decadal-scale simulations of water, energy, and carbon fluxes were conducted at high temporal (30 min) and spatial (4 km) resolution over the period 2002-2012. At local scales, model results without calibration compare well against AmeriFlux tower data. At regional scales, differences between the DCHM-V estimates and the Moderate Resolution Imaging Spectroradiometer GPP product reflect the spatial organization of soil hydraulic properties and soil moisture dynamics by physiographic region, highlighting the links between the water and carbon cycles. To isolate the contribution of TC precipitation to SE U.S. productivity, control forcing simulations are contrasted with simulations where periods of TC activity in the atmospheric forcing data were replaced with climatology. During wet years, TC activity impacts productivity in 40-50% of the SE U.S. domain and explains a regional GPP increase of 3-5 Mg C/m 2 that is 9% of the warm season total. In dry years, 23-34% of the domain exhibits a smaller positive response that corresponds to 4-8% of the seasonal carbon uptake, depending on TC timing and trajectory.

Decadal climate variability and the spatial organization of deep hydrological drought

“…Campbell et al 2010). Combining information based on physiographic region including climate, topography and underlying geology is necessary to interpret long-range baseflow response, and to examine nonlinear interactions among climate forcing and hydrologic drought (Bachman et al 1998, Bloomfield et al 2009, Lowman and Barros 2016. Here, baseflow was first extracted using a two-pass filter algorithm applied to streamflow records from gauges at the outlet of small and medium-size USSA watersheds (area < 5000 km 2 ), followed by space-time modal analysis using EOF and wavelet methods.…”

Seeing the Disturbed Forest for the Trees: Remote Sensing Is Underutilized to Quantify Critical Zone Response to Unprecedented Disturbance

“…Understanding and predicting the drivers of water stress on plant health and productivity is critical in light of increasing drought events from climate change (Meehl & Tebaldi, 2004). Derived estimates of plant phenology and gross primary productivity from satellite remote sensing depend solely on the spectral reflectance of the surface, and do not consider soil water conditions (Lowman & Barros, 2016, 2018). However, satellite estimates may be improved by assimilating the data into models that incorporate soil water stress (Lowman & Barros, 2018).…”

“…The AmeriFlux GPP data show that photosynthesis rates were severely reduced during the period where both soil moisture falls below SM min and VPD exceeds VPD max (Figure 9). The MODIS GPP data did not show the same reduction in plant productivity, because the satellite remote sensing product did not account for changes in soil moisture that influence plant photosynthesis (Lowman & Barros, 2016, 2018).…”

Unraveling phenological responses to extreme drought and implications for water and carbon budgets

“…Before scaling to the DCHM grid and time scale, the data for each pixel were smoothed using a Savitsky-Golay (Savitzky and Golay, 1964) algorithm following Chen et al (2004) in order to preserve seasonality and reduce noise associated with cloud cover and other atmospheric effects (Cihlar et al, 1997;Tanré et al, 1997). We use the m=6 scaling window and d=4 degree of the interpolating polynomial as in Chen et al (2004) and Lowman and Barros (2016). After the smoothing filter is applied, the data is up-scaled from 500 m to 4 km resolution and linearly interpolated to a 1 h temporal resolution.…”