README - Data availability 

20 October 2023
Nick Corak

The purpose of this document is to explain the attached .csv files. The .csv files included accompany a manuscript titled "Unraveling phenological responses to extreme drought and implication on the water and carbon budgets" submitted to the journal Hydrology and Earth System Sciences.

We ran model simulations for three AmeriFlux sites in Kansas, USA (US-KFS, US-KLS, US-Kon) using a land surface hydrology model with phenological states forced with remote sensing data and then coupled the model to a predictive phenology model. 

Included in this repository are daily averaged model outputs with and without predictive phenology for soil moisture, root water uptake, gross primary productivity, and evapotranspiration. Also include predictive phenology outputs of leaf area index and fraction of photosynthetically active radiation. 

Updates to the original version of the repository include additions of infiltration, stomatal conductance, and vapor pressure deficit. Due to new analysis, we also include more years of data for all three sites. 

Naming and Labeling
Files are labeled as Site_OutputVariable(s)_year.csv
Example.  US-KFS_SoilMoisture_3Layers_2012.csv

The first column in each .csv file is a date string with format 'yyyymmdd'. *See note below*

Abbreviations
DCHM-V - land surface hydrology model without predictive phenology
DCHM-PV - land surface hydrology model with predictive phenology
FPAR - fraction of photosynthetically active radiation
LAI - leaf area index
ET - evapotranspiration
GPP - gross primary productivity
Infil - infiltration (of water into soils)
RU - root water uptake (three layers--> RU1 = top, RU2 = middle, RU3 = deep)
SC - stomatal conductance
SM - soil moisture (three layers--> SM1 = top, SM2 = middle, SM3 = deep)
VPD - vapor pressure deficit

WET - refers to the parameters used in the preditive phenology routine generated using a wet year, 2005
DRY - refers to the parameters used in the preditive phenology routine generated using a dry year, 2003
3YR - refers to the parameters used in the preditive phenology routine generated using a continuous three year period that encompassed the wet and dry years, 2003-2005 
SD - standard deviation. All SDs listed in the data sets are ONE standard deviation

Units
SM [m^3/m^3]
RU [mm/day]
ET [mm/day]
GPP [gC/m^2/day]
FPAR [-]
LAI [m^2/m^2]
VPD [kPa]
SC [mm/s]
Infil [mm]

Preditive Phenology
In order to run the predictive phenology routine, we need several parameters. We generated these parameters froma data assimilation routine (see manuscript) using three different precipitation regimes: a single wet year, a single dry year, and a three year period which encompassed the wet and dry years. We present FPAR and LAI from the three different simulations. Other output data presented are from simulation which used parameters from the 3YR simulations.  

Ensemble means
For each DCHM-PV we ran 2000 Monte Carlo simulations. Presented here are ensemble means and standard deviations used to generate the plots within the accompanying manuscript. Column headers in the .csv files identify which simulation, DCHM-V or DCHM-PV. DCHM-PV are labeled as WET, DRY, or 3YR. If no label exists for DCHM-PV, it refers to the DCHM-3YR simulation. Column headers are indicate mean or standard deviations (SD)
 
SM and RU each have three layers whose order and names can be found as column headers in the .csv files. The three layers used in this study are top (0-8 cm), middle (8-89 cm), and deep (89-183 cm). SM and RU is only reported for DCHM-PV-3YR.
Top = Layer1
Middle = Layer2
Deep = Layer3

ET can be separated into evaporation and transpiration. Transpiration = RU1 + RU2 + RU3. 

Water use efficiency (WUE) can be computed from GPP and ET as WUE = GPP/ET. 

Years of Data
Some figures in the manuscript show data outputs from a flash drought year (2012), a drought year (2018) and a non-drought year (2019). Other figures contain data from the water year (April-October) for all years 2006-2019 broken down into flash drought, drought, and non-drought years. 2012 is the only flash drought year. Drought years are 2006, 2011, 2013, 2014, and 2018. Non-drought years are 2007-10, 2015-2017, and 2019. Drought and non-drought years were determined form the US Drought Monitor for Kansas.   

SM data are provided for 2012, 2018, and 2019 only. All other data are provide for 2006-2019.

Hourly/Daily/Monthly
To investigate sub-daily trends, we provide hourly estimates of GPP, SC, and VPD. 

Most data is provided as a daily average or accumulation total. 

GPP is also provided as daily averages and as hourly data. 

** Note on data without date stamps
There is no date stamp for infiltration. Infiltration data is only provided as monthly totals for Apr - Oct for all years 2006-2019 (7 rows and 14 columns).

There is no date stamp for stomatal conductance daily average. Stomatal conductance is provided as daily averages from April 1 - Oct 31 (214 days) for all years 2006-2019 (214 rows and 14 columns).  

There is no date stamp for hourly data (stomatal conductance, GPP, and VPD). Column headers have the year. Each row represents the hourly data for the year. If the year is a leap year, the last 24 rows will have NaN values (8784 rows and 14 columns).