Water vapour above Switzerland over the last 12 years June Morland*, Martine Collaud**, Klemens Hocke*, Pierre Jeannet**, Christian Mätzler* *Institute of Applied Physics, University of Bern **MeteoSwiss 1
Overview Introduction Motivation Datasets used Homogenisation of radiometer IWV time series Comparison of monthly and diurnal climatology Comparison of trends Least mean squares analysis Mann Kendall analysis 2
Why is it important to measure water vapour? Greenhouse Gas Increase Earth warms + Increase in water vapour At ambient temperatures and constant relative humidity: water vapour increases by ~6 % per degree K increase in air temperature. ''Water vapour feedback acting alone approximately doubles the warming from what it would be for fixed water vapour.'' IPCC 2001: The Scientific Basis. But water vapour changes: On short time scales Seasonally Rapidly with altitude 3
Data used in the analysis TROWARA microwave radiometer, Bern (575 m) Radiosonde, Payerne (490 m) 00 and 12 UT soundings GPS Receiver Bern hourly measurements, all weather conditions Sun Photometer Bern daytime and sunny conditions only ECMWF operational analysis closest gridpoint to Bern, 850-200 mb 4
The TROWARA microwave radiometer TROWARA is on the roof of the IAP, Bern Measurements since 1994 but not continual. Observations at 21.3 and 31.4 GHz yield estimates of ILW and IWV. ILW Integrated Liquid Water (cloud) IWV Integrated Water Vapour Calibration by internal hot and cold loads and tipping curves. 5
TROWARA: changes in measurement setup From 1994 to April 2002 TROWARA was outside on the roof. Large temperature changes (13 K in summer) inside radiometer Rain gathered on lens Since November 2002 TROWARA has been indoors Diurnal changes now 2 K in summer Observation of IWV possible in light rain conditions Summer indoors Winter indoors Summer outdoors Winter outdoors 6
A look back in time with TROWARA: the uncorrected time series Water vapour has a strong seasonal cycle Model seasonal cycle plus trend component. Estimate IWV decrease of 0.14 mm per year - 10 % per decade! Not expected given Alpine temperature increase of 0.9 C over the 1995 to 2002 period (Philipona and Dürr, GRL, 31, 2004) 7
Bias in TROWARA hourly average data Bias in TROWARA IWV, mm Bias in TROWARA relative to other instruments 4 0 Amplifier change Field trip Voltage correction P1 A P1 B P2 P3 P4-4 Jan94 Jan08 P1A (January to April 1994) IWV bias -0.7 to -0.3 mm Gap in measurements P1B (April 1995 to August 2000) IWV bias +1.9 to 3.0 mm. Amplifier change P2 (May to June 2001) TROWARA bias +0.2 to 0.8 mm. P3 (Oct 2001 to Jan 2002) IWV bias of -0.6 to -2.6 mm. Instrument removal for campaign led to saturated hot load voltage P4 (Feb 2002 to June 2005) IWV bias of -0.7 to +0.4 mm. 8
Homogenisation of TROWARA data Use statistical technique to find breakpoints in data series and to estimate correction (Vincent, Journal of Climate, 11, 1094-1104, 1998). Reference series: Radiosonde (launched from Payerne, 491 m) IWV differences: Bantiger-Bern and Zimmerwald Neuchatel Check correction by comparing TROWARA, sonde, sun photometer and GPS IWV Sonde, launched from Payerne IWV = 377 m Bantiger, 942 m Neuchatel, 485 m IWV = 421 m Bern, 565 m Zimmerwald, 906 m Sun photometer GPS 9
Monthly climatology IWV, mm 25 20 15 10 IWV monthly climatologies 1996 to 2007 5 1 2 3 4 5 6 7 8 9 10 11 12 Average TROWARA IWV 14.4 mm Radiosonde is +0.3 mm higher GPS is +1.4 mm higher samples in rain ECMWF is 33 % lower start at 850 mb Bern avg(p)=952 mb Sun photometer (SPM) is 19 % lower samples only in sunny conditions 10
GPS IWV calculation 20 200 km + 2 m, N 2 O 2 + 0.25 m H 2 O ZTD Zenith Total Delay ZHD Zenith Hydrostatic Delay ZWD -Zenith Wet Delay IWV T m P v T dz P v T dz 2 K T m ZWD T m water vapour weighted mean atmospheric temperature ZWD ZTD ZHD Estimate T m from surface temp T s T m 70.2 0.72 T s 11
Diurnal climatology IWV, mm 15.4 14.4 IWV hourly climatology 2003 to 2008 13.4 0 5 10 15 20 TROWARA diurnal cycle amplitude 0.32 mm ECMWF 0.39 mm Radiosonde 0.87 mm Solar radiation error? GPS 0.32 mm IWV calculation requires mean atmospheric temp Tm estimate from surface temp, Ts Best results for Ts with damped diurnal cycle 12
TROWARA homogenised time series IWV, mm 45 TROWARA water vapour Jan 1994 to Nov 2008 Least mean squares fit to seasonal cycle and trend TROWARA homogenised time series show trend of 0.06 mmy -1 4 % per decade 0 Jan95 Jan00 Jan05 13
Least mean squares trend analysis Trend mmyr -1 0.16 0.1 0.0-0.04 Least Mean Squares Trend Comparison 1996 to 2007 Midnight All Midday All datasets show positive trends ECMWF and radiosonde show higher midnight than midday trend TROWARA shows opposite But TROWARA (2003-2008) trend agrees well with radiosonde Problem may be TROWARA outdoors from 1996-2002? 14
Mann Kendall Monthly Trend Analysis Trend mmyr -1 Mann Kendall Monthly Trend Comparison 1996 to 2007 0.8 0-0.8 1 2 3 4 5 6 7 8 9 10 11 12 All datasets show same trend direction except November Winter (Dec Mar) negative trends TROWARA and ECMWF significant decrease in Dec Apr Oct mainly positive trends Sep all three datasets show significant increase Aug negative trends Surface drying and less evaporation? 15
Conclusions and outlook Homogenisation of IWV time series from mw radiometer Monthly IWV climatology is dependent on instrument sampling characteristics (day/night, fair weather/rain) TROWARA diurnal IWV climatology agrees with GPS with improved estimate of mean atmospheric temperature All datasets with all times included show positive trends between 0.25 and 0.45 %y -1 Radiosonde midnight trend significant at 90 % level All datasets show IWV increase in Sep (95 % significance) More time needed to establish cause of trends Important to archive quality data 16
Acknowledgements NCCR Climate provided funding for research GPS Zenith Total Delay data from Federal Swiss Office of Topography Surface meteorological measurements and soundings from Meteo Swiss Thanks to Eddie Graham for helpful discussions Online water vapour database: http://www.iapmw.unibe.ch/research/projects/startwave/database/index.html 17