Great Lakes Algorithms

NASA Workshop for Remote Sensing
of Coastal & Inland Waters
Madison, Wisconsin
June 20-22, 2012
Dr. Robert Shuchman, MTRI
George Leshkevich, NOAA GLERL
Contributors:
Michael Sayers, MTRI
Colin Brooks, MTRI
1
Great Lakes Algorithms
-A Work in Progress-
2
2
Specific Great Lakes Optical Satellite
Algorithm Suite Under Development
CPA for chlorophyll (chl), dissolved organic carbon (doc)
and suspended minerals (sm)
Sediment plume model (extent, constituents,
concentrations, and load)
Primary productivity (daily, monthly, and annual averages)
Lake bottom type (Cladophora/SAV, rocks, sand)
Harmful Algal Blooms (HABs)
Optical water properties (clarity, Kd, Photosynthetically
Active Radiation (PAR), and photic zone)
Wetlands mapping (combined optical + radar)
3
Status of Satellite Remote Sensing
Algorithm Development
Algorithm
CPA
Sediment Plumes
Primary Productivity Lake Bottom Mapping
Retrieval
Status
Next Steps
Remarks
Chl, doc, and sm concentrations
Algorithm for each lake developed and tested (JGLR paper submitted); HO models updated
Evaluation for CoastWatch operational production
Algorithm outperforms standard OC3 chl approach
& provides add’l DOC & SM info
Area extent, constituents,
concentration, and load
Algorithm developed and has undergone preliminary evaluation
Quantify concentration and continue validation
Addresses river plumes,
bays and complex basins
Daily/monthly/yearly gC/m2/d or m or yr
Algorithm is developed and undergoing evaluation
Upgrade code to account for chl and temp and Kd variability within photic zone
Existing algorithm using NASA Kd works well for
spring and fall conditions
Bottom type (sand, SAV inc. Cladophora, or rock) in optically shallow water
Fully developed and operational; Lakes Michigan & Huron done
Collect additional “truth” to update biomass
Mapping of Submerged Aquatic Vegetation (SAV) for Lakes Ontario and Erie underway
Location and extent
Work in progress in
collaboration with EPA GLRI, GLOS, NOAA GLERL
Utilize GLERL Lake Erie 2011 dataset for algorithm evaluation (GLRI – 2008‐
2012 baseline)
HABs mapping is utilizing knowledge gained from CPA and plume mapping
Clarity (optically shallow water) Kd, PAR, photic zone
Algorithm is operational Run additional time‐series presently using coastal and compare to ship based ocean constants to calculate IOP
values
Initial evaluation of algorithm indicates it is quite robust
NWI wetland type update for coastal Great Lakes (US + Canada)
4 year GLRI project in year 1 – obtaining imagery, field data
Combined SAR+optical
algorithm
HABs
Optical Water Properties
Wetlands
Use field data, existing data sources to start updated mapping work
44
Lake Huron CPA
Station t
HU 48
CPA CHL
(ug/L)
CPA vs.
EPA (ug/L)
OC3 CHL
(ug/L)
OC3 vs.
EPA (ug/L)
0.33
0.27
-0.06
0.19
-0.14
HU 45M
0.71
0.42
-0.29
0.44
-0.26
HU 37
0.33
0.38
0.05
0.20
-0.13
HU 38
0.31
0.33
0.02
0.17
-0.14
HU 32
0.34
0.37
0.03
0.02
-0.32
HU 27
0.33
0.34
0.00
0.25
-0.08
HU 15M
0.32
0.25
-0.08
0.20
-0.13
HU 93
0.36
0.40
0.04
0.20
-0.16
HU 12
0.41
0.37
-0.04
0.22
-0.20
HU 09
0.37
0.32
-0.05
0.25
-0.12
HU 06
Averages
Lake Huron CPA and
OC3 chl retrievals for
August 12, 2010. The
red dots indicate the
locations of the EPA
sampling stations. In
general the open Lake
OC3 chl derived values
are on the low side
while the nearshore
values are artificially
high.
EPA CHL
(ug/L)
0.53
0.37
-0.16
0.24
-0.29
0.39
0.35
-0.05
0.22
-0.18
August 2010 Lake Huron CPA and
OC3 derived chl values compared to
EPA in situ measurements for the
stations indicated. Individual station
values are presented along with
differences and averages. A negative
value indicates an under prediction.
Note all the OC3 retrievals were
underestimations when compared to
the EPA observations. The CPA
average chl observation compared
quite favorably with the EPA truth.
5
MODIS Derived Saginaw Bay
Sediment Plume – 15 Apr 2009
TSSI_GL
High
Moderate
6
Lake Michigan Primary Productivity
7
Lake Michigan SAV/Cladophora
In Lake Michigan, 24% of the
visible bottom consists of SAV,
mostly Cladophora (1024 km2
out of 4210 km2)
―
Some areas of Chara, other
diatoms
The optical depth varied from
7m to 18m depth
Conservative estimate of wet
weight biomass is 375,000
metric tonnes
Huron, Erie, Ontario coming
Info site inc. mapping at:
http://www.mtri.org/cladophora.html
8
MODIS Derived HABs Extent
09 Oct 2011
In collaboration with
NOAA GLERL, GLOS,
EPA GLRI
EPA GLRI – 2008-2012
baseline maps
Floating Algae
Sediment
Sediment/Algae Mix
9
Optical Example
MODIS-Derived
Lake Michigan
Kd(490)
10
Optical Example
MODIS-Derived
Lake Michigan
Kd(PAR)
11
Optical Example
MODIS-Derived
Lake Michigan
Photic Depth
12
Sayers2
Optical Example
Landsat-Derived
Lake Michigan Kd
13
Slide 13
Sayers2
Change kd490 to simply KD
Mike Sayers, 6/14/2012
Optical Example
Landsat-Derived
Lake Michigan
Kd(PAR)
14
Optical Example
Landsat-Derived
Lake Michigan Photic
Depth
15
Sayers3
Preliminary Lake Michigan Visibility
Estimates from 1974-2009
Water Bottom
Visibility Depth
(meters)
Optical Depth
0
5
10
15
20
25
1979
1984
1989
Kd (m-1)
1999
2004
2009
1999
2004
2009
1999
2004
2009
1999
2004
2009
Kd
0.15
0.1
0.05
0
1979
1984
1989
1994
KPAR
0.15
0.1
0.05
ZEU (meters)
Kd(PAR) (m‐1)
1994
0
1979
20
30
40
50
60
70
1979
1984
1989
1994
Photic Depth
1984
1989
1994
16
Slide 16
Sayers3
Fix this figure
Mike Sayers, 6/14/2012
Atmospheric Correction Example
Lake Huron Atmospheric Correction Comparison
0.012
White Aerosol Extraction
2 Band NIR Black Pixel
Remote Sensing Reflectance
0.010
Standard
0.008
Fixed Model Pair
Fixed Model Pair NIR Iterative
0.006
MUMM NIR Correction
0.004
NIR/SWIR Switch
0.002
No Aerosol Subtraction
In situ
0.000
400
450
500
550
Wavelength (nm)
600
650
700
17
Great Lakes Inherent Optical Properties
Geospatial Database (GLIOPGD)
Collection of the majority of optical properties of all the Great
Lakes made from 1997 to the present during NOAA/EPA
cruises
Developed in partnership with NOAA/GLERL
Database was essential to development of Color Producing
Agent (CPA), Great Lakes Primary Productivity Model
(GLPPM), plume and HABs algorithms
Time-series of changing optical properties in the Great Lakes as
a result of climate change, invasive species, and anthropogenic
forcing can be generated
GLIOPGD is used for algorithm validation
Soon to be available to community at www.gliopgd.org and
other portals such as NOAA Coastwatch
Incorporate into SeaBASS?
18
Ground-Truth/Calibration Issues
(cont.)
Share inherent optical properties – large data set - with
community
– www.gliopgd.org
– Available soon
3.5
Erie
3.0
Ontario
2.0
1.5
Michigan
Huron
cpg(440)
apg(440)
1.0
0.5
0.0
MI31B
MI48B
MI49B
MI53B
MI11
MI32
MI41
MI47
HU96B
HU06
HU38
HU48
ER58
ER91M
ER78M
ER95B
ER09
ON64B
ON12
ON49
ON60
-1
(m )
2.5
Example of CHL, FSS, VSS, TSS, CDOM, absorption,
& attenuation data now stored in relational IOP
geospatial database
Station
19
Moving Forward in the Great Lakes
with Algorithm Development
Optimize MODIS HABs algorithm
Utilize in situ values to calibrate plume model to provide
TSS and TSM concentrations
Validate Primary Productivity (PP) algorithm for Lakes
Erie, Huron, Ontario, and Superior
Estimate PP during Lake Stratification
Continue optimizing atmospheric correction procedure
Transition research algorithms to operational products
(NOAA Coastwatch)
20