Search

The mixing layer height (MLH) indicates the change between vertical mixing of air near the surface and less turbulent air above. MLH is important for the dispersion of air pollutants and greenhouse gases, and assessing the performance of numerical weather prediction systems. Existing lidar-based MLH detection algorithms typically do not use the full resolution of the ceilometer, require manual feature engineering, and often do not enforce temporal consistency of the MLH profile. Given the large-scale availability of lidar remote sensing data and the high temporal and spatial resolution at which it is recorded, this domain is very suitable for machine learning approaches such as deep learning. This presentation introduces a completely new approach to estimate MLH: the Deep-Pathfinder algorithm, based on deep learning techniques for image segmentation.The concept of Deep-Pathfinder is to represent the 24-hour MLH profile as a mask (i.e., black indicating the mixing layer, white indicating the non-turbulent atmosphere above) and directly predict the mask from an image with lidar observations. Range-corrected signal (RCS) data at 12-second temporal and 10-meter vertical resolution was obtained from Lufft CHM 15k ceilometers at five locations in the Netherlands (2020–2022). High-resolution annotations were created for 50 days, informed by a visual inspection of the RCS image, the manufacturer's layer detection algorithm, gradient fields, thermodynamic MLH estimates, and humidity profiles of the 213-meter mast at Cabauw.Our model is based on a customised U-Net architecture with MobileNetV2 encoder to ensure fast inference times. A nighttime variable indicated whether the sample occurred between sunset and sunrise and hence, whether an estimate of the stable or convective boundary layer was required. Model calibration was performed on the Dutch National Supercomputer Snellius. First, input samples were randomly cropped to 224x224 pixels, covering a 45-minute period and maximum altitude of 2240 meters. Then, the model was pre-trained on 19.4 million samples of unlabelled data. Finally, the labelled data was used to fine-tune the model for the task of mask prediction. Performance on a test set was compared to MLH estimates from ceilometer manufacturer Lufft and the STRATfinder algorithm.Results showed that days with a clear convective boundary layer were captured well by all three methods, with minimal differences between them. The Lufft wavelet covariance transform algorithm contained a slight temporal shift in MLH estimates. Further, it had more missing data in complex atmospheric conditions. STRATfinder estimates for the nocturnal boundary layer were consistently low due to guiding restrictions in the algorithm. In contrast, Deep-Pathfinder followed short-term fluctuations in MLH more closely due to the use of high-resolution input data. Path optimisation algorithms like STRATfinder have good temporal consistency but can only be evaluated after a full day of ceilometer data has been recorded. Deep-Pathfinder retains the advantages of temporal consistency by assessing MLH evolution in 45-minute samples, however, it can also provide real-time estimates. This makes a deep learning approach as presented here valuable for operational use, as real-time MLH detection better meets the requirements of users in aviation, weather forecasting and air quality monitoring.

In 2011 the UK Met Office established the Weather Observations Website (WOW) initiative, a global-coverage project in which users of personal weather stations (PWS) can contribute their weather observations to a central repository. In this decade, more than 10,000 PWS around the world have contributed 2 billion measurements to this project, with a remarkable presence of WOW users in Europe. The Dutch Met Office (KNMI) joined this initiative as partner in 2015. In the past 8 years, 1,000+ PWS located in the Netherlands have collected 250+ million observations of the most relevant weather variables, and the interest of the Dutch public in this network continues growing. In this context, the KNMI has two main roles with respect to WOW-NL observations: Platform Facilitator and Data Researcher. The KNMI facilitates WOW-NL to the public via the portal http://wow.knmi.nl, which enables visualizing the latest observations in a map, allows querying to inspect the historical data contributed by each station, and provides a space for news. As platform facilitator, the KNMI aims for a measurement system of PWS that provides optimal added value to our science and services. The Data Research teams at KNMI have dedicated continuous efforts to develop quality controls (QC) enabling a full quality assessment of the WOW-NL observations. The latest results show that the application of QC methods yields promising results for air temperature, rainfall, and wind speed measurements. This means that WOW-NL observations may have sufficient quality to be incorporated into successive research or operational workflows and become part of the ‘daily business’ of the organization. These two roles are designed to work independently, but we believe that bringing them together would positively and effectively impact quality of data for the organization’s science and services. Hence, how can we interlace them most optimally in a feedback loop and take them to the next level? How can we expand the Platform Facilitator role, to stimulate and provide guidance for citizens to obtain quality of crowd sourced data most optimal for our science and services? How to enable the Data Researcher role to deliver peer-reviewed scientific content to a broader audience and in a real-world set up? Last but not least, how to establish a dialogue with the users to create a community ensuring long-term data provision for national meteorological services? In this work we investigate the relationship between the Platform Facilitator and the Data Researcher roles to balance investment in actions “upstream” (e.g. network design, PWS location) vs “downstream” (e.g. metadata, statistical QC procedures). We also elaborate on how the inclusion of WOW-NL in operational workflows might require revisiting or creating new policies for crowdsourced data or assessing the readiness of the digital infrastructure of the organization.

The mixing layer height (MLH) indicates the change between vertical mixing of air near the surface and less turbulent air above. MLH is important for the dispersion of air pollutants and greenhouse gases, and for assessing the performance of numerical weather prediction systems. Existing lidar-based MLH detection algorithms typically do not use the full resolution of the ceilometer, require manual feature engineering, and often do not enforce temporal consistency of the MLH. To address these limitations, a novel MLH detection approach has been developed based on deep learning techniques for image segmentation. The concept of our Deep-Pathfinder algorithm is to represent the 24-hour MLH profile as a mask and directly predict it from an image with lidar observations. Therefore, range-corrected signal data was obtained from Lufft CHM 15k ceilometers at five locations in the Netherlands that were part of the operational ceilometer network. Input samples of 224 × 224 pixels were extracted, each covering a 45-minute observation period. A customised U-Net architecture was developed with a nighttime indicator and MobileNetV2 encoder for fast inference times. The model was pre-trained on 19.4 million samples of unlabelled data and fine-tuned using 50 days of high-resolution annotations. Qualitative and quantitative results showed competitive performance compared to two benchmark models: the Lufft and STRATfinder algorithms. Existing path optimisation algorithms have good temporal consistency, but can only be evaluated after a full day of ceilometer data has been recorded. Deep-Pathfinder retains the advantages of temporal consistency but can also provide real-time estimates. This makes our approach valuable for operational settings, as real-time MLH detection better meets the requirements of users such as in aviation, weather forecasting and air quality monitoring.

Recent studies indicate that global warming changes the global hydrological cycle and may trigger drought or expand and deepen existing drought conditions at our planet. During the summer of 2018 the Netherlands experienced extreme drought conditions, matching the previous drought record from 1976. This climatic extreme has been monitored using a cumulative metric based on the difference between (potential) evaporation and precipitation. In an effort to provide exhaustive drought monitoring facilities, the Netherlands Meteorological Service (KNMI) developed a drought monitor based on the Standard Precipitation Index (SPI) using 40 years of daily rainfall (1971-2010) from our official network of rain gauges for calibration. The daily SPI maps help decision makers to assess the status of meteorological drought in the Netherlands, thus enabling preventive measures mitigating its negative impacts on different socio-economic sectors. In the past two decades our global society has witnessed the advent of new technological and scientific advances that have reshaped the way we collect weather observations. Increasing numbers of citizens are joining the effort of monitoring the weather by installing citizen weather stations (CWS) in private spaces (e.g., home, schools), thus conforming novel sources of weather data. In 2015, the KNMI joined as a partner the Weather Observations Website (WOW) consortium, a citizen science initiative promoted by the UK Met Office bringing together weather enthusiasts all around the world. WOW-NL CWS have collected 100+ million observations between 2015-2019. However, it is still unclear how to use this remarkable volume of observations, or what is the added value (e.g., economic, operational, research) they provide with respect to the official network. In this ongoing work, we combined the newly developed SPI drought monitor with WOW observations from the Netherlands to obtain an ‘SPI-WOW’ indicator. Our goal is threefold: 1) illustrating how to turn WOW-NL data into operational value; 2) assessing the possibility of providing higher resolution drought maps including WOW-NL rainfall data; 3) enable the possibility for underrepresented regions to obtain (relevant) local drought metrics. We extracted 12 million precipitation observations for 2019 and, for each day of the year, we computed the daily rainfall accumulations for the previous 30 days (i.e., SPI-1). Note that the precipitation observations are not quality-controlled (QC). The calibrated model is tested with these newly created rainfall accumulations to obtain the SPI-WOW values. Our preliminary results compare the official vs alternative values of SPI at the location of each WOW-NL CWS. For each month we observe a moderate positive correlation, and there are CWS in the network capable of providing measurements close to the official ones. Further work to achieve the above-mentioned goal should include a) the application of a QC to the rainfall data to remove the outliers beforehand; b) thoroughly comparing the values of both networks in space and time across different scenarios; c) identifying the WOW-NL stations providing the best SPI metrics and its characteristics; d) assess the inclusion of radar data for the hi-res maps.

In the Agile Way of Working (AoW), a group of developers jointly work to efficiently realize a project. Here we report on the application of AoW in meteorological research and development (R&D) outside of the software engineering environment. Three projects were formulated, derived from the observations strategy (2015) of the Royal Netherlands Meteorological Institute (KNMI). An initial phase of preparation consisted of breaking down the workload into tasks to be accomplished by individual project members and achievable in two one-week sprints. Sprints consisted of daily stand-ups, where accomplishments, work intentions, and obstacles were discussed, followed by project work in a joint working environment. The three projects identified were 1) flying a drone to detect boundary layer evolution, 2) monitoring the quality of the precipitation measurement system, and 3) realizing a platform for merging third-party data with meteorological observations. The preparation phase proved to be vitally important to each of the projects. The roles of the product owner and Scrum master in streamlining and guiding these projects were essential to the success of the sprint weeks, but the joint group settings worked well for only two of the three projects. While team members were positive about their experience with the AoW, the challenge remains to fuse the traditional individual work practice of researchers with that of software engineers, who are experienced in working in a group setting.

Inspection of wall thickness of the evaporator walls is usually performed from the inside of the boiler with standard ultrasonic (UT) spot testing. The costs for this type of inspection is largely determined by the cost for scaffolding and grinding and by safety measures which sometimes even include radiation control. The necessity for grinding also slightly impacts the accuracy of the readings. A new non-destructive testing application has been developed which provides the possibility to do this inspection from the outside of a boiler and thus eliminating the need for scaffolding. Windows of interest of about 10 to 120 ft2 are selected and provisions in the boiler insulation are made. The tubes in the selected areas are tested through UT scanning. The minimum wall thickness of each tube in that window is graphically presented in a report and repeat measurements enable the plant owner to compare these values in time. Trends in wall thickness reduction provide a basis for boiler life time prediction and to take preventive action. As a tool for preinspection to assess required maintenance measures (such as overlay welding or tube replacement), this test application has already shown to be very valuable. Recent applications include wall thickness assessment of refractory covered wall tubes, testing of boiler wall tubing with internal wall thickness loss at the fire sided tube face and wall thickness assessment of Inconel clad tubes at waste incinerators (EU project: NextGen Biowaste). This paper will discuss the method of non-destructive testing of wall thickness of the evaporator walls from the outside and will present examples where this method is applied.Copyright © 2010 by ASME

Stress corrosion cracking (SCC) in retaining rings of generators has been a major issue since large generator manufactures started to advise generator owners to replace and inspect retaining rings made from certain materials in the mid 1990’s. KEMA has been inspecting generator retaining rings with automated systems from the seventies on. Since 1995 new developments led to the possibility of also testing rings in situ, eliminating the need of removing the rotor from the stator and thus saving considerable down-time. Over the years, step by step improvements to the system and its application to generator retaining ring inspections have been implemented, one of which is the use of smaller probes and manipulators. Another improvement is the implementation of eddy current testing (ECT) to increase the detection capability for cracking at the outside surface in parallel to the automated ultrasonic testing (AUT). Experience obtained with this method contradicts some of the common ideas about reliability of other Non-destructive testing (NDT) methods, in particular the application of Dye Penetrant inspection (DPI) which appears to be less reliable. By performing fracture mechanics calculations to evaluate detected and sized flaw indications, predictions of remaining life of the rings can be made (Fitness-for-Purpose analysis). This paper presents results of applying the in situ retaining ring inspection system in the last 14 years and gives an overview of cases — some of those from recent inspections — where defect indications were detected, sized and evaluated. In some cases this resulted in recommendations to replace the retaining rings.Copyright © 2009 by ASME