Answer: The damage amount and the link between a car and an accident are not in the four tables given, so this is the standard insurance schema with a fifth table Participated(driver_id, license, reportno, damage_amount). With it, the queries are below.
Schema used (the # in driver_id# only marks the key; it cannot be part of a column name in SQL):
Person(driver_id, name, address)
Car(license, model, year)
Accident(reportno, date, location)
Owns(driver_id, license)
Participated(driver_id, license, reportno, damage_amount)
1. Total number of people who owned cars that were involved in accidents in 2004
SELECT COUNT(DISTINCT o.driver_id) AS total_people
FROM Owns o
JOIN Participated p ON o.license = p.license
JOIN Accident a ON p.reportno = a.reportno
WHERE a.date BETWEEN '2004-01-01' AND '2004-12-31';
(In MySQL, WHERE YEAR(a.date) = 2004 also works.) DISTINCT makes sure an owner is counted once even if his car had several accidents.
2. Total number of accidents in which cars belonging to "HT" were involved (HT is taken as the owner's name)
SELECT COUNT(DISTINCT a.reportno) AS total_accidents
FROM Person pr
JOIN Owns o ON pr.driver_id = o.driver_id
JOIN Participated p ON o.license = p.license
JOIN Accident a ON p.reportno = a.reportno
WHERE pr.name = 'HT';
3. Update the damage amount of car mum2011 in accident AR2011 to Rs. 4000
UPDATE Participated
SET damage_amount = 4000
WHERE license = 'mum2011' AND reportno = 'AR2011';
Tested on sample data in SQLite (accidents AR2004 and AR2005 in 2004 involving cars owned by three people; HT owns mum2011 and mh04cd, which were in four accidents):
Query 1: total_people = 3
Query 2: total_accidents = 4
Query 3: Participated row (D1, mum2011, AR2011) now has damage_amount = 4000;
the other row of mum2011 (AR2012, 800) is unchanged