IRI@AMS 2016: Schedule of Events

From crowd-sourcing tornado data to teaching Harlem high-school students about climate change and climate justice, IRI scientists will be sharing a number of fascinating projects at the annual meeting of the American Meteorological Society (AMS) next week in New Orleans.  Below is a schedule of their presentations and posters. Presenting authors appear in bold.

Crowd-Sourcing the Storm: A New Approach for Obtaining and Collating Scientific Tornado Observations

John T. Allen

The scientific community has underutilized the use of visual documentation of severe storms by storm chasers. John Allen presents a case study from the May 31, 2013 El Reno tornadic supercell to demonstrate the high research value of this imagery. He also outlines the methods that can be applied to make such visuals scientifically useful. The El Reno Storm produced the widest tornado and potentially the strongest near-surface wind speeds ever recorded. The event was documented in real-time by research and operational radars, lightning detection networks, with still and video imagery also being recorded by a multitude of storm chasers. To collate available data, Allen and his colleagues created the El Reno Survey as a first-ever effort to crowd-source imagery from storm chasers and compile submitted materials in a quality-controlled, open-access research database. Allen will also demonstrate a novel online visualization tool that displays multi-perspective video and Phased Array Radar with user-selected geographic referencing.  Read the rest of the abstract.

Seasonal Predictability of Severe Thunderstorms Based on ENSO: Methodology and Evaluation of the 2015 Forecast

John T. Allen, Michael Tippett and Adam Sobel

Climate scientists can spot El Niño and La Niña conditions developing months ahead of time, and they use this knowledge to make more accurate forecasts of droughts, flooding and even hurricane activity around the world. Allen presents work that shows how El Niño and La Niña conditions can also help predict the frequency of tornadoes and hail storms in some of the most susceptible regions of the United States. The El Niño-Southern Oscillation has long been hypothesized to influence severe thunderstorm occurrence over the U.S., but limitations in the observation records, combined with large year-to-year variability have made demonstrating such a relationship difficult, particularly during spring, the peak hail and tornado season. Findings by Allen and his coauthors show that fewer hail and tornado events occur over the central United States during El Niño and conversely more occur during La Niña. Based on this relationship, Allen made a forecast for the 2015 tornado season back in March, which he will discuss and evaluate during his presentation. Read the rest of the abstract.

Enhancing National Climate Services for Malaria Control in Eastern Africa

Aisha Muhammad, Madeleine Thomson, Adugna Woyessa, Tufa Dinku, Rémi Cousin, Anthony Barnston, and Brad Lyon

The current El Niño provides a suitable test for the impact of malaria interventions in East Africa. Controlling malaria at a time of increasing malaria risk will provide substantive evidence of the effectiveness of current investments and shine light on additional strategies that are needed. The 1997/98 El Niño event, the strongest on record, produced above-normal rainfall and temperatures in Eastern Africa and was associated with widespread and devastating malaria epidemics across the region. The strong El Niño event currently underway is considered comparable in strength to the 1997/98 event, with predicted above-normal air temperatures around the tropical belt and an increased chance of above-normal rainfall in Eastern Africa during the October-December season, both of which may increase the climate suitability for malaria across the region. 

Aisha Muhammad gives an overview of a new data driven approach to decision-making and interventions made possible by the Enhancing National Climate Services (ENACTS) initiative. ENACTS combines rigorously evaluated weather-station observations from national meteorological agency archives and operational system with globally available satellite and model reanalysis data. The resulting products are then disseminated via ‘maprooms’ housed on meteorological agency websites or through direct transfer, and can be used for detailed research and local decision-making. Read the rest of the abstract.

Teaching Climate Change from a Climate Justice Perspective in Harlem

Evelyn Roman-Lazen, Alessandra Giannini

Roman-Lazen and Giannini developed an elective course on climate change in the science department at Frederick Douglass Academy (FDA), a public high school in Harlem. The motivation for the setting and approach was to connect local and global impacts of pollution – from increased asthma incidence in a low-income neighborhood in upper Manhattan to drought across the Sahel, the semi-arid southern edge of the Sahara desert. The majority of participants in the high school course are first-generation college-bound students from predominantly Black and Latino backgrounds. The main goal of the course is therefore to expose underrepresented youth to the complexities involving climate change and its impact on communities and ecosystems locally and globally. For the past three years, self-selected groups of 20 to 25 students have met five times a week and engaged in climate research, and in sustainable design and engineering projects cast in a social/climate justice perspective to help foster environmental stewardship. The course has been so successful and popular that it is now a stable part of the school electives roster. The school is looking into using this model to design and implement other science courses, as well as train science teachers on how to incorporate some of these hands-on strategies of teaching into all of their classes. Read the rest of the abstract.

January 2015 Malawi Floods from a Remote Sensing Perspective

Andrew Kruczkiewicz, Jerrod Lessel, Erin Coughlan de Perez, Pietro Ceccato

In January 2015, extended periods of extreme rainfall caused a series of flood events throughout southern Africa. With little or no warning, the floods took communities by surprise, resulting in 276 deaths and the displacement of more than 230,000 residents. The meteorological situation was complex, with both riverine floods and flash floods occurring in various parts of the region within a short time. 

Disaster management groups and humanitarian organizations responded using information available to them in the days after the flood, in the form of flood maps and anecdotal evidence. These organizations relied on remotely-sensed satellite data to evaluate initial disaster impact and design response programs. 

Kruczkiewicz and his colleagues worked with the Malawi Red Cross National Society to study how the locations of shelters and origins of displaced people compared with flood distribution data derived from satellites as well as other information.

Their findings suggest that flood signal variations across flood map can be great, creating a situation of potential uncertainty when rapid response actions need to be taken by humanitarian and government organizations. Furthermore, relative to ground truth data, certain maps able to detect certain types of floods at a higher skill than others. Also, the relationship between soil moisture and floods vary based on flood type, with antecedent soil moisture conditions potentially acting as a primary driver in the prediction of heightened risk of flash floods, while appearing to be of little importance in the prediction of river floods.

In the same way that the geophysical characteristics of flood vary spatially and temporally by flood type, so too does the linked humanitarian action. The results of this study will increase the ability to forecast and monitor flood events in Africa, benefiting organizations involved with disaster preparedness and relief efforts, with a specific emphasis on providing analysis for the development of forecast based financing mechanisms to release funding before a disaster based on pre determined triggers and actions. Read the rest of the abstract.

Statistical Significance: The Wrong Answer to the Wrong Question

Simon Mason

Statistical significance tests have become a widely adopted procedure for assessing whether a particular research result is “meaningful” or “correct”. The basic idea involves calculating a test statistic–perhaps a correlation, or a difference in means–and then calculating the probability that the result could have been equaled or bettered by chance. If this probability is sufficiently low (less than 5%, or in some cases 10%) then the result is considered sufficiently strong to be proof of …, well at this point the logic typically starts to become a bit hazy and questionable! What does a significance test actually tell us, and is what it tells us even interesting? Problems with significance testing are beginning to be seen as so egregious that some journals, notably in the statistics and medical literature, will not publish articles that use them. The problems are so bad that it can be demonstrated that most claimed research findings are false (although results in some disciplines are more susceptible to being falsely positive than in others). Why is this true? Some of the problems with significance tests are widely recognised (correlation does not imply causation, for example), but others apparently are less widely acknowledged. In this presentation I will detail what the p-value does mean, and why it remains possible for such a large proportion of published research results to be false despite apparently rigorous significance testing. I will outline the reasons why significance testing should be discouraged, pointing to two main issues: the fact that the p-value does not really address the question we are ultimately interested in, and secondly that the tests are invariably invalid whether because of violated assumptions and/or because of inherent biases in the way science proceeds (we are much more inclined to look for relationships between two or more sets of data than to demonstrate that such relationships do not exist). Of course, the fact that statistical significance testing does not work very well is not an excuse for ignoring the questions we are falteringly trying to address with them, and so some alternative procedures for assessing whether our research results are “meaningful” will be proposed. Read the rest of the abstract.

Exploration of the ENSO-Sahel Relationship on Intraseasonal-to-Interannual Timescales and its Relevance to Local Climate Services

Catherine Pomposi, Alessandra Giannini, Y. Kushnir

Previous studies have identified and elucidated a relationship between the El Niño Southern Oscillation (ENSO) climate mode in the Eastern Tropical Pacific and rainfall in the West African Sahel, the semi-arid grassland which sits directly south of the Sahara desert and exhibits large variability on a number of timescales. In particular, during warm events (El Niño), the Sahel is expected to be anomalously dry while during cold events (La Niña), rainfall anomalies are of the opposite sign and the region tends to be wetter than normal. However, the seasonal total of rainfall that occurs in West Africa during an ENSO event cannot simply be understood in terms of thinking of the magnitude and warm or cold nature of the event itself. For example, during the extraordinary 1997 El Niño event, the Sahel was only modestly dry, and some of the strongest La Niña years (e.g. 1973 & 1975) were not the wettest seasons for the Sahel during the 20th Century. It is with inconsistencies such as these in mind that the Sahel’s behavior during ENSO events is studied with a diagnostic framework, utilizing newly available high-resolution observations such as the CHIRPS dataset. Specifically, we complete this study with the following objectives, to study whether the precipitation response in the Sahel is different when El Niño is growing versus when it is on the decline, and to understand whether the state of the Tropical Atlantic can modulate the ENSO teleconnection to the Sahel. Results from this work have particular relevance for informing seasonal forecasting applications as these forecasts are primarily made using the current state of ENSO to inform local experts on the probability of having a normal, above normal, or below normal rainfall season. An example of how the increased knowledge of ENSO’s effects on Sahel rainfall may be used will be discussed by detailing current efforts of the Agence Nationale de l’Aviation Civile et de la Météorologie (ANACIM) du Sénégal, host to the lead author’s visit in country to work on delivering climate information and build resilience to climate variability locally. 

Read the rest of the abstract.

Towards a Forecast Based Financing Framework to Trigger Humanitarian Action Before Floods

Andrew Kruczkiewicz, Jerrod Lessel, Erin Coughlan de Perez, Pietro Ceccato,

Forecast based financing is a novel approach to automatically trigger pre-established humanitarian actions based on forecasts and observations at various timescales. In Africa, this system will be of particular interest to humanitarian and government disaster managers as preparedness actions prior to a hydro-meteorological disasters can be ad hoc. 

In the days after heavy flooding in southern Africa in January 2015, humanitarian organizations and government level disaster mangers made decisions using information available to them, many time relying on satellite and flood model driven flood maps and anecdotal evidence from contacts in the field. Many maps were available, however, decision-making was hampered by uncertainty in the validity of the flood maps and challenging communication with the most impacted regions. 

Kruczkiewicz and his colleagues worked with the Malawi Red Cross National Society to study how the locations of shelters and origins of displaced people compared with flood distribution data derived from satellites as well as other information.

They find that in the same way the geophysical characteristics of flood vary spatially and temporally by flood type, so too does the linked humanitarian action. The results of their study will increase the ability to forecast and monitor flood events in Mozambique, Malawi and across Africa, benefiting organizations involved with disaster preparedness and relief efforts, with a specific emphasis on providing analysis for the development of forecast based financing mechanisms to release funding before a disaster based on pre-determined triggers and actions. Read the rest of the abstract.

Linking Past, Present and Future Climate Change to Adaptation in the African Sahel

Alessandra Giannini

Enhancing National Climate Services for Development in Africa

Tufa Dinku, Andrew Kruczkiewicz, Pietro Ceccato, Rémi Cousin, John del Corral, Madeleine Thomson, Aisha Muhammad.

The ENACTS (Enhancing National Climate Services) initiative is an ambitious effort to simultaneously improve the availability, access and use of climate information by working directly with national meteorological and hydrological services. It enables these agencies to provide enhanced services by overcoming the challenges of data quality, availability and access – while at the same time fostering stakeholder engagement and use. 

By integrating ground-based observations with proxy satellite and other data, ENACTS products and services introduce quality-assessed and spatially complete data services into national meteorological agencies to serve stakeholder needs. One of the strengths of ENACTS is that it harnesses all local observational data, incorporating high definition information that globally produced or modelled products rarely access. The resulting spatially and temporally continuous datasets allow for the characterization of climate risks at a local scale , and potentially offer a low-cost, high impact opportunity to support applications and research. 

ENACTS has so far been implemented in Ethiopia, Madagascar, Tanzania (including Zanzibar), Zambia, Rwanda, Ghana, Mali and The Gambia at national levels, and at regional level for the CILSS countries (West African Sahel). The ENACTS initiative strengthens policy analysis, relevant for multiple sectors, by providing relevant data with national coverage with much greater accuracy at smaller spatial and temporal scales. The climate products developed through ENACTS (which include a historical time series, routine monitoring and forecast products) are disseminated in “Maprooms” via the web sites of the national meteorological agencies using the powerful IRI Data Library software http://iridl.ldeo.columbia.edu. Products are currently being tested for use in impact assessment and forecasting in agriculture, health and disasters. Read the rest of the abstract.

Australian Tornadoes: Climatology 1795-2014 Compared to a ‘Record’ 2013

John T. Allen

A new climatology for the occurrence of tornadoes in Australia has been developed for the period 1795 to 2014, the second largest single country record. However, extensive media coverage in 2013 raised the question ‘Was 2013 a record tornado year?’ Like many places outside of the United States, the historical records for tornadoes are poorly documented. Existing data from the Australian Bureau of Meteorology National Severe Storms Archive also suffer from observer-driven spatial limitations, and biases related to institutional policy of event documentation. Recently, extensive library archives of scanned newspapers, and digitization of the original severe thunderstorm reports material have become available for Australia that can offer insight into historical events and extend the existing climatology. 

Keyword optimization was used to identify tornadoes from the scanned data while reflecting changes to terms used in the historical vernacular. Additional metadata relating to intensity, time of occurrence, path characteristics, injuries, fatalities and damage were inferred from newspaper accounts. Further, tornadoes from the existing Severe Storms Archive were cross-validated and additional metadata determined for inclusion in the new climatology. Based on documentary evidence, tornadoes were rated via the Fujita scale using three categorizations to reflect uncertainty in historical strength determination (Weak F0-F1, Strong F2-F3 and Violent F4-F5). The quality of record for each identified event was categorized into three levels (Possible, Likely or Definite) based on the reliability of observations, as well as documentation of characteristics indicating the presence of a tornadic event. 

The climatology in context of a recent observed year (2013) will be presented, highlighting that the annual frequency of tornadoes in Australia ranges between 30 and 80 observed tornado events per year but likely underestimates the total frequency given underreporting due to population density. Numerous tornado outbreak cases have also been identified throughout the length of the record. To further illustrate the risk posed for Australia by tornadoes, cases from 2013 encompassing the broad spectrum of tornado formative environments will be discussed. These results reveal that Australia is subject to tornadoes from most environmental sources on a relatively frequent basis, and this should play a greater role in the forecasting and warning process.

Read the rest of the abstract.

Forecasts of “Normal”

Simon Mason

The difficulty of forecasting “normal” weather and climate conditions is demonstrated in the context of bivariate normally distributed forecasts and observations. Functional relationships are derived between many deterministic and probabilistic verification scores and the skill of the forecasts in the bivariate normal setting as measured by Pearson’s correlation. The deterministic and probabilistic skill scores for the “normal” category are less than for the outer category for all but perfect models. There are two important mathematical properties of the normal category in a three-category climatologically equiprobable forecast system that affect the scores for this category. Firstly, the normal category can achieve the highest probability less frequently than the outer categories, and far less frequently in contexts of weak to moderate skill. Secondly, there are upper limits to the probability the normal category can reach. Read the rest of the abstract.