Object-based cloud and cloud shadow detection in Landsat imagery

doi:10.1016/J.RSE.2011.10.028

Home
/
Papers
/
Object-based cloud and cloud shadow detection in Landsat imagery

Journal Article•DOI•

Object-based cloud and cloud shadow detection in Landsat imagery

Zhe Zhu¹, Curtis E. Woodcock¹•Institutions (1)

Boston University¹

15 Mar 2012-Remote Sensing of Environment (Elsevier)-Vol. 118, pp 83-94

TL;DR: The goal is development of a cloud and cloud shadow detection algorithm suitable for routine usage with Landsat images and as high as 96.4%.

read less

About: This article is published in Remote Sensing of Environment.The article was published on 2012-03-15. It has received 1620 citations till now. The article focuses on the topics: Cloud top & Cloud fraction.

...read moreread less

Citations

PDF

Open Access

More filters

Journal Article•DOI•

High-resolution mapping of global surface water and its long-term changes

[...]

Jean-François Pekel, Andrew Cottam, Noel Gorelick¹, Alan Belward•Institutions (1)

Google¹

15 Dec 2016-Nature

TL;DR: Using three million Landsat satellite images, this globally consistent, validated data set shows that impacts of climate change and climate oscillations on surface water occurrence can be measured and that evidence can be gathered to show how surface water is altered by human activities.

...read moreread less

Abstract: A freely available dataset produced from three million Landsat satellite images reveals substantial changes in the distribution of global surface water over the past 32 years and their causes, from climate change to human actions. The distribution of surface water has been mapped globally, and local-to-regional studies have tracked changes over time. But to date, there has been no global and methodologically consistent quantification of changes in surface water over time. Jean-Francois Pekel and colleagues have analysed more than three million Landsat images to quantify month-to-month changes in surface water at a resolution of 30 metres and over a 32-year period. They find that surface waters have declined by almost 90,000 square kilometres—largely in the Middle East and Central Asia—but that surface waters equivalent to about twice that area have been created elsewhere. Drought, reservoir creation and water extraction appear to have driven most of the changes in surface water over the past decades. The location and persistence of surface water (inland and coastal) is both affected by climate and human activity1 and affects climate2,3, biological diversity4 and human wellbeing5,6. Global data sets documenting surface water location and seasonality have been produced from inventories and national descriptions7, statistical extrapolation of regional data8 and satellite imagery9,10,11,12, but measuring long-term changes at high resolution remains a challenge. Here, using three million Landsat satellite images13, we quantify changes in global surface water over the past 32 years at 30-metre resolution. We record the months and years when water was present, where occurrence changed and what form changes took in terms of seasonality and persistence. Between 1984 and 2015 permanent surface water has disappeared from an area of almost 90,000 square kilometres, roughly equivalent to that of Lake Superior, though new permanent bodies of surface water covering 184,000 square kilometres have formed elsewhere. All continental regions show a net increase in permanent water, except Oceania, which has a fractional (one per cent) net loss. Much of the increase is from reservoir filling, although climate change14 is also implicated. Loss is more geographically concentrated than gain. Over 70 per cent of global net permanent water loss occurred in the Middle East and Central Asia, linked to drought and human actions including river diversion or damming and unregulated withdrawal15,16. Losses in Australia17 and the USA18 linked to long-term droughts are also evident. This globally consistent, validated data set shows that impacts of climate change and climate oscillations on surface water occurrence can be measured and that evidence can be gathered to show how surface water is altered by human activities. We anticipate that this freely available data will improve the modelling of surface forcing, provide evidence of state and change in wetland ecotones (the transition areas between biomes), and inform water-management decision-making.

...read moreread less

2,469 citations

Journal Article•DOI•

Improvement and expansion of the Fmask algorithm: cloud, cloud shadow, and snow detection for Landsats 4–7, 8, and Sentinel 2 images

[...]

Zhe Zhu¹, Shixiong Wang¹, Curtis E. Woodcock¹•Institutions (1)

Boston University¹

15 Mar 2015-Remote Sensing of Environment

TL;DR: In this article, a new cirrus band was introduced for detecting clouds, especially for thin cirrus clouds, and a new version of the Fmask algorithm was developed for use with Landsat 8 images.

...read moreread less

1,018 citations

Cites background or methods from "Object-based cloud and cloud shadow..."

...Most of the cloud detection algorithms are heavily dependent on the thermal band, as cloud pixels are much colder than clear-sky pixels (Huang, Thomas, et al., 2010; Irish et al., 2006; Zhu & Woodcock, 2012)....
[...]
...The pixels surrounding clouds and cloud shadows are masked out in the Fmask algorithm because many of these pixels may still be influenced by the thin edges of clouds and their shadows (Zhu & Woodcock, 2012)....
[...]
...%) for cloud detection over water (Zhu & Woodcock, 2012)....
[...]
...…that cloud shadows are usually dark in the Near Infrared (NIR) band, the original Fmask algorithm uses the difference between the NIR band and the flood-fill transformation (Soille, 1999; Soille, Vogt, & Colombo, 2003) of the same NIR band to extract potential cloud shadows (Zhu & Woodcock, 2012)....
[...]
...The Fmask algorithm matches clouds with their shadows based on similarity measurements (Zhu & Woodcock, 2012)....
[...]

Journal Article•DOI•

Continuous change detection and classification of land cover using all available Landsat data

[...]

Zhe Zhu¹, Curtis E. Woodcock¹•Institutions (1)

Boston University¹

25 Mar 2014-Remote Sensing of Environment

TL;DR: In this article, a two-step cloud, cloud shadow, and snow masking algorithm is used for eliminating noisy observations and a time series model that has components of seasonality, trend, and break estimates surface reflectance and brightness temperature.

...read moreread less

981 citations

Cites methods from "Object-based cloud and cloud shadow..."

...The robust feature of the RIRLS method reduces the influence of ephemeral changes, or pixels affected by clouds, shadows, or snow that were not identified by Fmask. ρ̂ i; xð ÞRIRLS ¼ a0;i þ a1;i cos 2π Τ x þ b1;i sin 2π Τ x þ a2;i cos 2π ΝΤ x þ b2;i sin 2π ΝΤ x ð1Þ where, x Julian date I the ith Landsat Band T number of days per year (T = 365) N number of years of Landsat data a0,i coefficient for overall values for the ith Landsat Band a1,i,b1,i coefficients for intra-annual change for the ith Landsat Band a2,i,b2,i coefficients for inter-annual change for the ith Landsat Band ρ̂ i; xð ÞRIRLS predicted value for the ith Landsat Band at Julian date x based on RIRLS fitting....
[...]
...The cloud cover estimates are derived from a newly developed Fmask algorithm (Zhu & Woodcock, 2012) for Path 12 Row 31 between 1982 and 2011....
[...]
...All available Level 1 Terrain (corrected) (L1T) Landsat TM/ETM+ images for Worldwide Reference System (WRS) Path 12 and Row 31 with cloud cover less than 80% (based on the Fmask results) were used (Fig....
[...]
...For observations acquired after model initialization, only the Fmask algorithm is used. ρ 2; xð Þ−ρ̂ 2; xð ÞRIRLSN0:04 OR ρ 5; xð Þ−ρ̂ 5; xð ÞRIRLSb−0:04 ð2Þ where, x Julian date ρ(i,x) Observed value for the ith Landsat Band at Julian date x. ρ̂ i; xð ÞRIRLS Predicted value for the ith Landsat Band at Julian date x based on RIRLS fitting (Eq....
[...]
...Though the Fmask algorithm provides relatively accurate masks for clouds, cloud shadows, and snow, it is not perfect....
[...]

Journal Article•DOI•

Optical remotely sensed time series data for land cover classification: A review

[...]

Cristina Gómez¹, Joanne C. White², Michael A. Wulder²•Institutions (2)

University of Aberdeen¹, Natural Resources Canada²

01 Jun 2016-Isprs Journal of Photogrammetry and Remote Sensing

TL;DR: In this article, the authors present the issues and opportunities associated with generating and validating time-series informed annual, large-area, land cover products, and identify methods suited to incorporating time series information and other novel inputs for land cover characterization.

...read moreread less

Abstract: Accurate land cover information is required for science, monitoring, and reporting. Land cover changes naturally over time, as well as a result of anthropogenic activities. Monitoring and mapping of land cover and land cover change in a consistent and robust manner over large areas is made possible with Earth Observation (EO) data. Land cover products satisfying a range of science and policy information needs are currently produced periodically at different spatial and temporal scales. The increased availability of EO data—particularly from the Landsat archive (and soon to be augmented with Sentinel-2 data)—coupled with improved computing and storage capacity with novel image compositing approaches, have resulted in the availability of annual, large-area, gap-free, surface reflectance data products. In turn, these data products support the development of annual land cover products that can be both informed and constrained by change detection outputs. The inclusion of time series change in the land cover mapping process provides information on class stability and informs on logical class transitions (both temporally and categorically). In this review, we present the issues and opportunities associated with generating and validating time-series informed annual, large-area, land cover products, and identify methods suited to incorporating time series information and other novel inputs for land cover characterization.

...read moreread less

784 citations

Cites background from "Object-based cloud and cloud shadow..."

...Automated processes can also be implemented for cloud and shadow detection and subsequent masking (e.g. Fmask; Zhu and Woodcock, 2012; Zhu et al., 2015a)....
[...]

Journal Article•DOI•

Cloud detection algorithm comparison and validation for operational Landsat data products

[...]

Steve Foga¹, Pat L. Scaramuzza¹, Song Guo¹, Zhe Zhu¹, Zhe Zhu², Ronald D. Dilley¹, Tim Beckmann¹, Gail L. Schmidt¹, John L. Dwyer¹, M. Joseph Hughes³, Brady Laue¹ - Show less +7 more•Institutions (3)

United States Geological Survey¹, Texas Tech University², Oregon State University³

01 Jun 2017-Remote Sensing of Environment

TL;DR: A workflow to evaluate cloud and cloud shadow masking algorithms using cloud validation masks manually derived from both Landsat 7 Enhanced Thematic Mapper Plus and Landsat 8 OLI/TIRS data is created, finding that CFMask, C code based on the Function of Mask (Fmask) algorithm, and its confidence bands have the best overall accuracy among the many algorithms tested using validation data.

...read moreread less

648 citations

Cites background or methods or result from "Object-based cloud and cloud shadow..."

...…active development by the creator, adoption from the academic community, physically meaningful attributes, ongoing development to utilize the Landsat 8 OLI cirrus band, development to support Sentinel-2 bands (Zhu et al., 2015) and a reported high cloud accuracy of 96.41% (Zhu and Woodcock, 2012)....
[...]
...The C programming language implementation of Function of Mask, known as CFMask, is a translation from Function of Mask, known as Fmask (Zhu and Woodcock, 2012; Zhu et al., 2015)....
[...]
...The C programming language implementation of Function of Mask, known as CFMask, is a translation from Function of Mask, known as Fmask (Zhu and Woodcock, 2012; Zhu et al., 2015)....
[...]
...We validated algorithms that already existed in both experimental and operational environments (Section 4), though we chose to implement Fmask (Zhu and Woodcock, 2012) in an operational environment, where we renamed it CFMask and created variants of it to produce cloud confidence bits, incorporate a cirrus test, remove the thermal tests from the algorithm, and add cirrus band tests to recover some accuracy from the lack of thermal data (Table 5)....
[...]
...• Function of Mask (Fmask) algorithm (Zhu and Woodcock, 2012)...
[...]

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
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200

Collapse

References

PDF

Open Access

More filters

Book•

Morphological Image Analysis: Principles and Applications

[...]

Pierre Soille

22 Dec 2012

TL;DR: This self-contained volume will be valuable to all engineers, scientists, and practitioners interested in the analysis and processing of digital images.

...read moreread less

Abstract: From the Publisher: The purpose of this book is to provide readers with an in-depth presentation of the principles and applications of morphological image analysis. This is achieved through a step by step process starting from the basic morphological operators and extending to the most recent advances which have proven their practical usefulness. This self-contained volume will be valuable to all engineers, scientists, and practitioners interested in the analysis and processing of digital images.

...read moreread less

4,018 citations

"Object-based cloud and cloud shadow..." refers background in this paper

...Therefore, a morphological transformation called flood-fill is performed for Band 4 reflectance (NIR band) which brings the intensity values of dark areas that are surrounded by lighter areas up to the same intensity level as the surrounding pixels (Soille, 1999)....
[...]

Journal Article•DOI•

A Landsat surface reflectance dataset for North America, 1990-2000

[...]

Jeffrey G. Masek¹, Eric Vermote², Nazmi Saleous¹, Robert E. Wolfe¹, Forrest G. Hall³, Karl F. Huemmrich³, Feng Gao¹, J. Kutler¹, Teng-Kui Lim¹ - Show less +5 more•Institutions (3)

Goddard Space Flight Center¹, University of Maryland, College Park², University of Maryland, Baltimore County³

16 Jan 2006-IEEE Geoscience and Remote Sensing Letters

TL;DR: Initial comparisons with ground-based optical thickness measurements and simultaneously acquired MODIS imagery indicate comparable uncertainty in Landsat surface reflectance compared to the standard MODIS reflectance product.

...read moreread less

Abstract: The Landsat Ecosystem Disturbance Adaptive Processing System (LEDAPS) at the National Aeronautics and Space Administration (NASA) Goddard Space Flight Center has processed and released 2100 Landsat Thematic Mapper and Enhanced Thematic Mapper Plus surface reflectance scenes, providing 30-m resolution wall-to-wall reflectance coverage for North America for epochs centered on 1990 and 2000. This dataset can support decadal assessments of environmental and land-cover change, production of reflectance-based biophysical products, and applications that merge reflectance data from multiple sensors [e.g., the Advanced Spaceborne Thermal Emission and Reflection Radiometer, Multiangle Imaging Spectroradiometer, Moderate Resolution Imaging Spectroradiometer (MODIS)]. The raw imagery was obtained from the orthorectified Landsat GeoCover dataset, purchased by NASA from the Earth Satellite Corporation. Through the LEDAPS project, these data were calibrated, converted to top-of-atmosphere reflectance, and then atmospherically corrected using the MODIS/6S methodology. Initial comparisons with ground-based optical thickness measurements and simultaneously acquired MODIS imagery indicate comparable uncertainty in Landsat surface reflectance compared to the standard MODIS reflectance product (the greater of 0.5% absolute reflectance or 5% of the recorded reflectance value). The rapid automated nature of the processing stream also paves the way for routine high-level products from future Landsat sensors.

...read moreread less

1,389 citations

"Object-based cloud and cloud shadow..." refers methods in this paper

...For Landsat L1T images, Digital Number (DN) values are converted to TOA reflectances and BT (Celsius degree) with the LEDAPS atmosphere correction tool (Masek et al., 2006; Vermote & Saleous, 2007)....
[...]

Journal Article•DOI•

Discriminating clear sky from clouds with MODIS

[...]

Steven A. Ackerman, Kathleen I. Strabala, W. Paul Menzel, Richard A. Frey, Christopher C. Moeller, Liam E. Gumley - Show less +2 more

27 Dec 1998-Journal of Geophysical Research

TL;DR: The MODIS cloud mask algorithm as discussed by the authors uses several cloud detection tests to indicate a level of confidence that the MEDIS is observing clear skies, which is ancillary input to MEDIS land, ocean, and atmosphere science algorithms to suggest processing options.

...read moreread less

Abstract: The MODIS cloud mask uses several cloud detection tests to indicate a level of confidence that the MEDIS is observing clear skies. It will be produced globally at single-pixel resolution; the algorithm uses as many as 14 of the MEDIS 36 spectral bands to maximize reliable cloud detection and to mitigate past difficulties experienced by sensors with coarser spatial resolution or fewer spectral bands. The MEDIS cloud mask is ancillary input to MEDIS land, ocean, and atmosphere science algorithms to suggest processing options. The MEDIS cloud mask algorithm will operate in near real time in a limited computer processing and storage facility with simple easy-to-follow algorithm paths. The MEDIS cloud mask algorithm identifies several conceptual domains according to surface type and solar illumination, including land, water, snow/ice, desert, and coast for both day and night. Once a pixel has been assigned to a particular domain (defining an algorithm path), a series of threshold tests attempts to detect the presence of clouds in the instrument field of view. Each cloud detection test returns a confidence level that the pixel is clear ranging in value from 1 (high) to zero (low). There are several types of tests, where detection of different cloud conditions relies on different tests. Tests capable of detecting similar cloud conditions are grouped together. While these groups are arranged so that independence between them is maximized, few, if any, spectral tests are completely independent. The minimum confidence from all tests within a group is taken to be representative of that group. These confidences indicate absence of particular cloud types. The product of all the group confidences is used to determine the confidence of finding clear-sky conditions. This paper outlines the MEDIS cloud masking algorithm. While no present sensor has all of the spectral bands necessary for testing the complete MEDIS cloud mask, initial validation of some of the individual cloud tests is presented using existing remote sensing data sets.

...read moreread less

1,198 citations

"Object-based cloud and cloud shadow..." refers background in this paper

...Though it works sometimes, most of the time it will inevitably include other dark surfaces that have similar spectral signatures (like topographic shadows or wetlands) and exclude cloud shadows that are not dark enough (Ackerman et al., 1998; Hutchison et al., 2009)....
[...]
...These sensors are usually equipped with more than one thermal band, or with water vapor/CO2 absorption bands, both of which are useful for thin semitransparent cloud detection (Ackerman et al., 1998; Derrien et al., 1993; Saunders & Kriebel, 1998)....
[...]

Journal Article•DOI•

Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and the input data

[...]

Yuanchong Zhang¹, William B. Rossow², Andrew A. Lacis², Valdar Oinas, Michael I. Mishchenko² - Show less +1 more•Institutions (2)

Columbia University¹, Goddard Institute for Space Studies²

16 Oct 2004-Journal of Geophysical Research

TL;DR: Zhang et al. as discussed by the authors used a more advanced NASA Goddard Institute for Space Studies (GISS) radiative transfer model and improved ISCCP cloud climatology and ancillary data sets.

...read moreread less

Abstract: [1] We continue reconstructing Earth’s radiation budget from global observations in as much detail as possible to allow diagnosis of the effects of cloud (and surface and other atmospheric constituents) variations on it. This new study was undertaken to reduce the most noticeable systematic errors in our previous results (flux data set calculated mainly using International Satellite Cloud Climatology Project–C1 input data (ISCCP-FC)) by exploiting the availability of a more advanced NASA Goddard Institute for Space Studies (GISS) radiative transfer model and improved ISCCP cloud climatology and ancillary data sets. The most important changes are the introduction of a better treatment of ice clouds, revision of the aerosol climatology, accounting for diurnal variations of surface skin/air temperatures and the cloud-radiative effects on them, revision of the water vapor profiles used, and refinement of the land surface albedos and emissivities. We also extend our previous flux results, limited to the top of atmosphere (TOA) and surface (SRF), to also include three levels within the atmosphere, forming one integrated vertical atmospheric flux profile from SRF to TOA, inclusive, by combining a new climatology of cloud vertical structure with the ISCCP cloud product. Using the new radiative transfer model and new input data sets, we have produced an 18-year at 3-hour time steps, global at 280-km intervals, radiative flux profile data set (called ISCCP-FD) that provides full- and clear-sky, shortwave and longwave, upwelling and downwelling fluxes at five levels (SRF, 680 mbar, 440 mbar, 100 mbar, and TOA). Evaluation is still only possible for TOA and SRF fluxes: Comparisons of monthly, regional mean values from FD with Earth Radiation Budget Experiment, Clouds and the Earth’s Radiant Energy System and Baseline Surface Radiation Network values suggest that we have been able to reduce the overall uncertainties from 10–15 to 5–10 W/m 2 at TOA and from 20–25 to 10– 15 W/m 2 at SRF. Annual mean pressure-latitude cross sections of the cloud effects on atmospheric net radiative fluxes show that clouds shift the longwave cooling downward in the Intertropical Convergence Zone, acting to stabilize the tropical atmosphere while increasing the horizontal heating gradient forcing the Hadley circulation, and shift the longwave cooling upward in the midlatitude storm zones, acting to destabilize the baroclinic zones while decreasing the horizontal heating gradient there. INDEX TERMS: 1620 Global Change: Climate dynamics (3309); 3309 Meteorology and Atmospheric Dynamics: Climatology (1620); 3359 Meteorology and Atmospheric Dynamics: Radiative processes; KEYWORDS: Earth radiation budget, surface radiation budget (SRB), cloud vertical structure, ERBE, CERES, BSRN Citation: Zhang, Y., W. B. Rossow, A. A. Lacis, V. Oinas, and M. I. Mishchenko (2004), Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and the input data, J. Geophys. Res., 109, D19105, doi:10.1029/2003JD004457.

...read moreread less

1,076 citations

"Object-based cloud and cloud shadow..." refers background in this paper

...The International Satellite Cloud Climatology Project-Flux Data (ISCCP-FD) data set estimates the global annual mean cloud cover is approximately 66% (Zhang et al., 2004)....
[...]

Journal Article•DOI•

Spectral signature of alpine snow cover from the Landsat Thematic Mapper.

[...]

Jeff Dozier¹, Jeff Dozier²•Institutions (2)

University of California, Santa Barbara¹, California Institute of Technology²

01 Apr 1989-Remote Sensing of Environment

TL;DR: In this article, the spectral signatures of the Landsat TM images of the Sierra Nevada were analyzed to distinguish several classes of snow from other surface covers, and a number of TM images were used for automatic analysis of alpine snow cover.

...read moreread less

907 citations

"Object-based cloud and cloud shadow..." refers background in this paper

...However, if visible bands become saturated at a small value, for example 0.5 (Dozier, 1989), while NIR and SWIR bands do not (close to 1), it would make the absolute values of NDSI and NDVI much larger than 0, making probability of spectral variability lower for cloud pixels....
[...]
...5 (Dozier, 1989), while NIR and SWIR bands do not (close to 1), it...
[...]