Effects of Green-Up on Spring Maximum Temperatures

A look at the effect of Green-Up on high temperature extremes in northen New England

Background

In 2014, in my last year of my undergraduate Meteorology degree at Plymouth State University. I completed a Senior Research Project titled “Effects of Green-up on Spring and Summer Maximum Temperatures in Northern New Hampshire from 1989 to 2012”; I later presented this research at the 39th Northeast Storm Conference in Rutland, Vermont, USA. My original research project was done using Microsoft Excel, but used the same data sets. The goal of this project specifically was to redo the project, with an additional eight years of data, in R.

Introduction

The Hubbard Brook Experimental Forest (HBEF) is a 7,800-acre valley located within the White Mountain National Forest in New Hampshire, USA. Research started by the Northern Research Station office of the US Forest Service because in 1956, the Hubbard Brook Ecosystem Study (HBES) began in 1953, and HBEF was designated a National Science Foundation Long-Term Ecological Research Forest in 1988. (Hubbard Brook Ecosystem Study, 2022; USDA Forest Service Northern Research Station, 2022)

Image 1: Site map of the Hubbard Brook Experimental Forest in the White Mountain National Forest, New Hampshire. (Campbell, et. al., 2007)

In the sizable HBES Data Catalog, there are over 60 years of daily temperature records, and yearly documentation of tree growth in the spring. When the leaves of deciduous trees have fully bloomed is called “Green-Up”; in the mountains of northern New England, this is late May through early June. With Green-Up comes a tempering of temperature extremes, as the data below shows; there are more days where the maximum temperature is significantly greater than normal before Green-Up than after. (Hubbard Brook Ecosystem Study, n.d.)

Data

Phenology Data

Import

phen1 <- read.csv('https://pasta.lternet.edu/package/data/eml/knb-lter-hbr/51/12/9f623c83fb1da7595c6d2d498bde15df')

#phen1 <- read.csv('C:/Users/ahammond1/OneDrive - University of Massachusetts/DataSci/HBEF_Phenology_longform.csv')

** Spring Phenology Codes: ** [0] No change from winter conditions, unexpanded buds only [1] Bud swelling noticeable [2] Small leaves or flowers visible, initial stages of leaf expansion, leaves about 1 cm long [3] Leaves 1/2 of final length, leaves obscure half of sky as seen through crowns [3.5] Leaves 3/4 expanded, sky mostly obscured through crown, crowns not yet in summer condition [4] Canopy appears in summer condition leaves fully expanded little sky visible through crowns

(USDA Forest Service Northern Research Station, 2021b)

Cleaning

New Column: YEAR

phen1 <- phen1 %>%                      # YYYY-MM-DD to YYYY
   mutate(phen1, YEAR = substr(DATE, 0, 4))

phen1$YEAR <- as.integer(phen1$YEAR)    # Define as integer
phen1$DATE <- as.Date(phen1$DATE)       # Define & format as a Date
phen1$DOY <- as.integer(phen1$DAY)      # Redefining "DAY" to DOY (Day Of Year)

phen1$dayCode = paste(phen1$YEAR, phen1$DOY, sep = ",")   # Merge Year and Day into one item, easier for coding.

phen1 <- phen1 %>%
  arrange(YEAR, DOY)

Filter: Spring Records Only

phenSpring <- filter(phen1, `SEASON` == "SPRING") %>%           # Removes FALL Values
  select(`YEAR`, `DOY`, `dayCode`, `Phenology_Stage`, `DATE`)   # remove unnecessary columns

remove(phen1)

Calculate: Mean Phenology Stage

SpringCal <- phenSpring %>%       #Min, Mean, Max Stage by Year and Day, non-weighted, does not account for Site/Species
  group_by(dayCode) %>%
  summarise(minStage = min(Phenology_Stage), 
            averageStage = mean(Phenology_Stage), 
            maxStage = max(Phenology_Stage)) 

SpringCal <- SpringCal %>%    #Split DayCode into Year and DOY
  mutate(YEAR = substr(dayCode, 0, 4), 
         DOY = substr(dayCode,6,8) )

SpringCal$YEAR <- as.integer(SpringCal$YEAR) #Define as INT
SpringCal$DOY <- as.integer(SpringCal$DOY) #Define as INT

SpringCal <- SpringCal %>%
  arrange(YEAR, DOY)

Define and Filter for Green-Up

Green-Up defined as Phenology_Stage >= 3.75

GreenUp <- SpringCal %>%
  filter(`averageStage` >= 3.75) %>%
  group_by(YEAR)

GreenUp = GreenUp[!duplicated(GreenUp$YEAR), ] # Remove duplicated >> keep first instance of Stage >= 3.75

GreenUp$date <- as.Date(GreenUp$DOY, origin = "2019-01-01")  # Non-leap Year
GreenUp$date <- format(GreenUp$date, format = "%b-%d")       # Monday-Day format

GreenUp = select(GreenUp, `YEAR`, `date`, `DOY`)

remove(phenSpring, SpringCal)

Temperature Data

Import

dailyTemp <- read_csv('https://pasta.lternet.edu/package/data/eml/knb-lter-hbr/59/10/9723086870f14b48409869f6c06d6aa8')

#dailyTemp <- read_csv('C:/Users/ahammond1/OneDrive - University of Massachusetts/DataSci/HBEF_air_temp_daily_1957-2021.csv')

(USDA Forest Service Northern Research Station, 2021a)

Cleaning

dailyTemp <- dailyTemp %>% # Use only HQ station for temp
  filter(STA == "HQ")      # HQ starts in 1956, other stations/locations start in later years

dailyTemp$date <- as.Date(dailyTemp$date, "%Y-%m-%d")

Normals

Using Daily Temperature Records from 1959 to 2021

normals <- dailyTemp %>%
  group_by(month = month(date), day = (day(date))) %>%
  summarise(count1 = n(),                         # How many years averaged together
            Tmin = mean(MIN), sdMin = sd(MIN),    # sd = standard deviation
            Tave = mean(AVE), sdAve = sd(AVE),
            Tmax = mean(MAX), sdMax = sd(MAX),
            sigma = sdMax,
            oneSigma = (Tmax + sigma),            # Events >= 1 standard deviation greater than normal
            twoSigma = (Tmax + (2*sigma)))         # Events >= 2 standard deviations greater than normal) 

normals$month <- as.integer(normals$month)
normals$dateX = paste(normals$month, normals$day, sep = "-")
normals$dateX <- as.Date(normals$dateX, format = "%m-%d")
normals$DOY <- format(normals$dateX, format = "%j")
normals$DOY <- as.integer(normals$DOY)

Filter: Spring Normals

Spring defined as January 1 through July 31

normalSPRING <- normals %>%
  filter(month >= 1 & month <= 7) %>%
  select(month, day, dateX, DOY, 
         Tmax, oneSigma, twoSigma)

Filter: Maximum Daily Temperatures

tempSpring <- dailyTemp %>%
   mutate(YEAR = substr(`date`, 0, 4))

tempSpring <- tempSpring %>%
   mutate(MONTH = substr(`date`, 6, 7))

tempSpring$YEAR <- as.integer(tempSpring$YEAR)
tempSpring$MONTH <- as.integer(tempSpring$MONTH)

tempSpring <- tempSpring %>%
  filter(STA == "HQ" & 
         YEAR >= 1989 & YEAR <=2020 &
         MONTH >= 1 & MONTH <= 7)

tempSpring$date <- as.Date(tempSpring$date, format = "%m-%d")
tempSpring$DOY <- format(tempSpring$date, format = "%j")
tempSpring$DOY <- as.integer(tempSpring$DOY)

#remove(dailyTemp)

Megatable Formated for tidyr

Merge Green-Up and Daily Normals

tempSpringNORMALS <- tempSpring %>%
  inner_join(
    x = tempSpring, y = normalSPRING, 
    by = 'DOY')

tempSpNorGreen <- tempSpringNORMALS %>%
  inner_join(
    x = tempSpringNORMALS, y = GreenUp,
    by = 'YEAR')

remove(tempSpringNORMALS)

Determine Maximum Temperature Events

• Days where the Recorded Daily Maximum Temperature is Greater Than or Equal To the Normal Maximum Temperature
• Then Counted by if the Event is Greater Than or Equal To 1 Standard Deviation or Two Standard Deviations above Normal

  eventsB <- tempSpNorGreen %>%
    group_by(YEAR) %>%
    filter(DOY.x < DOY.y & MAX >= Tmax) %>%
    summarise(bGreenCt = n())
  
  eventsB1 <- tempSpNorGreen %>%
    group_by(YEAR) %>%
    filter(DOY.x < DOY.y & MAX >= oneSigma) %>%
    summarise(bGreen1 = n())
  
  eventsB <- eventsB %>%
    left_join(
      x = eventsB,
      y = eventsB1,
      by = 'YEAR')
  
  eventsB1 <- tempSpNorGreen %>%
    group_by(YEAR) %>%
    filter(DOY.x < DOY.y & MAX >= twoSigma) %>%
    summarise(bGreen2 = n())
  
  eventsB <- eventsB %>%
    left_join(
      x = eventsB,
      y = eventsB1,
      by = 'YEAR')
  
  remove(eventsB1)
  
  eventsA <- tempSpNorGreen %>%
    group_by(YEAR) %>%
    filter(DOY.x >= DOY.y & MAX >= Tmax) %>%
    summarise(aGreenCt = n())
  
  eventsA1 <- tempSpNorGreen %>%
    group_by(YEAR) %>%
    filter(DOY.x >= DOY.y & MAX >= oneSigma) %>%
    summarise(aGreen1 = n())
  
  eventsA <- eventsA %>%
    left_join(
      x = eventsA,
      y = eventsA1,
      by = 'YEAR')
  
  eventsA1 <- tempSpNorGreen %>%
    group_by(YEAR) %>%
    filter(DOY.x >= DOY.y & MAX >= twoSigma) %>%
    summarise(aGreen2 = n())
  
  eventsA <- eventsA %>%
    left_join(
      x = eventsA,
      y = eventsA1,
      by = 'YEAR')
  
  remove(eventsA1)
  
  events <- eventsB %>%
    left_join(
      x = eventsB,
      y = eventsA,
      by = 'YEAR')
  
  remove(eventsA, eventsB)
  
  events[is.na(events)] <- 0  # If NA, make 0
  
  events <- events %>%        # Rename Columns
    rename(
      "Before Green-Up" = bGreenCt,
      "Before + 1σ" = bGreen1,
      "Before + 2σ" = bGreen2,
      "After Green-Up" = aGreenCt,
      "After + 1σ" = aGreen1,
      "After + 2σ" = aGreen2)