Flood Forecast accuracy has decreased while number of forecasts increased
Sai Krishna Muthyanolla
October 27, 2020
Accurate forecasts play an important role in mitigating the losses caused by floods. Government data indicates that while the number of level & inflow forecasts has more than doubled between 2015 & 2019, the accuracy has reduced from 98% to 86.6%.
Recent floods due to depression in Bay of Bengal wreaked havoc in the states of Telangana, Andhra Pradesh, Karnataka & Maharashtra leading to loss of human life as well as property. Earlier, Assam was hit by floods which has now become an annual affair. It is estimated that around 2.25 lakh people were affected by these floods in the state of Assam.
Rains during the monsoon season and unseasonal rains in different parts of the country have over the years, resulted in immense loss to human life, crops, property etc.  Governments at the Centre and the States have developed mechanism to predict the floods and mitigate the potential loss. Despite all these measures, significant number of human lives are lost in addition to loss of property, and death of animals etc.  In this story, we take a look at the magnitude of losses caused due to floods in India.
More than 8,800 deaths are estimated to be caused due to floods between 2017 & 2020
Responding to a question in Lok Sabha in November 2019, the Union Minister for Jal Shakti provided the details of Death toll caused due to floods. The total losses due to floods was provided as a response to a different question in July 2019.
As per the information, in 2016-17, 1550 human lives were lost due to rain oriented natural calamities. This number increased in the year 2017-18, when a total of 2,494 deaths were recorded as caused due to floods across India. In the following  year of 2018-19, the number reduced slightly with 2,400 deaths being attributed to floods.  As per the available provisional estimates, there were around 2391 deaths that were cased due to floods in 2019-20 up to November 2019.
Over this four-year period, the highest number of deaths were reported in the states of Bihar & West Bengal. Both these states have recorded over thousand deaths each between 2017 & 2020.
While, the number of deaths in West Bengal have been consistently high in each of these years, the overall numbers of Bihar are skewed with 2017-18 along recording 649 deaths.
Madhya Pradesh reported around 674 deaths in 2019-20, while Kerala & Himachal Pradesh lost more than 400 people each due to floods in 2018-19.
Increase in numbers of Livestock lost and Houses Damaged in 2018-19, while there is a fall in crop area affected
In a response provided in July 2019, the union government states that the measures taken by Central & State governments have improved disaster management and reduced the causalities during natural calamities.
However, the data provided by the government in the same response, provides a contrasting picture. The latest data available for 2019-20 is only an estimate, hence only the information for the three-year period i.e. 2016-19 is considered for the analysis.
States with regular problem of floods lost fewer livestock in the last two years
The inconsistent nature of the occurrence of floods does not present an accurate picture to compare the trends over the years. For example, the increase in the number of livestock deaths in 2018-19, can be attributed to the large-scale floods that occurred in Kerala during that year. Out of the 1.22 lakh livestock deaths, 76.8 thousand deaths were in Kerala. Tamil Nadu which also had severe floods in 2018-19 reported more than 20 thousand livestock deaths. However, in the previous years, these states had significantly fewer deaths making it difficult to  understand effectiveness of disaster management.
However, improvement was observed in the case of Assam & West Bengal – two states which are regularly affected by the floods. The loss of livestock in 2018-19 was significantly less than in previous years in Assam. Similar was the case with West Bengal. Bihar did not report any loss of livestock due to foods in 2018-19, but the trends in previous years do point to a decline.
In most of the other states, the higher numbers are because of a higher number in a single year.
Similar trend observed in the case of Damaged houses
A similar trend was observed in the case of damaged houses, with Kerala & Tamil Nadu reporting a damage of 6.54 lakhs & 5.67 lakhs in 2018-19 out of the total 15.57 lakh houses damaged in that year.
Even the numbers for West Bengal, Bihar & Assam are inconsistent over the years and does not point to any specific trend. The variations in specific states, the severity of flood in those areas etc. makes it a challenging to compare the numbers across the years.
The fall in the area of affected crop land in 2018-19, can be attributed to minimal impact to crop land in the major flood affected states of Kerala & Tamil Nadu that year. Other states which usually are affected by floods did not witness severe floods in 2018-19.
Forecast accuracy falls over the years while the number of forecasts increases
The Central Water Commission (CWC) has established a flood forecasting network across the country. As per the update provided in Lok Sabha in September 2020, there are a total of 328 such network stations. Out of these, 198 are level stations i.e. they provide the forecast for the Villages/Town on the banks of rivers. The remaining 130 are inflow forecast stations for Dams & Barrages.
Bihar, Uttar Pradesh & Assam have the highest number of level stations with 40,39 & 30 respectively. Over the years, the number of forecasts issued by these stations has increased, but there is a fall in the accuracy of these forecasts.
For Level forecasts, if the forecasted level is +/- 0.15 m of actual level, it is considered as correct.  In 2015, 3500 level forecasts were made with an accuracy on 97.97%. In 2019, the number of forecasts increased to 6004, but the accuracy fell to 96.5%.
A much steeper fall in forecast accuracy can be observed in the case of inflow forecasts. In 2015, a total of 572 inflow forecasts issued with an accuracy of 98.25%. By 2019, the number of forecasts issued increased to 3750, but of these only 2678 were accurate forecasts i.e. an accuracy of only 71.41%.  For an inflow forecast, the forecasted inflow needs to be with +/- 20% of the actual inflow attained in the reservoir when the forecast is taken, for it to be considered correct.
In the five years between 2015 & 2019, out of the yearly average of 6,643 forecasts made by flood forecasting network (both level & inflow), a total of 6,157 were within the acceptable deviation i.e. accuracy of 94 %.  However, the accuracy has consistently reduced from 98.01% in 2015 to 86.64% in 2019 while the number of forecasts has increased.
Preparedness for unexpected floods remains a challenge
Recent trends of the last four years  (2016 to 2019) indicate that the major contributor to the loss in human life, livestock and property are the states which experienced unexpected floods during a particular year (as seen in case of Kerala & Tamil Nadu in 2018-19, Rajasthan, West Bengal in 2017-18 etc.).
On the other hand,  there seems to be an improvement in states like Assam, Uttar Pradesh & Bihar which face the problem nearly every year. However, it is too early to deduce any major conclusions based on the numbers of these four years.
The primary responsibility of disaster management lies with the State governments, with the Centre providing the required technical and financial assistance. The Government of India (GOI) launched Flood Management Plan (FMP) to provide financial assistance to State governments for taking up various works – river management, flood control, anti-erosion works etc.
Over the last three years (2017-20), the Central government has released over 2 thousand crores to various states as per the data share in the Lok Sabha.
While the Centre & States seem to be taking  proactive steps in forecasting floods, preventive actions etc. there seems to be a challenge in case of unexpected floods. Inconsistent rainfall, lack of proper drainage etc. results in occurrence of immediate flood situations which the authorities are unable to respond to quickly. These are more prominent in the urban areas.  Further, the decreasing forecast accuracy as observed from the data is also a cause for concern especially with regards to inflow of flood water, as this results in flooding downstream.
While there are larger institutional and policy challenges like – town planning, forest management, drainage development etc. which could help managing the flood situation, better forecast mechanisms could help mitigate the losses in the near term.
Featured Image: Flood Forecast