Nithiya Streethran

Environmental research and random stuff

Airborne pollen concentration maps can be a useful indicator of areas to avoid during pollen season for people who suffer from hay fever1.

Pollen concentration in May. Pollen concentration in June. Pollen concentration in July. Pollen concentration in August. Pollen concentration in September.
Pollen concentration maps.

Tools used

Data (licenced under the Open Government Licence, version 3.0 (OGLv3.0)):

  • Abundance of airborne pollen for nine grass species, measured by qPCR, UK, 2016-20172
  • OS Open Zoomstack, layer names3
  • OS Boundary-Line™, layer country_region4

Software:

Data processing

First, the pollen data must be restructured into a GeoPackage using the Python script shown in the section below.

The vector point data is interpolated using the Inverse Distance Weighted (IDW) algorithm in QGIS with the following Python snippet:

layerList = ["May", "Jun", "Jul", "Aug", "Sep"]
extent = (
    "5512.998000000,655989.000000000,5333.809600000,1220310.000000000" +
    " [EPSG:27700]"
)

for layer in layerList:
    interpolation_data = (
        "data/pollen-data.gpkg|layername=" + layer + "::~::0::~::16::~::0"
    )
    params = {
        "INTERPOLATION_DATA": interpolation_data,
        "EXTENT": extent,
        "PIXEL_SIZE": 100,
        "OUTPUT": "data/idw_" + layer + ".tif"
    }
    processing.run("qgis:idwinterpolation", params)

Each interpolated raster layer is then clipped to Great Britain’s boundary:

for layer in layerList:
    params = {
        "INPUT": "data/idw_" + layer + ".tif",
        "MASK": "data/bdline_gb.gpkg|layername=country_region",
        "SOURCE_CRS": QgsCoordinateReferenceSystem("EPSG:27700"),
        "TARGET_CRS": QgsCoordinateReferenceSystem("EPSG:27700"),
        "OUTPUT": "data/idw_" + layer + "_clipped.tif"
    }

    processing.run("gdal:cliprasterbymasklayer", params)

Mapping procedure

Pollen station point styling:

  • Use a rule-based labelling to exclude Belfast using the filter Site!='Belfast'
  • Use the ‘effect drop shadow’ symbol available in QGIS’ ‘Showcase’ point symbol list
  • Change the fill colour to #fdbf6f and outline to #ff7f00
Point styling 1. Point styling 2. Point styling 3.
Point styling.

Basemap styling:

  • Use OS stylesheets for the appropriate layers5 6

  • The stylesheets can be applied using the following process:

    # Open Zoomstack
params = {
    "INPUT": "data/OS_Open_Zoomstack.gpkg|layername=names",
    "STYLE": "data/names.qml"
}
processing.run("native:setlayerstyle", params)

# Boundary-Line
params = {
    "INPUT": "data/bdline_gb.gpkg|layername=country_region",
    "STYLE": "data/country_region.qml"
}
processing.run("native:setlayerstyle", params)

Raster layer styling:

  • Singleband pseudocolour
  • Discrete interpolation
  • Label precision of 1
  • Quantile classification mode
  • 10 classes
  • Cubic resampling
Raster styling 1. Raster styling 2. Raster styling 3.
Raster layer styling.

Raster colour ramps:

Mako (inverted). PiYG (inverted). YlGn. Spectral (inverted). Rocket (inverted).
Raster colour ramps.

Mako and Rocket are Seaborn colour maps7 8, while the rest are from ColorBrewer9.

Python script to restructure pollen data

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

# import libraries
import pandas as pd
import geopandas as gpd

# read data
pollen = pd.read_csv("data/qPCR_copy_number_abundance_data_aerial_DNA.csv")

# create well known text from coordinates
pollen["wkt"] = (
    "POINT (" + pollen["Long"].astype(str) + " " +
    pollen["Lat"].astype(str) + ")"
)

# drop cells with no value
pollen = pollen.dropna(subset=["Lat", "Long", "MaxPoaceaeConc", "year-month"])

# use full pollen monitoring site name
pollen = pollen.replace({
    "EXE": "Exeter", "EastR": "East Riding", "ESK": "Eskdalemuir",
    "LEIC": "Leicester", "CAR": "Cardiff", "IOW": "Isle of Wight",
    "IPS": "Ipswich", "BNG": "Bangor", "WOR": "Worcester",
    "KCL": "King's College London", "YORK": "York", "ING": "Invergowrie",
    "BEL": "Belfast"
})

# create a geo data frame
pollen = gpd.GeoDataFrame(
    pollen, geometry=gpd.GeoSeries.from_wkt(pollen["wkt"]), crs="EPSG:4326"
)

# drop unnecessary columns
pollen = pollen.drop(columns=["Lat", "Long", "wkt"])

# reproject to BNG
pollen = pollen.to_crs("epsg:27700")

# get list of months with available data
monthList = list(pollen["year-month"].str[0:3].unique())

# get list of pollen monitoring sites
pollen_sites = pollen[["Site", "geometry"]].drop_duplicates()

# aggregate monthly data for each site and save as GeoPackage layers
for m in monthList:
    pollen_month = pollen[pollen["year-month"].str.contains(m)]
    pollen_month = pollen_month.groupby(["Site"]).mean("MaxPoaceaeConc")
    pollen_month = pd.merge(pollen_month, pollen_sites, on="Site")
    pollen_month.to_file("data/pollen-data.gpkg", layer=m)

Footnotes

  1. Met Office. 2022. Pollen Allergies. ↩

  2. Brennan, G., S. Creer, and G. Griffith. 2020. Abundance of Airborne Pollen for Nine Grass Species, Measured by qPCR, UK, 2016-2017 [Dataset] [CSV]. NERC Environmental Information Data Centre. ↩

  3. Ordnance Survey (GB). 2021. OS Open Zoomstack [Dataset] [GeoPackage] [v2021-12].. ↩

  4. Ordnance Survey (GB). 2021. OS Boundary-Line™ [Dataset] [GeoPackage] [v2021-10]. ↩

  5. Ordnance Survey (GB). 2021. OrdnanceSurvey/OS-Open-Zoomstack-Stylesheets/GeoPackage/QGIS Stylesheets (QML)/Outdoor style/names.qml.. GitHub. ↩

  6. Ordnance Survey (GB). 2021. OrdnanceSurvey/Boundary-Line-stylesheets/Geopackage stylesheets/QGIS Stylesheets (QML)/country_region.qml. GitHub. ↩

  7. Rudis, B., N. Ross, and S. Garnier. 2021. Introduction to the viridis color maps. ↩

  8. Waskom, M. 2021. Choosing color palettes - seaborn 0.11.2 documentation. seaborn. ↩

  9. Brewer, C. and M. Harrower. 2021. ColorBrewer: Color Advice for Maps. The Pennsylvania State University. ↩