gremlin97/RemoteSensingCorpus · Datasets at Hugging Face

text stringlengths 0 820
(ETM+)Landsat 6
(ETM)Landsat 4–5
(TM)Landsat 1–5
(MSS)Landsat 3
(RBV)Landsat 1–2
(RBV)
89
7 65 4
32187 5 432187 5 43217 5 432143211321
11 121530100153060153012030120804080
Resolution (m)
11 10666
Figure 6: Spectral wavelengths and spatial resolutions of each band captured by all Landsat sensors.
On Landsats 1–3, MSS bands were actually numbered 4–7. Landsat 9 introduced new and improved
OLI-2/TIRS-2 sensors, but the bands are identical, so the sensors were combined in this figure.
A.7 Data visualization
(a) Spring (2021-03-22)
(b) Summer (2022-06-13)
(c) Fall (2022-10-19)
(d) Winter (2022-12-22)
Figure 7: Example location showing the time-series nature of SSL4EO-L. Each location has imagery
from 4 different seasons. Images are selected from a 60-day window centered about the vernal and
autumnal equinoxes and the summer and winter solstices in order to maximize seasonal changes.
Images are limited to a 2-year window to minimize man-made changes. Image location is Ci County,
Handan, Heibei, China.
20
A.8 Model complexity
Table 6: Complexity of backbone models used in this paper. Includes the number of parameters,
memory requirements, floating point operations per second (FLOPS), and multiply-accumulate op-
erations (MACs) of each model. All experiments were performed on an NVIDIA A100 GPU.
Model # Params (M) Memory (MB) FLOPS (G/s) MACs (G)
ResNet-18 11.21 44.87 622.49 136.21
ResNet-50 23.56 94.46 366.32 281.72
ViT-S16 22.46 89.83 423.61 281.28
A.9 Sampling algorithm
Procedure DownloadSSL4EO( N= 250 ,000, S= 4, σ= 50 km)
Data: M={µ}centroids of 10K most populous cities in the world
Result: Downloads non-overlapping, cloud-free, nodata-free images from Nlocations during
Sseasons
X← {}
while len( X)< N :
µ∼ U(M)
x∼ N(µ, σ)
# Ensure x does not overlap with existing sampled patches
ifOverlaps (x,X):# 264 px buffer
continue
# Look for S cloud-free, nodata-free images at location x
T← {}
t←0
fortinrange( S):# 60-day and 2-year window around equinoxes/solstices
ifCloudCover( x, t):# 20% threshold
continue
ifNoData( x, t):
continue
T←T∪ {t}
iflen( T)< S:
continue
# Download from Google Earth Engine
Download( x,T)
X←X∪ {x}
21
Detecting trend and seasonal changes in satellite image time series
Jan Verbesselta,⁎, Rob Hyndmanb, Glenn Newnhama, Darius Culvenora
aRemote Sensing Team, CSIRO Sustainable Ecosystems, Private Bag 10, Melbourne VIC 3169, Australia
bDepartment of Econometrics and Business Statistics, Monash University, Melbourne VIC 3800, Australia
abstract article info
Article history:
Received 24 June 2009Received in revised form 13 August 2009Accepted 18 August 2009
Keywords:Change detectionNDVI
Time series
Trend analysisMODISPiecewise linear regressionVegetation dynamicsPhenologyA wealth of remotely sensed image time series covering large areas is now available to the earth science
community. Change detection methods are often not capable of detecting land cover changes within time
series that are heavily in ﬂuenced by seasonal climatic variations. Detecting change within the trend and
seasonal components of time series enables the classi ﬁcation of different types of changes. Changes occurring
in the trend component often indicate disturbances (e.g. ﬁres, insect attacks), while changes occurring in the
seasonal component indicate phenological changes (e.g. change in land cover type). A generic changedetection approach is proposed for time series by detecting and characterizing Breaks For Additive Seasonal
and Trend (BFAST). BFAST integrates the decomposition of time series into trend, seasonal, and remainder
components with methods for detecting change within time series. BFAST iteratively estimates the time andnumber of changes, and characterizes change by its magnitude and direction. We tested BFAST by simulating16-day Normalized Difference Vegetation Index (NDVI) time series with varying amounts of seasonality and
noise, and by adding abrupt changes at different times and magnitudes. This revealed that BFAST can
robustly detect change with different magnitudes (>0.1 NDVI) within time series with different noise levels(0.01 –0.07σ) and seasonal amplitudes (0.1 –0.5 NDVI). Additionally, BFAST was applied to 16-day NDVI
Moderate Resolution Imaging Spectroradiometer (MODIS) composites for a forested study area in south
eastern Australia. This showed that BFAST is able to detect and characterize spatial and temporal changes in aforested landscape. BFAST is not speci ﬁc to a particular data type and can be applied to time series without
the need to normalize for land cover types, select a reference period, or change trajectory. The method can be
integrated within monitoring frameworks and used as an alarm system to ﬂag when and where changes
occur.
Crown Copyright © 2009 Published by Elsevier Inc. All rights reserved.
1. Introduction
Natural resource managers, policy makers and researchers de-
mand knowledge of land cover changes over increasingly large spatial
and temporal extents for addressing many pressing issues such as
global climate change, carbon budgets, and biodiversity ( DeFries et al.,
1999; Dixon et al., 1994 ). Detecting and characterizing change over
time is the natural ﬁrst step toward identifying the driver of the
change and understanding the change mechanism. Satellite remote
sensing has long been used as a means of detecting and classifying
changes in the condition of the land surface over time ( Coppin et al.,
2004; Lu et al., 2004 ). Satellite sensors are well-suited to this task
because they provide consistent and repeatable measurements at a
spatial scale which is appropriate for capturing the effects of many
processes that cause change, including natural (e.g. ﬁres, insect

(ETM+)Landsat 6

(ETM)Landsat 4–5

(TM)Landsat 1–5

(MSS)Landsat 3

(RBV)Landsat 1–2

(RBV)

89

7 65 4

32187 5 432187 5 43217 5 432143211321

11 121530100153060153012030120804080

Resolution (m)

11 10666

Figure 6: Spectral wavelengths and spatial resolutions of each band captured by all Landsat sensors.

On Landsats 1–3, MSS bands were actually numbered 4–7. Landsat 9 introduced new and improved

OLI-2/TIRS-2 sensors, but the bands are identical, so the sensors were combined in this figure.

A.7 Data visualization

(a) Spring (2021-03-22)

(b) Summer (2022-06-13)

(c) Fall (2022-10-19)

(d) Winter (2022-12-22)

Figure 7: Example location showing the time-series nature of SSL4EO-L. Each location has imagery

from 4 different seasons. Images are selected from a 60-day window centered about the vernal and

autumnal equinoxes and the summer and winter solstices in order to maximize seasonal changes.

Images are limited to a 2-year window to minimize man-made changes. Image location is Ci County,

Handan, Heibei, China.

20

A.8 Model complexity

Table 6: Complexity of backbone models used in this paper. Includes the number of parameters,

memory requirements, floating point operations per second (FLOPS), and multiply-accumulate op-

erations (MACs) of each model. All experiments were performed on an NVIDIA A100 GPU.

Model # Params (M) Memory (MB) FLOPS (G/s) MACs (G)

ResNet-18 11.21 44.87 622.49 136.21

ResNet-50 23.56 94.46 366.32 281.72

ViT-S16 22.46 89.83 423.61 281.28

A.9 Sampling algorithm

Procedure DownloadSSL4EO( N= 250 ,000, S= 4, σ= 50 km)

Data: M={µ}centroids of 10K most populous cities in the world

Result: Downloads non-overlapping, cloud-free, nodata-free images from Nlocations during

Sseasons

X← {}

while len( X)< N :

µ∼ U(M)

x∼ N(µ, σ)

# Ensure x does not overlap with existing sampled patches

ifOverlaps (x,X):# 264 px buffer

continue

# Look for S cloud-free, nodata-free images at location x

T← {}

t←0

fortinrange( S):# 60-day and 2-year window around equinoxes/solstices

ifCloudCover( x, t):# 20% threshold

continue

ifNoData( x, t):

continue

T←T∪ {t}

iflen( T)< S:

continue

# Download from Google Earth Engine

Download( x,T)

X←X∪ {x}

21

Detecting trend and seasonal changes in satellite image time series

Jan Verbesselta,⁎, Rob Hyndmanb, Glenn Newnhama, Darius Culvenora

aRemote Sensing Team, CSIRO Sustainable Ecosystems, Private Bag 10, Melbourne VIC 3169, Australia

bDepartment of Econometrics and Business Statistics, Monash University, Melbourne VIC 3800, Australia

abstract article info

Article history:

Received 24 June 2009Received in revised form 13 August 2009Accepted 18 August 2009

Keywords:Change detectionNDVI

Time series

Trend analysisMODISPiecewise linear regressionVegetation dynamicsPhenologyA wealth of remotely sensed image time series covering large areas is now available to the earth science

community. Change detection methods are often not capable of detecting land cover changes within time

series that are heavily in ﬂuenced by seasonal climatic variations. Detecting change within the trend and

seasonal components of time series enables the classi ﬁcation of different types of changes. Changes occurring

in the trend component often indicate disturbances (e.g. ﬁres, insect attacks), while changes occurring in the

seasonal component indicate phenological changes (e.g. change in land cover type). A generic changedetection approach is proposed for time series by detecting and characterizing Breaks For Additive Seasonal

and Trend (BFAST). BFAST integrates the decomposition of time series into trend, seasonal, and remainder

components with methods for detecting change within time series. BFAST iteratively estimates the time andnumber of changes, and characterizes change by its magnitude and direction. We tested BFAST by simulating16-day Normalized Difference Vegetation Index (NDVI) time series with varying amounts of seasonality and

noise, and by adding abrupt changes at different times and magnitudes. This revealed that BFAST can

robustly detect change with different magnitudes (>0.1 NDVI) within time series with different noise levels(0.01 –0.07σ) and seasonal amplitudes (0.1 –0.5 NDVI). Additionally, BFAST was applied to 16-day NDVI

Moderate Resolution Imaging Spectroradiometer (MODIS) composites for a forested study area in south

eastern Australia. This showed that BFAST is able to detect and characterize spatial and temporal changes in aforested landscape. BFAST is not speci ﬁc to a particular data type and can be applied to time series without

the need to normalize for land cover types, select a reference period, or change trajectory. The method can be

integrated within monitoring frameworks and used as an alarm system to ﬂag when and where changes

occur.

1. Introduction

Natural resource managers, policy makers and researchers de-

mand knowledge of land cover changes over increasingly large spatial

and temporal extents for addressing many pressing issues such as

global climate change, carbon budgets, and biodiversity ( DeFries et al.,

1999; Dixon et al., 1994 ). Detecting and characterizing change over

time is the natural ﬁrst step toward identifying the driver of the

change and understanding the change mechanism. Satellite remote

sensing has long been used as a means of detecting and classifying

changes in the condition of the land surface over time ( Coppin et al.,

2004; Lu et al., 2004 ). Satellite sensors are well-suited to this task

because they provide consistent and repeatable measurements at a

spatial scale which is appropriate for capturing the effects of many

processes that cause change, including natural (e.g. ﬁres, insect