ased on Surfaces Generated from Ground Weather Stations ... · This Koppen system is founded on the...

1
Introducon India’s current climate classificaon method is almost a century old. This Koppen system is founded on the princi- ple that vegetaon is a proxy for temperature and precipi- taon, which are proxies for climate. Ideally satellites could be used to gather temperature and precipitaon over large spaal and temporal scales. Unfortunately, India’s mon- soon season causes much of the country to be cov- ered in clouds making remote sensing inaccurate. Interpolaon of weather variables from ground staons offer a way to circumvent these is- sues, providing current, accurate, and connuous data for climate classifi- caon. While remote sensing is sll preferable, this method could be used for reci- procity year round or as a fill in during the monsoon season. Methods Monthly weather staon data from 1980 to present was down- loaded as .csv’s from NOAA’s: Naonal Climate Data Center (NCDC) and scrubbed for errors in Excel. The data was imported into ArcMap and geolocated. The staons were then fed into a tool created with Model Builder, that iterated the data grouping by date (month) before feeding it into IDW interpolaon (basic method based off distances), and outpung mean raster sur- faces. These surfaces could then have cell stascs run on them for different monthly groupings depending on the user’s needs. I began with running stascs on all the rasters to create a “30 year normal”, which is considered in the literature the re- quired amount of me to obtain long term trends from. Surfaces were then generated from hot vs. cold (Apr-Aug vs. Sep-Mar) months, and seasons were made from every three months starng from January. Climate Classificaon of India, Based on Surfaces Generated from Ground Weather Staons Ground Weather Staons In India Mean Temperature and Precipitaon Surfaces (1980 - present) Surface data strafied biannually Surface data strafied biannually Strafied by season Strafied by season Conclusions Beginning with the two “30 year normals” for temperature and precipitaon, you can see that India’s usually prey hot (mostly between 70 and 80⁰F) and gets a reasonable amount of total rain per month (between 0.5 and 1m). When each of these are strafied into hot vs. cold months you can begin to understand the variability in climate per me of year. The cold poron sees less heat (70-80⁰F) and very lile rain (0-0.5m). Note: snow was not included in this analysis. The hot poron of the year is usually 10⁰ warmer and has more rain (0.5- 2m). When these are further strafied by sea- son, addional informaon can be drawn out. Noce how as the year progresses the climate changes, when reading from leſt to right (Note: autumn was placed at the front so as to held keep “cold” months grouped together). Sum- mer is the hoest, followed closely by the monsoon season, while there is praccally no rain in the winter as compared to the monsoon season. The straficaon capability by user preference is the importance of this work, and it was only for the sake of easy interpretaon that the legend scales were chosen. This data is available in as much detail as supplied by the source. The main limitaon was the use of IDW interpolaon and temporal scale, which were only used so as to limit computaonal me. Future Work Addional work was done, that for the sake of presentaon was not included on this poster. This used Alex Liss’s the creaon of zones based off of vegetaon spectral signature which were mosaicked, water masked, PCA analyzed, classified, and smoothed via the remote sensing soſt- ware ENVI before being imported into GIS where zonal stascs were performed on the raster surfaces. The zones, and results can be seen below: These could be summarized in a table or placed into a bubble plot, where color represents zone, each variable is an axis, and the size of the bubble is related to cell counts. This would shows that each zone is indeed a separate cluster indicang unique climate. Climate could then be related to populaon and health outcomes per zone, such as heat exposure in India. Data: NOAA NCDC GSOD Projecon: World Geodec System 1984 UTM Zone 44 Acknowledgements: Alex Liss for help troubleshoong and allowing me to build off his past work. 12/12/2014 Madeline Wrable UEP232—Introducon to GIS

Transcript of ased on Surfaces Generated from Ground Weather Stations ... · This Koppen system is founded on the...

Page 1: ased on Surfaces Generated from Ground Weather Stations ... · This Koppen system is founded on the princi-ple that vegetation is a proxy for temperature and precipi-tation, which

Introduction

India’s current climate classification method is almost a

century old. This Koppen system is founded on the princi-

ple that vegetation is a proxy for temperature and precipi-

tation, which are proxies for climate. Ideally satellites could

be used to gather temperature and precipitation over large

spatial and temporal scales. Unfortunately, India’s mon-

soon season causes much of the country to be cov-

ered in clouds making remote sensing inaccurate.

Interpolation of weather variables from ground

stations offer a way to circumvent these is-

sues, providing current, accurate, and

continuous data for climate classifi-

cation. While remote sensing is

still preferable, this method

could be used for reci-

procity year round or

as a fill in during

the monsoon

season.

Methods Monthly weather station data from 1980 to present was down-

loaded as .csv’s from NOAA’s: National Climate Data Center

(NCDC) and scrubbed for errors in Excel. The data was imported

into ArcMap and geolocated. The stations were then fed into a

tool created with Model Builder, that iterated the data grouping

by date (month) before feeding it into IDW interpolation (basic

method based off distances), and outputting mean raster sur-

faces. These surfaces could then have cell statistics run on

them for different monthly groupings depending on

the user’s needs. I began with running statistics on

all the rasters to create a “30 year normal”,

which is considered in the literature the re-

quired amount of time to obtain long

term trends from. Surfaces were

then generated from hot vs. cold

(Apr-Aug vs. Sep-Mar)

months, and seasons

were made from every

three months

starting from

January.

Climate Classification of India, Based on Surfaces Generated from Ground Weather Stations

Ground Weather Stations

In India

Mean Temperature and Precipitation Surfaces

(1980 - present)

Surface data stratified biannually

Surface data stratified biannually

Stratified by season Stratified by season

Conclusions Beginning with the two “30 year normals” for

temperature and precipitation, you can see

that India’s usually pretty hot (mostly between

70 and 80⁰F) and gets a reasonable amount of

total rain per month (between 0.5 and 1m).

When each of these are stratified into hot vs.

cold months you can begin to understand the

variability in climate per time of year. The cold

portion sees less heat (70-80⁰F) and very little

rain (0-0.5m). Note: snow was not included in

this analysis. The hot portion of the year is

usually 10⁰ warmer and has more rain (0.5-

2m). When these are further stratified by sea-

son, additional information can be drawn out.

Notice how as the year progresses the climate

changes, when reading from left to right (Note:

autumn was placed at the front so as to held

keep “cold” months grouped together). Sum-

mer is the hottest, followed closely by the

monsoon season, while there is practically no

rain in the winter as compared to the monsoon

season. The stratification capability by user

preference is the importance of this work, and

it was only for the sake of easy interpretation

that the legend scales were chosen. This data

is available in as much detail as supplied by the

source. The main limitation was the use of IDW

interpolation and temporal scale, which were

only used so as to limit computational time.

Future Work Additional work was done, that for the sake of presentation was not included on this poster.

This used Alex Liss’s the creation of zones based off of vegetation spectral signature which were

mosaicked, water masked, PCA analyzed, classified, and smoothed via the remote sensing soft-

ware ENVI before being imported into GIS where zonal statistics were performed on the raster

surfaces. The zones, and results can be seen below:

These could be summarized in a table or placed into a bubble plot, where color represents

zone, each variable is an axis, and the size of the bubble is related to cell counts. This would

shows that each zone is indeed a separate cluster indicating unique climate. Climate could then

be related to population and health outcomes per zone, such as heat exposure in India.

Data: NOAA NCDC GSOD

Projection: World Geodetic System 1984 UTM Zone 44

Acknowledgements: Alex Liss for help troubleshooting and allowing me to build off his past work.

12/12/2014

Madeline Wrable

UEP232—Introduction to GIS