Lake surface temperature as a proxy of climate change: satellite observations versus multi probe data

(1)

Lake surface temperature as a proxy of climate change

Satellite observations versus multi probe data

S. Pareeth

1,2,4

, M. Metz

1 , D. Rocchini

1 , N.Salmaso

2 , R. Adrian

3,4

, and M. Neteler

1 Introduction - WarmLakes

Sensors and Thermal data

_Methods

1

2

3

4

5 Lakes under

study

1. Lake Garda

2. Lake Iseo

3. Lake Como

4. Lake Maggiore

5. Lake Trasimeno

●

Study the long term warming trends of Sub-Alpine lakes using surface

temperature derived from satellite data

●

Leveraging the availability of daily thermal imageries for last 2 decades from

multiple sensors aboard satellites

●

Lake specific validation and model development using field data

●

Develop daily homogenized Lake Surface Water Temperature (LSWT) for last

2 decades.

●

Time series analysis linking the trend with climatic tele-connection index -

Eastern Mediterranean Pattern (EMP)

●

EMP explains well, variation in water surface temperature over Mediterranean

area than NAO (North Atlantic Oscillation)

●

Presenting here the preliminary results for Lake Garda

Global products for daily Surface Temperature

~ 2014 AVHRR June 1991 April 2012 2000 2014~ ATSR/A(A)TSR MODIS 4:36 and 16:36 local solar time

~

0130 and 1330 , 10:30 and 22:30 local solar time

10:00 and 22:00 local solar time

~

1980,s ₁₉₉₈

Geocoding issues

MODIS Land Surface Temperature (LST) products

– MOD11A1, MYD11A1 @ 1km , daily 2 observations, from 2002 – Covers all the lakes globally

– 1km spatial resolution

– https://lpdaac.usgs.gov/products/modis_products_table

MODIS Sea Surface Temperature (SST) products

– 4 km spatial resolution, daily 2 observations, from 2002 – few lakes

– http://oceancolor.gsfc.nasa.gov/

AVHRR pathfinder SST products

– 4 km spatial resolution, daily

– Not suitable for Sub-alpine Lakes – longest time series (from 1981)

ArcLakes – Lake Surface Water Temperature(LSWT) from ATSR/AATSR

– 0.05 degrees, 1995 – 2012, daily

– by School of Geosciences, University of Edinburg – daily recostructed data, day and night

– covers1600 lakes globally

– http://www.geos.ed.ac.uk/arclake/

Raw thermal data from

MODIS;A(A)TSR;AVHRR Brightness temperatures

Global LST/SST products

Optimized split window SST algorithm for Lakes Lake Surface Water Temperature (LSWT)

Level 1

(Calibration)

G.C. Hulley et al. / Remote Sensing of Environment 115 (2011) 3758–3769 Lake/Sensor specific coefficients (clear sky)

Level 2

Cloud mask /QC layers Statistical Reconstruction Methods

Gap filled seamless Time series

data set

Level 3

Validation/Model development using field data

Modeled Time series of LSWT

Level 4

Cloud mask /QC layers

Mutiple Spatio-temporal regression approach using secondary datasets (Metz et.al, 2014)

– Daily four observations of MODIS LST at 250m for entire Europe – Using hierarchial temporal and spatial interpolation

– Regression model using climatology parameters, Digital Elevation Model (DEM) etc

Harmonic ANalysis of Time series (HANTS) (Roerink et.al)

– Fourier Analysis

– Temporal interpolation

– Implemented in GRASS – “r.hants”

15 17 20 22 25 °C 15 17 20 22 25 °C 0.5 1.2 1.8 2.4 3.0 °C

2003 Summer Mean 2008 Summer Mean

2003-2008 Summer Mean difference

- 2003 – 2013 - MODIS SST

- Reconstructed using HANTS

October 27 2006, MODIS-Aqua – 13:30; Metz et.al, 2014

Raw LSWT Reconstructed LSWT Difference map (raw-reconst)

Results

Reconstruction of time series

Validation

Shortcomings

 LST algorithm not suitable for

water surface

 Gaps in time series due to clouds and bad raw data

 Coarse spatial resolution of

the available products

 Scope of using lake/sensor specific coefficients to derive Lake Surface Water

Temperature (LSWT)

Gap Reconstruction Algorithms

GRASS

Open source data processing environment

http://grass.osgeo.org/ http://r-project.org/ LST vs SST (Clear Sky) 7 9 11 13 15 17 19 21 23 25 5 10 15 20 25 f(x) = 1.05x - 0.32 R² = 0.98 Field data M O D IS S S T - 2003 – 2013 - MODIS SST

- Monthly field data

5 7 9 11 13 15 17 19 21 23 25 f(x) = 0.94x + 0.36 R² = 0.95 M O D IS L S T - 2003 – 2013 - MODIS LST

- Monthly field data

-1.5 -1 -0.5 0 0.5 1 1.5 7 9 11 13 15 17 19 21 23 25 5 10 15 20 25 f(x) = 1.03x - 0.09 R² = 0.96 2003 - 2013 monthly Field data M O D IS L S W T -2 -1.5 -1 -0.5 0 0.5 1 1.5 Garda

Trend of daily mean deviation

Lake Garda- MODIS LSWT, 2003 – 2013 daily

Garda

Lake Garda- A(A)TSR - ArcLakes LSWT, 1995 – 2011 daily

p = 0.13 Sen slope = 0.02 p = 0.002 Sen slope = 0.06 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 7.5 8 8.5 9 9.5 10 10.5 f(x) = - 0.35x + 9.2 R² = 0.34 EMP vs LSWT EMP M o d e lle d A T S R L S W T

EMP vs Winter Average Temp from LSWT

Conclusions/Future directions

References

● Thermal imageries from Satellite sensors can measure LSWT accurately, with an overall R

squared above 0.90

● SST algorithms are more accurate for water surfaces than LST based data

● Out of the available products MODIS SST and ArCLakes data (A(A)TSR) performs well. ● Products are of very coarse resolution, not suitable for inter lake studies

● The available products does not have the data for all the sub/alpine lakes

● The negative correlation of satellite derived winter averages with EMP is proven trend

explaining the local variation

● Extending back the database to begin from 1980 using NOAA AVHRR data ● Extending the study to other sub/alpine lakes

● Homogenizing the LSWT data based on Split window SST algorithm and optimized lake specific coefficients (Hulley et.al, 2011) to develop daily LSWT from 1980 - 2013

Metz, M.; Rocchini, D.; Neteler, M. Surface Temperatures at the Continental Scale: Tracking Changes with Remote Sensing at Unprecedented Detail. Remote Sens. 2014, 6, 3822-3840

Glynn C. Hulley, Simon J. Hook, Philipp Schneider, Optimized split-window coefficients for deriving surface temperatures from inland water bodies, Remote Sensing of Environment, Volume 115, Issue 12, 15

December 2011, Pages 3758-3769, ISSN 0034-4257

Roerink, G. J., Menenti, M. and Verhoef, W., 2000. Reconstructing cloudfree NDVI composites using Fourier analysis of time series. International Journal of Remote Sensing, 21 (9), 1911-1917

Salmaso N (2012) Influence of atmospheric modes of variability on a deep lake south of the Alps. Clim Res 51:125-133

MODIS - Moderate Resolution Imaging Spectroradiometer, NASA A(A)TSR - Advanced Along-Track Scanning Radiometer, ESA AVHRR - Advanced Very High Resolution Radiometer, NOAA

1_{GIS and Remote Sensing unit, Department of Biodiversity and Molecular Ecology,}

The Research and Innovation centre (CRI), Fondazione Edmund Mach (FEM), Trento, Italy

2_{Limnology and River Ecology unit, Department of Sustainable Agro-Ecosystems and Bioresources,}

The Research and Innovation centre (CRI), Fondazione Edmund Mach (FEM), Trento, Italy

3_{Department of Ecosystem Research, Leibniz Institute of Freshwater Ecology and Inland}

Fisheries (IGB), Muggelseedamm, Berlin, Germany

4_{Department of Biology, Chemistry and Pharmacy, Freie Universität, Berlin, Germany}