Satellite-Based Detection of Fog and Very Low Stratus A High-Latitude Case Study Centred on the Helsinki Testbed Experiment J. Cermak 1, J. Kotro 2, O. Hyvärinen 2, V. Nietosvaara 2, J. Bendix 1 1: Laboratory for Climatology and Remote Sensing (LCRS), Marburg/Germany 2: Finnish Meteorological Institute (FMI) 7 September 2006 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 1 / 25
Outline 1 Introduction Fundamentals The Problem 2 Approach and Data Helsinki Testbed Satellite Algorithm 3 Tests and Analysis Data Comparison Conclusions Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 2 / 25
Motivation Introduction Fundamentals Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 3 / 25
Aims Introduction Fundamentals Fog detection system should offer High spatial resolution (fog margins) High temporal resolution (nowcasting) Long time series (climatology) Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 4 / 25
Aims Introduction Fundamentals Fog detection system should offer High spatial resolution (fog margins) High temporal resolution (nowcasting) Long time series (climatology) Only geostationary satellite data suited! Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 4 / 25
Introduction Fundamentals Satellite system: Meteosat 8 SEVIRI 3 km spatial resolution 15 min repeat rate 11 spectral channels Operational since 2004 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 5 / 25
High Latitudes: Finland Introduction The Problem Latitude southern Finland: 61 Satellite zenith angle: 70 Reduced spatial resolution Low solar elevation Enhanced atmospheric effect Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 6 / 25
Introduction High Latitudes: Difficulties The Problem Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 7 / 25
Introduction The Problem High Latitudes: Difficulties Sensor sensitivity 1 0.8 0.6 0.4 0.2 1.6 1.2 0.8 0.4 CO 2 absorbance 0 3.4 3.6 3.8 4 4.2 4.4 4.6 0 Wavelength (µm) Terra MODIS ch. 20 NOAA 16 AVHRR ch. 3 Meteosat SEVIRI ch. 4 CO 2 absorbance Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 7 / 25
Introduction High Latitudes: Difficulties The Problem Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 8 / 25
Research Question Introduction The Problem Problem Is the detection of fog and very low stratus possible from a geostationary platform at high latitudes? To be investigated with detailed measurements. Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 9 / 25
Helsinki Testbed Approach and Data Helsinki Testbed Figure: J. Poutiainen, 2006 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 10 / 25
Helsinki Testbed Approach and Data Helsinki Testbed 2005 2007 Mesoscale weather research System integration Model development Data and info on www.fmi.fi/testbed Figure: J. Poutiainen, 2006 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 10 / 25
Measurements Approach and Data Helsinki Testbed Targeted measurements for Nowcasting applications Precipitation type Sea breeze Convection Stable boundary layer Figure: J. Poutiainen, 2006 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 11 / 25
SOFOS Approach and Data Satellite Algorithm SOFOS Satellite-Based Operational Fog Observation Scheme Fog as a cloud in the water phase low above ground with a stratiform surface impairing visibility at the ground. Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 12 / 25
SOFOS Approach and Data Satellite Algorithm SOFOS Satellite-Based Operational Fog Observation Scheme Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 13 / 25
SOFOS Approach and Data Satellite Algorithm SOFOS Satellite-Based Operational Fog Observation Scheme Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 14 / 25
Tests and Analysis Test Procedure: Data Selection Data Comparison 12 days, November 2005 February 2006 Data availability Data quality Includes fog / low stratus situations Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 15 / 25
Tests and Analysis Data Comparison Test Procedure: Point/Pixel Comparisons 23 24 25 26 27 Ground station observations Low clouds High clouds Clear 61 60 Satellite classification Low water clouds Other clouds Clear 61 60 Visibility and ceilometer data for low cloud presence Compared with satellite classification (no cloud / very low stratus cloud / high cloud) 59 23 24 25 26 27 59 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 16 / 25
Tests and Analysis Data Comparison Test Procedure: Contingency Tests Observation Yes Observation No Classification Yes A B Classification No C D Hit Rate: HR = A A+C [0 1] False Alarm Rate: FAR = B A+B [0 1] Critical Success Index: CSI = A A+B+C [0 1] Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 17 / 25
Test Results: By Day Tests and Analysis Data Comparison 1 0.8 Indicator Value Best Worst HR 0.66 1.00 0.52 FAR 0.25 0.00 0.67 CSI 0.54 1.00 0.31 n 9374 Hit rate 0.6 0.4 0.2 0 0 0.2 0.4 0.6 0.8 1 False alarm rate Figure: Results by day Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 18 / 25
Tests and Analysis Test Results: By Station Data Comparison 1 0.8 Indicator Value Best Worst HR 0.66 0.78 0.51 FAR 0.25 0.00 0.60 CSI 0.54 0.77 0.36 n 9374 Hit rate 0.6 0.4 0.2 0 0 0.2 0.4 0.6 0.8 1 False alarm rate Figure: Results by station Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 19 / 25
Analysis Tests and Analysis Data Comparison Possible reasons for misclassifications: Missed situations: Cloud overlap Low cloud cover False alarms: Station data not representative of pixel Station data quality (Visibility and ceiling don t always agree) Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 20 / 25
Tests and Analysis Test Results: CSI Distribution Data Comparison 23 24 25 26 27 0.75 1.00 0.50 0.75 0.25 0.50 0.00 0.25 61 61 Lowest category not present No obvious spatial pattern 60 60 59 23 24 25 26 27 59 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 21 / 25
Tests and Analysis Data Comparison European Context: CSI Distribution 60 350 0 10 20 60 0.75 1.00 0.50 0.75 0.25 0.50 0.00 0.25 50 50 583 METAR stations 24 days 2 nd half 2005 1030 satellite scenes No latitudinal effect! 40 40 0 10 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 22 / 25
Conclusions Tests and Analysis Conclusions High latitude detection of very low stratus works well Small-scale situations may be missed Coordinated measurement efforts useful Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 23 / 25
Outlook Tests and Analysis Conclusions 60 350 0 10 20 60 0.75 1.00 0.50 0.75 0.25 0.50 0.00 0.25 50 50 Ground fog will be tested for Helsinki testbed Qualitative analysis of cases to follow 40 40 0 10 Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 24 / 25
Contact Tests and Analysis Conclusions Helsinki Testbed: www.fmi.fi/testbed Laboratory for Climatology and Remote Sensing: www.lcrs.de cermak [at] lcrs.de Cermak et al. (LCRS, FMI) High Latitude Fog Detection 7 September 2006 25 / 25