ADAPTIVE OBSERVATIONS AT NCEP: PAST, PRESENT, AND FUTURE

1. INTRODUCTIONIn this paper we provide an overview of activities re-lated to the use of adaptive observations at the NationalCenters for Environmental Prediction (NCEP). First a briefsummary is given on past developments (section 2). Thisis followed by a discussion on current activities related tothe recent introduction of adaptive observations into theNational Weather Service (NWS) operations in the frame-work of the Winter Storm Reconnaissance (WSR) program(section 3). Section 4 is devoted to a short discussion of theimpact adaptive observations collected in the latest fieldprogram (Winter Storm Reconnaissance 2001, WSR01)had on forecast quality. A short description of planned ac-tivities is given in section 5, while section 6 offers someconclusions.2. HISTORICAL OVERVIEWTargeted observations, where data are collected inspecific areas at specific times with the aim of improvingthe quality of preselected Numerical Weather Prediction(NWP) forecast features in the extatropics has only a shorthistory (Toth et al. 2001). The idea was first dicussed pub-licly at a workshop in 1995 (Snyder 1996). Related re-search at several NWP centers has first organized arounda major field program, the Fronts and Atlantic Storm TrackExperiments (FASTEX, Joly et al. 1999). NCEP, in collabo-ration with Pennsylvania State University (PSU) scientists,contributed by developing the Ensemble Transform tech-nique (ET, Bishop and Toth 1999), later superseded by theEnsemble Transform Kalman Filter technique (ETKF,Bishop et al. 2000; Majumdar et al. 2001a), along with ad-joint and quasi–inverse linear techniques (Pu et al. 1997).In 1997 these techniques were used, in a collaborative ef-fort with Massachusetts Institute of Technology (MIT) sci-entists, in real time during the FASTEX field experimentsto identify observational areas for the release of drop-sondes by manned aircraft. Results from the use of theseadaptive observational techniques in FASTEX are re-ported in Toth et al. (1998), Szunyogh et al. (1999a), andPu and Kalnay (1999).The ET technique was further tested, along with anadjoint technique developed by the Naval Research Labo-ratory (NRL), in the following winter within the North PacificExperiment (NORPEX, Szunyogh et al. 1999b; Langlandet al. 1999). NCEP also participated in the California Land-falling Jets Experiment (CALJET, Ralph et al. 1998), andlater in the Pacific Landfalling Jets Experiment (PACJET)regional field programs by providing experimental mesos-cale adaptive observational guidance (Toth et al. 2000).The success of these early field experiments led to the es-tablishment of the Winter Storm Reconnaissance field pro-gram in 1999 (Toth et al. 1999).The aim of the WSR pro-gram is to reduce forecast errors for significant winterweather events over the contiguous US and Alaska in the24–96 hour lead time range through the use of adaptiveobservations over the data sparse northeast Pacific. Forthis purpose dropsonde data are collected by the AircraftOperations Center (AOC) of the National Oceanic and At-mospheric Administration (NOAA) and the 53rd WeatherReconnaissance Squadron of the US Air Force (USAF),through the use of manned aircraft operating out of Hono-lulu, HI, and Anchorage, AL.Verification results indicate that the majority of the tar-geted forecasts are significantly improved (Szunyogh et al.2000; 2001). Based on these results WSR became a regu-lar program (Toth et al. 2001). Recognizing the successand importance of wintertime adaptive observations, inJanuary 2001 NWS made the WSR program operational.In the following section we discuss some details of the tran-sition of the adaptive observational work from the researchinto the operational environment.3. TRANSITION FROM RESEARCH INTO NWS OP-ERATIONSAs discussed by Toth et al. (2001) adaptive observa-tions involves three major steps: (1) Selection of a target-ing case, i. e., the identification of date/time and location ofa threatening weather event for which forecasts are to beimproved; (2) Identification of observational time and areafrom where extra observations can most benefit the fore-cast aspect defined in (1) above (sensitivity calculations);and (3) Collection of atmospheric measurements from theobservational area defined in (2). Here we will discuss howthe transition from research to operational mode is affec-ting each of the three major steps of adaptive observationsin the WSR program.In the first year (1999) of the WSR program, research-ers at the Environmental Modeling Center (EMC) of NCEP,in collaboration with PSU scientists, carried out the first twosteps of adaptive observations, case selection, and sensi-tivity calculations. The program became operationally ori-ented in 2000, hence the case selection task was taken up