geopredict predictive knowledge factory and forecasting services

Keyboard Arrow Down

Climate Forecasting


01 |

Unbiased Climate Intelligence from data.

Climate Modeling and Forecasting

Chart Line

The CLIMFOR Climate Solution

There is an ever-growing need for understanding the climate through modeling which will aid in future forecasting of climate for compounded events that has impact on life and environment on earth. 

Many of these climate models are general circulation models (GCMs) that are domain and theoretical based physical models using energy and mass balances with their complexities for weather and climate modeling and forecasting. It is difficult with such approach for forecasting seasonal to sub-seasonal and upto decades the weather and climate forecasts for global as well as for a specific location with high accuracy.

geopredict believes that a data driven climate modeling and forecasting approach is one of the best ways for such seasonal to sub-seasonal and decadal forecasting and through its technology it would also be complimentary for checking the domain-based modeling and forecasting systems.

Mathematical modeling is at the core of many decision support systems. Like many real-world problems the earth climate system is ill-defined and can be characterized by: 

  • insufficient a priori information about the system for adequately describing the inherent system relationships, 
  • possessing a large number of variables, many of which are unknown and/or cannot be measured, 
  • noisy data available in very small to very large data sets, 
  • vague and fuzzy objects whose variables has to be described adequately.

For ill-defined systems the classical hard approach that is based on the assumption that the world can be understood objectively and that knowledge about the world can be validated through empirical means needs to be replaced by a soft systems paradigm which can better describe vagueness and imprecision. 

Climate systems and in general any system can be modeled either by deductive logical-mathematical methods (hard or theory-driven approach) or by inductive modeling methods (soft systems or data-driven approach). Deductive methods have been used to advantages in the cases of well-understood problems and which obey well-known principles. The spectacular results in aerospace are prime examples of this approach. Here, the theory of the object being modeled is well known and it obeys the known physical laws. However, they are unduly restrictive to one end because of insufficient a priori knowledge, complexity and the uncertainty of the objects, as well as the exploding time and computing demands.

In contrast, inductive methods are used when data-driven approaches is the only alternative. These models are derived from the real physical data and they represent the relationships implicit within the system without prior knowledge of the physical processes or mechanisms involved. 

With these advanced possibilities with inductive methods, geopredict has developed an AI System, called CLIMFOR system, that is entirely based on SELF-ORGANIZING, HIGH-DIMENSIONAL KNOWLEDGE EXTRACTION TECHNOLOGIES to predict atmospheric parameters over different time horizons, that is used for Climate modeling and forecasting at different spatial temporal levels from a specific regional level to higher global levels.

The Challenge

02 |

Climate Forecasting services

geopredict offers a wide range of Climate Forecasting Services using CLIMFOR Technology.

Global as well as Location Specific Atmosphere Forecasts for

  • Global Temperatures
  • Green House Gases
  • Extreme weather 
  • Disaster risk management
  • Drought and Flooding

The forecast time horizon of few days, months, years to decades ahead are provided as services.


— A

Forecasting of Global Temperature

geopredict developed a global atmosphere model with CORE MODELING CLIMFOR system, that describes a single-node non-linear dynamic system model of the atmosphere for decadal monthly forecasting consisting of 5 drivers: ozone concentration, aerosol index, radiative cloud fraction, and global mean temperature as endogenous variables and sun activity as exogenous variable of the system. The model was built using, observational station data (HADCRUT, GISS), and SAT data (RSS, UAH), from October 1988 till April 2011 constituting of about 1000 input variables, which is a typical input space dimension for complex dynamic climate systems modeling.

Using this CLIMFOR model, we calculated a 24 months ex-ante prediction of the global temperature anomalies from May 2011 to April 2013. The ex-ante prediction accuracy from CLIMFOR model relative to the true observed station data is 84 %. For the same time period, we also compared our forecast with the moderate IPCC A1B forecast scenario published in its Assessment Report 4. This 100-years scenario trend shows an accuracy of 34% relative to the actual temperature data. This accuracy improvement from geopredict’s forecast demonstrates the high performance of our CLIMFOR system and underlines the CURRENT PREDICTIVE INFORMATION GAP.

global temperature ex post forecast and IPCC A1B projection

CLIMFOR ex post forecast (green; January 2002- April 2011; P50 percentile) of geopredict’s self-organized system model (as of April 2011) vs observed values (gray; HADCRUT3) vs IPCC A1B projection (dark green) published in 2007 in the IPCC AR4 report.

global temperature ex ante forecast and IPCC A1B projection

CLIMFOR ex ante forecast (green; May 2011 - April 2013; P50 percentile) vs observed values (gray) vs IPCC A1B projection (dark green).

— B

GHG Forecasting

monthly mauna loa atmospheric CO2 forecast

Atmospheric CO2: CLIMFOR ex ante forecast ranges from June 2011-June 2030. The period for forecast evaluation is June 2011-August 2022 with an ex ante forecast accuracy of 99.6%.  

In June 2011 geopredict developed a predictive model for long-term monthly ex ante forecast of atmospheric CO2 concentration based on the observed monthly Mauna Loa CO2 records measured at an altitude of 3400 m. The inter-annual fluctuations of CO2 concentration are due to seasonal variations in CO2 consumption by plants. In the northern hemisphere, there is a much larger forest area than in the south which is why more CO2 is removed from the atmosphere during northern hemisphere summer.

As of August 2022 the ex ante forecast accuracy (134 months) is 99.6% which demonstrates that CLIMFOR models are able to accurately and reliably predict atmospheric parameters over long forecast horizons.

geopredict predictive knowledge factory and forecasting services

Location Pin

Headquarters

Brandteichstr. 20

17489 Greifswald

Germany



Contact



Scale Balanced

CO2 Neutral 

Certified by the EIC Greenhouse Gas Programme

CO2 reduction certified by EIC greenhouse gase programme. geo-forecasting, knowledge factory, energy forecast, climate forecast, renewable energy production, forecast,


© geopredict GmbH, All rights reserved.

This website makes use of cookies. Please see our privacy policy for details.

Deny

OK