A Method for Calibrating Deterministic Forecasts of Rare Events

Marsh, Patrick T.; Kain, John S.; Lakshmanan, Valliappa; Clark, Adam J.; Hitchens, Nathan M.; Hardy, Jill

doi:10.1175/waf-d-11-00074.1

Cited by 12 publications

(3 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The forecasts were verified for the 4 year period 2011–2014. Even though the forecast skill of mesoscale models has been evaluated in many studies globally (Coniglio et al , ; Schmidli et al , ; Marsh et al , ; Wen et al , ; Stratman et al , ; Berner et al , ), as well as over the Indian region (Rakesh et al , , , ; Goswami et al , ; Vishnu and Francis, ), very few studies have evaluated rainfall forecast skill for specific agricultural applications. Another important aspect of the present study is that the rainfall forecasts are verified against the observations at comparable resolution from the telemetric rain gauge (TRG) network installed and maintained by Karnataka State Natural Disaster Monitoring Centre (KSNDMC).…”

Section: Introductionmentioning

confidence: 99%

An evaluation strategy of skill of high‐resolution rainfall forecast for specific agricultural applications

Rakesh

Goswami

2016

Meteorological Applications

View full text Add to dashboard Cite

ABSTRACT:A strategy for validation of rainfall forecasts for specific agricultural applications is presented. The focus is mainly on the design of specific forecast advisories that are risk-free and useful in spite of their inherent errors. The strategy works for these specific applications because the forecast advisories are based on when NOT to irrigate or apply fertilizer/pesticide because rain is predicted (risk-free because wrong forecast only delays irrigation/application of fertilizer/pesticide within tolerance). Thus, unlike in conventional forecast evaluation, a forecast is considered as valid if the forecasted rain (or no rain) is correct for the day of the forecast (D0C) or the next day or the day after (designated D1C and D2C, respectively), as the farmer can afford to postpone the field application for a couple of days beyond the scheduled date. The methodology has been evaluated for rainfall forecasts over Karnataka (a state in southwest India with nearly 56% of the workforce engaged in agriculture). Here, forecast validation against rain gauge observations is presented at comparable resolutions for the southwest (June to September) and the northeast (October to December) monsoon seasons during 2011-2014. Analyses demonstrate that forecasts over several areas which may appear to be less reliable based on conventional evaluation (D0C) are found to have useful skill for the specific agro-applications as evident from evaluation based on D1C and D2C criteria. Our analysis shows that the evaluation strategy presented is effective during the non-rainy (January-May) season also. It is pointed out that such an approach can help to meet the challenges in designing and implementing best practices in agriculture by combining immediate gains for the end users with long-term sustainability.

show abstract

Section: Introductionmentioning

confidence: 99%

An evaluation strategy of skill of high‐resolution rainfall forecast for specific agricultural applications

Rakesh

Goswami

2016

Meteorological Applications

View full text Add to dashboard Cite

show abstract

“…The skill depends on various approximations, parameters, parameterization schemes as well as initial and boundary data (Jankov et al , ; Etherton and Santos, ; Wen et al , ). It is recognized, therefore, that generic mesoscale models that may be available in the public domain require significant calibration and validation of configuration for optimum performance over a region (Das et al , ; Coniglio et al , ; Skok et al , ; Rakesh et al , ; Marsh et al , ). Further, even for a given model (or more precisely, for a particular model configuration), the skill of rainfall forecast depends on the geographical location, season (background state) and the lead and resolution of the forecasts (Mass et al , ; Jones and Carvalho, , Das et al , ).…”

Section: Introductionmentioning

confidence: 99%

Evaluation of high resolution rainfall forecasts over Karnataka for the 2011 southwest and northeast monsoon seasons

Rakesh

Goswami

Prakash

2014

Meteorological Applications

View full text Add to dashboard Cite

Evaluation of real-time rainfall forecast at the hobli (village cluster ∼10 km 2 ) level from an atmospheric mesoscale model over Karnataka (located at southwest India, with nearly 56% of the workforce engaged in agricultural activities) for the southwest (June-September) and northeast (October-December) monsoon seasons of 2011 is presented in this study. The forecast system has been operationally implemented through data assimilation from a number of local observatories for effective applications as well as for proof of concept. A highlight of the study is the validation of the rainfall forecasts against observations at comparable resolution established by Karnataka State Natural Disaster Monitoring Centre, Government of Karnataka (KSNDMC). An evaluation of the forecast skill against daily observed rainfall is presented. A number of statistical evaluations show that the forecasts have enough skill to be useful for end users. A few areas of systematic bias and relatively higher forecast error were identified for further improvement in forecast skill.

show abstract

“…Instead, output from each ensemble member can be used by statistical post-processing algorithms, which are much less computationally expensive and can be run independently of the numerical models, in order to produce probabilistic predictions of high impact weather events. Previous projects have focused on probabilities of heavy rain using grid point [12] or local neighborhood data [13]. There has been no work focusing on deriving severe hail probabilities from these ensembles up to this point.…”

Section: Introductionmentioning

confidence: 99%

Severe Hail Prediction within a Spatiotemporal Relational Data Mining Framework

Gagne¹,

McGovern

Brotzge³

et al. 2013

2013 IEEE 13th International Conference on Data Mining Workshops

View full text Add to dashboard Cite

Severe hail, or spherical ice precipitation over 1 inch in diameter, has caused billions of dollars in damage to crops, buildings, automobiles, and aircraft. Accurate predictions of severe hail with enough lead time can allow people to mitigate some hail damage by sheltering themselves and their vehicles and by rerouting their aircraft. Current pinpoint forecasts of severe hail rely on detection of hail in existing storms with radar-based methods. Predictions beyond an hour are limited to probabilistic predictions over larger areas based on expected environmental conditions. This paper describes a technique that could increase the accuracy of severe hail forecasts by incorporating output from an ensemble of storm scale numerical weather prediction models into a spatiotemporal relational data mining model that would produce probabilistic predictions of severe hail. The spatiotemporal relational framework represents the ensemble output as a network of storm objects connected by spatial relationships. Composites of the ensemble data show spatial biases in the placement of severe and non severe hail storms. The spatiotemporal relational model performs significantly better at discriminating between severe and non-severe hail compared to a traditional data mining model of the same type. Variable importance rankings show results physically consistent with previous studies and highlight the importance of the relational data.

show abstract

A Method for Calibrating Deterministic Forecasts of Rare Events

Cited by 12 publications

References 22 publications

An evaluation strategy of skill of high‐resolution rainfall forecast for specific agricultural applications

An evaluation strategy of skill of high‐resolution rainfall forecast for specific agricultural applications

Evaluation of high resolution rainfall forecasts over Karnataka for the 2011 southwest and northeast monsoon seasons

Severe Hail Prediction within a Spatiotemporal Relational Data Mining Framework

Contact Info

Product

Resources

About